Characterizing spurious (Harmonic) responses of the SDRPlay RSP1a (and other models)

The SDRPlay RSP2pro (left) and RSP1a receivers (right)

The SDRPlay RSP1a is a popular Software Defined Radio (SDR).  This device, connected to and powered by the computer via a USB cable covers from VLF through UHF and low microwave frequencies.

This receiver shares a similar internal architecture of similar devices such as the RTL-SDR dongle and the AirSpy in that an analog frequency converter (mixer) precedes the analog-to-digital converter:  In the case of the SDRPlay, the frequency to which the receiver is tuned is (usually) converted to baseband I/Q signals, with the "center" frequency being at zero Hz (DC). ¹

Note:

For the purposes of this discussion, there is no difference between the RSP1a and some of the other receivers in the product lineup (e.g. RSPDuo, RSPdx and the discontinued RSP1, RSP2 and RSP2pro) in terms of harmonic response across the 2-30 MHz range as they all have about the same 12 MHz and 30 MHz cut-off frequencies on their input filtering - properties that would affect HF reception across the 2-30 MHz range in terms of harmonic response.

This issue was noted at least as far back as 2017 in the SDRPlay forum - See this thread:  https://www.sdrplay.com/community/viewtopic.php?t=2280

Imperfect mixers

By its nature, a frequency mixer is a non-linear device.  Ideally, the two frequencies applied to a mixer would yield just two more - the sum and difference.  For example, if we applied a 5 MHz signal and a 1 MHz signal to a mixer, it would output both the sum of 6 MHz and the difference of 4 MHz - and this is true, but there's more to the story.

In our example - with a real-world mixer, we will also get additional products - including those related to the harmonics of the local oscillator and the applied signal.  Because of this, we will see weaker signals at:

• 11 MHz (2 * 5 MHz + 1 MHz) 
• 9 MHz (2 * 5 MHz - 1 MHz) 
• 7 MHz (5 MHz + 2 * 1 MHz) 
• 3 MHz (5 MHz - 2 * 1 MHz) 
• And so on.

Typically, these "other" signals will be quite a bit weaker than the original - but they will still be present, possibly at a high enough level to cause issues such as spurious signals - a problem with both receivers and transmitters.  Typically, these are tamed by proper design of the mixer, proper selection of frequencies and careful filtering around the mixer to limit the energy of these "extra" signals.

Note:  There will be a response at 5x the center frequency as well, but it is suppressed better than the 3x response by the mixer and - for the 80 meter amateur band and higher - these responses are suppressed reasonably well by the filtering.

SDRPlay's poor harmonic response suppression on 80 meters and below.

ANY receiver will experience spurious responses related to mixing products.  Typically, filtering is employed to remove/minimize such responses, but for a wide-bandwidth receiver such an SDR, doing this is complicated by the fact that being able to cover wide swaths of bandwidth would ideally require a large number of overlapping filters.

An example of a radio where this is done - albeit of different architecture - is the Icom IC-7300 which has nine overlapping band-pass filters that cover 160 through 10 meters.  While the reasons for the '7300 having many filters has as much to do with its being a "direct sampling" ² type of SDR, good filtering on the signal path of any type of receiver - SDR or "HDR" (Hardware Defined Radio - or an "old school" analog type) is always a good idea

If this many filters had been implemented on the SDRPlay, there would be enough filtering to prevent a significant harmonic response.  In the case of the RSP1a, this was not done - partly to allow 5-8 MHz of continuous coverage without being significantly impacted by the filters in many cases, but more likely it was done due to practical reasons of economics ³ :  There are just three filters used for covering all of the "HF" amateur bands 160 through 10 meters:  One that covers up to 2 MHz, another that covers 2-12 MHz and third that covers 12-30 MHz:  This information is covered in the RSP1a technical information document ( https://www.sdrplay.com/wp-content/uploads/2018/01/RSP1A-Technical-Information-R1P1.pdf )

The sensitivity to harmonics was tested with the RSP1a's local oscillator (but not necessarily the virtual receiver) tuned to 3.7 MHz ⁴ .  For reasons likely related to circuit symmetry, it is odd harmonics that will elicit the strongest response which means that it will respond to signals around (3.7 MHz * 3) = 11.1 MHz.  "Because math", this spurious response will be inverted spectrally - which is to say that a signal that is 100 kHz above 11.1 MHz - at 11.2 MHz - will appear 100 kHz below 3.7 MHz at 3.6 MHz.  (It's likely that there are also weaker responses at frequencies around 5 times the local oscillator, but these are - for the most part - adequately suppressed by the filtering.)

In other words, the response to spurious signals follow this formula:

Apparent signal = Center frequency + ((Center frequency * 3) - spurious signal) )

Where:

• Center frequency = The frequency to which the local oscillator on the RSP is tuned.  In the example above, this is 3.7 MHz.
  
• Spurious signal = The frequency of spurious signal which is approximately 3x the center frequency.  In the example above, this is 11.2 MHz.
  
• Apparent signal = Lower frequency where signal shows up.   In the example above, this is 3.6 MHz.
  

In our example - a tuned frequency of 3.7 MHz - the 3rd harmonic would be within the passband of the 2-12 MHz filter built into RSP1a meaning that the measured response at 11.2 MHz will reflect the response of the mixer itself, with little effect from the filter as the 2-12 MHz filter won't really affect the 11 MHz signal - and according to the RSP1a documentation (link), this filter really doesn't "kick in" until north of 13 MHz.

In other words, in the area around 80 meters, you will also be able to see the strong SWBC (Shortwave Broadcasting) signals on the 25 meter band around 11 MHz.

How bad is it?

Measurements were taken at a number of frequencies and the amount of attenuation is indicated in the table below.  These values are from measurement of a recent-production RSP1a and spot-checking of a second unit using a calibrated signal generator and the "HDSDR" program:

LO Frequency	Measured Attenuation at 3X LO frequency	Attenuation in "S" Units
2.1 MHz	21 dB (@ 6.3 MHz)  - Using 2-12 MHz filter	3.5
2.5 MHz	21 dB (@ 7.5 MHz)	3.5
3.0 MHz	21 dB (@ 9.0 MHz)	3.5
3.7 MHz	21 dB (@ 11.1 MHz)	3.5
4.1 MHz	23 dB (@ 12.3 MHz)	3.8
4.5 MHz	30 dB (@ 13.5 MHz)	5
5.0 MHz	39 dB (@ 15.0 MHz)	6.5
5.5 MHz	54 dB (@ 16.5 MHz)	9
6.0 MHz	54 dB (@ 18.0 MHz)	9
6.5 MHz	66 dB (@ 19.5 MHz)	11
12.0 MHz	21 dB (@ 36.0 MHz) - Using 12-30 MHz filter	3.5
12.5 MHz	21 dB (@ 37.5 MHz)	3.5
13.5 MHz	22 dB (@ 40.5 MHz)	3.7
14.5 MHz	26 dB (@ 43.5 MHz)	4.3
15.5 MHz	31 dB (@ 46.5 MHz)	5.2
16.5 MHz	35 dB (@ 49.5 MHz)	5.8
17.5 MHz	39 dB (@ 52.5 MHz)	6.5
18.5 MHz	43 dB (@ 55.5 MHz)	7.2
19.5 MHz	46 dB (@ 58.5 MHz)	7.7
20.5 MHz	50 dB (@ 61.5 MHz)	8.3
21.5 MHz	53 dB (@ 64.5 MHz)	8.8

Table 1:  Measured 3rd harmonic response of the RSP1a

Interpretation:

• In the above chart we see the local oscillator frequency in the left column, the measured attenuation of the 3rd harmonic response (and its frequency) in the center column, and that amount of attenuation expressed in "S" units.  Here, an "S" unit is based on the IARU standard (Technical recommendation R.1) of 6 dB per S unit, which is reflected in programs like SDRUNO, HDSDR and many others.
  
• The attenuation of the 3rd harmonic response was measured by first noting the signal level required to obtain a given reading - typically "S-9" near the fundamental frequency - and then observing the level required to obtain that same reading - within +/-1dB - near the 3rd harmonic frequency, using the relationship formula, above.
  
• Below the cutoff frequency of the relevant filter (nominally 12 MHz for receive frequencies in the range of 2 to 12 MHz, nominally 30 MHz for receive frequencies in the range of 12 to 30 MHz) the harmonic response is limited to that of the mixer itself, which is about 21 dB.
  
• We can see that on the 2 to 12 MHz segment, the attenuation related to the 3rd harmonic doesn't exceed 40 dB (which is the low end of what I would call "OK, but not great) until one gets above about 5 MHz (which translates to 15 MHz) and it doesn't get to the "goodish" range (50dB or more) until north of about 5.5 MHz which is borne out by the filter response charts published by SDRPlay.
• On the 12 to 30 MHz band the 2-12 MHz filter has practically negligible effect until one gets above about 20 meters, at which point it gets into the "OK, but not great" range by about 18 MHz, and it doesn't really get "goodish" until north of 20.5 MHz.
• Similarly, the "12-30 MHz filter isn't particularly sharp, either.  What this means is that strong 6 meter signals may well appear in the 16.5 to 17.5 MHz range as frequency inverted representations.
• If there is a relatively strong signal source in the area of the 3rd harmonic response, it will likely appear at the lower receive frequency where the attenuation of the filter is less than 40 dB or so.  The severity of this response will, of course, depend on the strength of that signal, the amount of attenuation afforded by the filters at that frequency, and the amount of noise and other signals present in the range of the fundamental frequency response.
  

Based on the above data, we can deduce the following:

• When the RSP1a is tuned between 2 MHz and (below) 12 MHz, it is using its "2-12 MHz" filter. In this range - and below approx. 4 MHz - the 12 MHz cut-off of the filter has negligible effect in reducing 3rd harmonic response.
• What this means is that signals from 6-12 MHz will appear more or less unhindered (aside from the 21 dB reduction afforded by the mixer) when the local oscillator of the receiver is tuned between 2 and 4 MHz.
• The 3rd harmonic response across 2-4 MHz - which is the 6-12 MHz frequency range - can contain quite a few strong signals and noise sources such as those from shortwave broadcast stations.
  
• When the RSP1a is tuned between 12 MHz and (below) 30 MHz, it is using its "12-30 MHz" filter.  Below about 14 MHz, the 30 MHz cut-off of the filter has negligible effect in reducing 3rd harmonic response.
• Signals from 36-40 MHz will appear with just 21-26 dB attenuation when tuned in the range of 12-13.5 MHz.
• In most cases there are probably few signals in the 36-40 MHz range that are likely to be an issue when tuning in the 12-13.5 MHz range.

80 meter example:

Connecting the RSP1 to a known-accurate signal generator set to -40dBm, the signal level at 3.6 MHz was measured:  Maintaining the signal level, the generator was retuned to 11.2 MHz and the resulting signal level was measured to be 21 dB (a bit more than 3 "S" units) lower than that at 3.6 MHz.

What this means is is that a "20 over S-9" signal at 11.2 MHz will show up as an S-9 signal at 3.7 MHz, and an S-9 signal at 11.2 MHz will be around S-6 at 3.7 MHz.  In other words, even a "weak-ish" signal at the 3rd harmonic will show up at the lower frequency.

80/60 meter example:

If you run the RSP1a in a wider bandwidth mode, it is possible to simultaneously see and tune a greater frequency range.  For example, let us presume that you wish to cover both 80 and 60 meters using a single RSP1.  To do this, you could set the center (LO) frequency to 4.5 MHz and set the sample rate to 5.376 MHz and use the 5 MHz band-pass filter built into the RSP1's converter/mixer chip (the Msi001) to prevent in-band aliasing.

In this configuration 20 meter signals will appear at the top of 80 meters owing to the relationship in the formula that we described above.  Taking the 20 meter FT-8 subband at 14.074 MHz as an example, we see that:

4.5 MHz + ((4.5 MHz * 3) - 14.074) = 3.926 MHz

In other words, we will see spectrally-inverted representations of 20 meter FT-8 signals around 3.926 MHz, and the rest of the (upper) portion of 20 meters across the rest (lower) portion of 80 meters where USB signals on 20 meters will show up as LSB signals on 80.  We know from the chart above that those signals will be attenuated by between 30 and 39 dB (about 5-6 S-units).  This might sound like a lot of attenuation, but it means that a "20 over" signal on 20 meters will appear at around S-7 to S-8 on 80 meters - still quite respectable most certainly very audible!

More about filtering and harmonic response

While these spurious responses may not be too much of a problem for the casual user, it will be necessary to add additional filtering to allow the RSP1a to function on par with a modern, SDR receiver from one of the major manufacturers.

Unfortunately, the filtering in the RSP1a is not sufficient in the 80 meter case mentioned above as it doesn't have octave filters (or similar) - but what about 60 or 40 meters?

The table above answers this question.  In the case of 60 meters - with the receiver tuned to 5.3 MHz - our 3rd harmonic will land on 15.9 MHz.  Based on measurements of the receiver the response of signals around 15 MHz - which corresponds to the 19 meter Shortwave Broadcast Band - will be a bit more than 40 dB down from 40 meters with about 20 dB of this being due to the roll-off of the 2-12 MHz filter - but because this frequency range is inhabited by very strong shortwave broadcasters they are likely to still be quite audible around 60 meters.

The situation is a bit better for 40 meters where the 3rd harmonic is around the 15 meter band.  There, the 2-12 MHz filter knocks signals down by 50dB or more, putting them about 70dB below the 40 meter response - on par with about any respectable receiver.

What this means is that for amateur bands below 40 meters it is suggested that additional filtering be applied.

The best solution - and recommended for any software-defined radio (or even older "hardware-defined radios") is to have band-pass filter designed for the specific amateur band in question. This will not only significantly attenuate the harmonic response, but it will also reduce the total amount of RF energy entering the receiver, reducing the probability of overload.  The obvious down-side is that it will reduce the flexibility of the receiver in that unless you change/remove it, you won't be able to receive signals well outside the filter's design range.

Another possibility is to add a low-pass filter that is designed to cut off signals above the band of interest.  For example, if you have a filter that cuts off sharply above 8 MHz, you will be able to tune 80-40 meters and get reasonable attenuation of the 3rd harmonic response across this entire frequency range.

In the case of 160 meters the RSP1a will automatically select the 0-2 MHz low-pass filter and the 3rd harmonic response will be a respectable 50-ish dB down, depending on frequency.

On 20 meters - where the 3rd harmonic is around 42 MHz - the "12-30 MHz" filter will be selected, but the published response of this filter shows that at 42 MHz its attenuation will be quite limited.  Practically speaking, it is unlikely that there will be any signals in this frequency range so there being "only" 20-30dB of attenuation is unlikely to cause a problem in most cases, but one should be aware of this.

What can be done:

In short, none of the currently-made SDRPlay receivers - by themselves - will offer very good performance in terms of harmonic rejection between 2 and 5 MHz and it will be particularly bad on the 80 meter band where strong 25 meter SWBC signals can appear:  It is interesting that the ARRL review of the RSPdx (Link here) didn't catch this issue.

It is unfortunate that the designers of the SDRPlay receivers did not add at least one additional low-pass filter in the signal path to quash what is a rather strong response in the 2-6 MHz range - particularly on 80 meters, one of the most popular bands.  A low-pass filter with a cut-off frequency of 6 MHz (with attenuation becoming significant above 7 MHz) would ameliorate the harmonic response when tuning across this band.  This problem is made even worse by the fact that even antennas that aren't particularly resonant at their harmonic responses (e.g. the antenna for 80 meters) will likely do quite a decent job of receiving signals in the 11-12 MHz area.

The only real "fix" for this is to install additional filtering between the SDRPlay receiver and the antenna.  If single-band operation is all that is desired, the best choice will be a band-pass filter designed for the frequency range in question ⁵ - but unless you are dedicating the receiver just for that one band, this isn't really desirable unless you can easily switch/bypass the filter when tuning elsewhere.

A more flexible solution would be to use a low-pass filter.  As we noted above, the 12 MHz roll-off of the built-in (2-12 MHz) filter just doesn't do much to suppress signals from 20 meters, but if we had a filter that had a sharp cut off beginning, say, at 8 MHz, we could use it for 80, 60 and 40 meters - such a filter is depicted schematically, below:

8 MHz low-pass filter schematic, designed using ELSIE

This filter is pretty easy to build:

• Capacitors 1 and 5 each consist of a 100pF and 470pF in parallel
• Capacitor 3 consists of a 680pf and 220pF in parallel - although you could probably get away with two 470pF capacitors in parallel in a pinch.
  
• Inductors 2 and 4 consist of 16 turns on a T50-2 (or 18 turns on a T37-2) toroid using small wire - 24-30 AWG (0.6-0.25mm dia.) is fine

A small scrap of PC board material - about 2"x4" (5x10cm) is more than large enough to accommodate with the capacitors soldered directly to the foil and inductors held aloft by the capacitors.  The connectors should be attached to the PC board directly - or with short lengths of coax, keeping the ground (shield) lead length to an absolute minimum to minimize the probability of ground-loop induced noise currents.

If you have access to a NanoVNA it's quite easy to check the performance.  If anything, it may be necessary to spread the turns across the toroid or remove one turn - but this design is quite forgiving and it should work adequate with no tuning at all.

The obvious down-side for this is that if you are tuning all over the HF spectrum (above 7.5-8 MHz in the case of the filter above) you'd have to manually remove or bypass any such filtering when you tuned beyond the range that the added filter would pass.

 

Footnotes:

1. The receivers mentioned at the beginning of the article (SDRPlay, AirSpy HF, RTLSDR, etc.) have analog-to-digital converters that cover only a portion of the HF spectrum, using a frequency mixer to convert a range of frequencies from the range of interest to a lower frequency, which is then fed into the converter.  Limiting the amount of spectrum being ingested by the receiver - particularly when appropriate filtering is used - can improve performance, reduce cost, and especially reduce the total amount of data, allowing a modest computer (older PC, Raspberry Pi) to be used with it.
   
2. A "direct sampling" type of receiver - such as that found in the Icom IC-7300, IC-7610, the KiwiSDR, Red Pitaya and the RX-888 (when used at HF) and others like them simply "inhale" large swaths of spectrum all at once.  Because the analog-to-digital converter itself has a limited amount of total RF signal power that it can handle, radios like the Icoms have filtering that allow the passage of only the (relatively) small portion of the HF spectrum around that to which the receiver is tuned, reducing the probability of overload from strong signals on frequencies well away from those of interest.  Other direct-sampling receivers such as the KiwiSDR, Red Pitaya and RX-888 do not necessarily have band-specific filtering as they are intended to be able to receive multiple frequencies across the entire HF spectrum at once and as such, much more care is required in implementation to prevent overload/distortion for these devices.
   
3. In the case of the (currently-produced) RSP receivers, the filtering varies depending on model:  In the case of the RSP1a, it has a band-pass filter that covers 2-12 MHz while other models have used just a 12 MHz low-pass - the former being capable of rejecting AM broadcast band (e.g. mediumwave) signals from the input of the receiver when tuned to HF, and the latter not.  Some units additionally have a separate "notch" (reject) filter that is designed to remove just AM broadcast-band signals.  The situation described in this article - the reception of signals around 11 MHz when tuned to 80 meters - is related to the fact that the 2-12 MHz filter represents a 6:1 frequency range which means that over the lower portion of this spectrum, the 12 MHz cut-off of this filter cannot possibly remove responses to the third harmonic, hence the issue described here.
4. If you are using a program like SDRUno it may not be readily apparent to what frequency the receiver's local oscillator is tuned.  If set to "Zero IF" mode, the local oscillator will be tuned at the same place as the center of the waterfall display when it is fully zoomed out - typically indicated by a slight line at the "Zero Hz" frequency there there is a slight amount of noise energy.  By default, one cannot directly tune the local oscillator ("Zero IF" frequency) in SDRUno.  If you use the "HDSDR" program by I2PHD (et al) you can independently tune the local oscillator and the frequency of the virtual receiver.
   
5. SDRPlay receivers are currently in use at a number of well known and public WebSDRs around the world as the "acquisition device" (e.g. receiver).  In most cases these receivers - because they are used only for specific amateur bands - are preceded by a band-pass filter for the band that they are covering, completely eliminating issue noted in this article.  It was during testing at one of these WebSDRs - a receiver on 80 meters that does not (yet) have additional filtering - that signals were noted across the 80 meter band in the middle of the day that should not have been there at all - and these signals were quickly realized to be the result of a harmonic response in the front end.  These responses were then verified and quantified using two other RSP1a receivers (of different production runs) and test equipment during the preparation of this article.  When it was convenient to do so, a low-pass filter with a cut-off frequency of 7.5 MHz was installed on this receiver, solving the image problems.
   

* * *

This page stolen from ka7oei.blogspot.com

[End]

A 2 meter and 222 MHz low-pass filter for the SocoTran ST-7900D

In an earlier post (link) I described a "hilariously bad" radio - the Socotran ST-7900D (a.k.a. the QYT "KT-7900D").  This radio, right out of the box, could not legally be used by U.S. amateurs on three of its four "bands".


One of these bands - that which covered 350-390 MHz - was understandably off-limits as there is no U.S. amateur band in this frequency range but the other two, the 2 Meter and the 1-1/4 meter (a.k.a. 222 MHz band), also covered by this radio, had poorly-filtered harmonic content:  It was even possible to key up a fairly-distant UHF repeater when one transmitted on a 2 meter frequency at precisely one-third of its input frequency!

The article noted that as it was shipped, the only band that might be legally used was the 70cm band as the strongest harmonics of the other bands weren't properly suppressed - not at all, actually...  It was observed that this radio seemed to have a single low-pass filter in its transmit path that was designed to start cutting off energy in the 550-575 MHz range - but this sort of filter would have no effect at all on the 2nd and 3rd harmonics on 2 meters and the 2nd harmonic on 222 MHz - which was the problem.

Besides just being cheap, one reason why someone might have been attracted to this radio is its ability to operate in the 222 MHz band - and paying $75 or so for a  25 watt radio that could only do 222 MHz might be a reasonable thing to do - so what about making some sort of low-pass filter that would kill two birds with one stone:  A single filter that would allow legal operation on both 2 meters and 222 MHz without having to switch filters?

Designing the filter:

What I needed was a filter that would pass the 222-225 MHz band with little attenuation, but still knock out the 2 meter band's 2nd harmonics in the 288-296 MHz range.  This sort of filter would permit operation on both the 2 meter and 222 MHz band without needing to switch or change anything - but if would, of course, preclude operation on 70cm unless it were removed.

Curious to see if this could be done I fired up the Elsie program, a software tool that is free for "student", non-commercial use (aren't we all students in this world?).  Designing a filter that would both adequately attenuate 2 meter's 2nd harmonics and pass 222-225 MHz would require at least a slight amount of complexity, so I went to work.

Knowing that a simple Butterworth or Chebychev filter would never meet the need for "sharpness", I immediately picked a Cauer (a.k.a. "Elliptical") low-pass filter design and plugged in the numbers, coming up with this:

Figure 1:
 Low-pass filter, inductor-input topology shown.  This "same" filter could have been constructed using capacitors on the input/output, but this version uses fewer capacitors and more inductors - which are both extremely cheap to make and are very easily adjusted - unlike fixed capacitors.
Click on the image for a larger version.

When I plotted the predicted response of this filter, Elsie showed me this:

Figure 2:
The predicted attenuation of the filter.  Part of the design goal was to place the 2 meters' second harmonics in the first "notch" in the filter.  While only 40dB was theoretically needed, a filter with 50dB attenuation was implemented knowing full-well that the real-world implementation of the filter may not do quite as well.
Click on the image for a larger version.

While I needed "only" 40dB to make this radio "clean enough" it is often the case that real-world filters aren't quite as good as their simulated counterparts, so I inputted 50dB into the program as the minimum attenuation.  If you look closely, you'll see that at the top of the flat part - just before it "rolls off" - the attenuation at 232 MHz, comfortably above the 222 MHz band, is just under 1dB while there is a deep "notch" at around 290 MHz - which is right where the 2nd harmonics of the 2 meter band will lie.  If this filter was, in fact, "build-able", it would neatly solve the problem of the harmonics from the 2 meter and 222 MHz bands.

When it came to specific filter types I had two choices for the "same" filter:  A capacitor-input low-pass filter and an inductor-input low-pass filter.  Both are theoretically equal in performance, but because the capacitor input version had 7 capacitors and the inductor input version had just 3 capacitors, I chose the latter as seen in Figure 1:  Inductors - which are just a few turns of wire - cost practically nothing to make and are easily adjusted!

Being familiar with VHF/UHF construction techniques I knew, when I saw the inductor values, that they would be very small, but easy to make.  For example, when 20 AWG wire is wound on a 3/16" (4.76mm) diameter drill bit with very short leads, you can expect to start with something along the lines of:

• 20-30nH:  2 turns
• 30-40nH:  3 turns
• 40-50nH:  4 turns

The precise value would then be obtained by squeezing/stretching the turns - or, possibly cutting a fraction of a turn off as necessary.  If you have ever looked at some commercially-made VHF/UHF gear you might have noticed that some of the coils look as though they were smashed or stretched, but this is a time-honored way of making fine adjustments to these circuits after assembly.

If one is constructing this using only small, surface-mount components the self inductance and stray capacitance of these tiny components on a well-designed board can almost be ignored at these frequencies - but I was going to use plain, old through-hole leaded disk ceramic capacitors, which would require a bit more consideration.

A good example of this is the first series-resonant section of the above filter - in the section marked "531.581M".  As you can guess, this is a series-tuned circuit that must be resonated at around 532 MHz using components of the approximate values shown.  Practically speaking, in this application one can "fudge" a bit on the values, so rather than trying to find a precision capacitor of about 17.5pF, I simply pulled a 18pF unit out of my capacitor bin with the idea that I would select the inductance to make it resonate somewhere in the area of 532 MHz.

But, there's a twist:  Noticing that resonating inductance is ideally 5.1nH, one may realize that even a rather short length of wire has a similar amount of inductance - and that is exactly what was done:   The capacitor's own lead - about 4 millimeters of it - plus the series inductance of the capacitor itself was enough to create a resonant circuit at the desired frequency.

What it takes to build this filter:

As you may have gathered, it is simply not possible to build this filter without some sort of test equipment at hand - and I used a spectrum analyzer with a tracking generator as I was building it.  In short, here's what I had to do:

• Fuss with the series L/C circuits to get the stated series resonant frequencies as indicated by deep notches on the sweep.
• Stretch/compress/adjust the other inductors as necessary to minimize the loss below the cut-off frequency
• Repeat the above two steps until it makes no difference - usually taking about a half-dozen iterations.

During construction I didn't bother breaking out any capacitance or inductance measuring gear - but very small inductors (those lower than a few hundred nanoHenries) can be very difficult to measure, anyway.  Using only "known" values of capacitance, by adjusting the inductors in the manner mentioned above, I have found via experience that such filters often "take care of themselves" when one takes a bit of care during assembly and adjustment - particularly when setting the resonant frequencies of the "notch" elements.

Amazingly, the filter went together without too much trouble with the test equipment indicating less than 1dB of insertion loss at either 2 meters or 222 MHz.  The hastily-kludged prototype looks like this:

Figure 3:
Constructed prototype.  This was constructed on a scrap piece of copper-clad PC board material using small PC board islands for some component support.  This version was built for testing the concept:  A "real" implementation of this filter would be crammed into a small metal box with the input/output inductors and ground plane soldered directly to the in/out RF connectors.
A circuit board could be designed, but to be effective it would need to be built on (at least) a double-sided board with a large-as-possible ground plane on the top connected to the bottom plane using lots of vias.
Click on the image for a larger version.

Figure 3 shows the prototype, constructed on a small piece of copper-clad circuit board material.  The input/output connections were made via some N connectors that were pre-attached to UT-141 rigid PTFE coaxial cable, being were used because they were on-hand.  Because we would need to handle "only" 25 watts, 100 volt NP0/C0G disk ceramic capacitors are more than adequate.

As can be seen in Figure 3 the junctions where the series L/C portions are attached are held off the ground plane with small pieces of circuit board (the capacitive effects of these are negligible at these frequencies) while the attaching hardline's center conductors supported the in/out inductors.  Also apparent is what looks like haphazard stretching/compressing of the various inductors to achieve the resonant frequencies for the three elements - which I marked on the board.

Once I connected it to a transmitter I did a bit of final tweaking, "adjusting" the series coils for minimum loss (as indicated on an RF power meter) on both 2 meters and 222 MHz but leaving the "notch" adjustments alone:  Only slight adjustments were needed.

Can I make such a filter?

Yes, you can - if you are familiar with VHF/UHF circuit techniques and have access to a spectrum analyzer with a tracking generator or some sort of equivalent.

If you don't have access to this sort of gear - and you don't know anyone else who does - then it is (unfortunately) not possible to properly "tweak" this filter for both harmonic attenuation and also to make its insertion loss and added VSWR low enough to both allow transmit power to pass through it without damaging the radio.

(And no, I won't build one for you...  Remember:  It's a $75 radio!)

Does it work?

Amazingly enough, it works pretty much as predicted!

After final tweaking the measured insertion loss was under half a dB:  With 25 watts in, around 20 watts exited the filter on both 2 meters and 222 MHz - hardly enough loss to worry about on receive or transmit.  After about a minute of solid key-down at 25 watts input the hottest of the filter's components were barely warm - lower than body temperature in a "not hot/not cold" room.

The real test was to put the filter inline and check it again on the spectrum analyzer - and the plots below show the results for 2 meters:

Figure 4:
The harmonics on 2 meters, through the filter.  The analyzer has been adjusted to read actual power, so the 2 meter fundamental is at about +43dBm.  The second marker (#2) shows the location and amplitude of where the 2nd harmonic would be - and this trace shows that it is at least 79dB down well within the FCC part 97 rules and  probably "cleaner" than your average "good" radio!
Click on the image for a larger version.

And here is the result from the 222 MHz band:

Figure 5:
The operation of the radio on the 222 MHz band.  The "2" marker shows the second harmonic - and other spurious signals may be seen at a similar level on the plot.  Like the plot in Figure 4, this is scaled to show the actual transmitter power (+43dBm) and thus the harmonics and spurious signals are around 80dB below the carrier - well within FCC part 97 rules!
Click on the image for a larger version.

As can be seen, the harmonics and other spurious signals are all but undetectable!

"Sweeping" the filter:

Curious as to how the actual attenuation curve of the filter looks?  Figure 6, below, shows its response over the frequency range of 100 through 400 MHz.

Figure 6:
A "sweep" of the prototype filter from 100 through 400 MHz.  On this plot marker #1 has been configured so that the difference in amplitude being measured and this indicates that the depth of attenuation is a bit over 53dB - but the settings of the analyzer used to make this plot likely reduce the apparent depth.  Outside that null the depth of attenuation is at least 40dB - more than enough to suppress the harmonics to meet FCC part 97 regulations.  If I'd taken a bit more care in building the filter (e.g. more thoughtful layout, some shielding between sections) more attenuation might have been obtained - but as it was, it worked better than was necessary.  Because of the configuration of the test jig, the absolute level of the passband portion of the response is arbitrary.
Click on the image for a larger version.

As can be seen, the "depth" of the low-pass filter is at least 40dB, but where the first "null" is located (which happens to be around the frequencies of 2 meter 2nd harmonics) the depth is much greater.

As expected, the simulated filter's ">=50dB" attenuation above the designed cut-off frequency wasn't quite met over its intended range (likely due to the physical layout of the filter - not to mention the difference between simulated and real-world components) but it is more than capable of rendering this radio "legal" when it is operating on the 2 meter and 222 MHz bands.

Using the filter:

Some time in the near future it is likely that this page will show a version of this filter that is built into a small box with UHF connectors which will allow the radio to be used legally on 2 meters and 222 MHz by U.S. amateurs - but having this filter inline precludes its use on 70cm:  To do that, the filter would have to be manually removed.

If that is the case, one would consider this to be a "2 band" radio - and for around $75, it would to OK - aside from the possible tendency for its receiver to overload from nearby signals.

What about automatically switching the filter?

In theory, it should be possible to build into the filter a "bypass" circuit using some UHF-rated relays.

In poking about inside the radio I quickly found a circuit that was powered on only when the UHF (400 MHz) band was selected - and this could be used to "key" a relay to bypass such a filter.  In reality, one would want to design such switching so that when the relay was un-powered, the filter would be bypassed, but when the radio was powered up and not on UHF, use the absence of the aformentioned signal to pull in the relays in insert the filtering.

If we decide to do this, I'll post it here - but at some point, trying to make this radio do what it should have been capable of doing by design becomes an exercise of "turd polishing" when the time and money spent exceeds the gain.

Then again, the effort is sometimes worth the journey if the goal is to build and learn something!

(No, this filter won't help with the problem of this receiver being easily-overloaded by other signals in the same "band".)

* * * * * * * * * * * *

This page stolen from ka7oei.blogspot.com

[End]

A hiliariously bad multi-band radio: The SocoTran ST-7900D

Spoiler:  DO NOT get one of these radios!
 
If you have one, don't use it!

(If you are of the "tl;dr" type, scroll down to the section titled "The Real Problem with this radio") 

A couple of days ago an "interesting" mobile transceiver crossed my path - the SocoTran ST-7900D.   (This radio is also sold under different brands and names, including the "QYT KT-7900D" - and many, many others like the "KT-7900" - usually with "7900" in the model number.)

I say "interesting" - but what I really mean is "scary" - for anyone who ends up using this radio unawares!

Note:  There is a follow-up to this article describing a 2 meter/222 MHz low-pass filter linked here.

Figure 1:
Tiny "quad band" radio - but not really a quad band radio as only three
of the ranges include valid amateur bands!
Can you spot which one of the frequencies shown on the display
is not in a U.S. amateur band?
(Answer at the bottom of the page.)
Click on the image for a larger version.

This radio is diminutive - a little tiny thing that, when you see the "<=25 watt" power rating, makes you wonder how much transmitting it would take to overheat it.  To be sure, it does have a cooling fan, but it is (literally) only slightly bigger than a postage stamp and is of dubious efficacy - but more on that later.  Also a bit alarming is that just sitting there, receiving, the radio gets quite warm - probably about 98 degrees F (37C) in a 70 F (21C) room.

This radio is billed as a "quad band" radio with its frequency coverage being listed as follows:

• 136-174 MHz:  This includes 2 meters, plus lots of other things.
• 220-270 MHz:  This includes the 1-1/4 meter band (a.k.a. the "222 MHz" band).
• 350-390 MHz:  Used for military comms. - There are no U.S. Amateur bands in this frequency range.
• 400-480 MHz:  This includes the 70cm amateur band.

 In reality, only three of its four "bands" are available to a law-abiding U.S. citizen... sort of.

Using the radio:

As is typical for inexpensive Chinese radios, the manual isn't very good - but it's "less bad" than many I've seen, but this isn't much help against the radio's shortcomings.

Problems with the menu system:

The menu system does not appear to be well thought-out.  Here are a few examples:

• Similar items are not necessarily grouped together.  If you want to set transmit offset, subaudible tone, offset direction, power, etc. you must awkwardly jump around between 10s of menu items to do this.  On the test radio, the front-panel knob (the one on the right) didn't seem to reliably change the menu item number up/down so I had to use the up-down button on the microphone or look up the menu item in the manual and enter its number on the microphone.

• The menu selection may not start at the current setting.  If you were to set the subaudible tone to 100.0 Hz and then later change it to, say, 123.0 Hz, you would go into the menu and see it at 100.0 Hz.  However, when you pressed the menu button again to allow the parameter to be changed it will start at 77.0 Hz - the "first" tone in the list, rather than where it had previously been set, requiring you to go through the list again.  For menu items with only a few selections this isn't too bad, but for something like the subaudible tone that has dozens of options this can be a pain!

• Menu settings do not take effect until you enter them.  If you go to the squelch setting in the menu, you hit "menu" again and can change the setting - but it doesn't actually take effect until you press "menu" again to save it.  In other words, to try several squelch settings you have to go back-and-forth several times.

  

  Figure 2:
  The top side of the board.
  The brass cover hides a large, surface-mount transistor
  that is the power amplifier.  The filtering may be
  seen in the lower-right corner of the board.
  Absent seem to be individual low-pass filter sections for
  2 meters and the 222 MHz bands - or any
  means of switching these filters in/out.
  The cover plate for this radio is appears to be
  genuine unreinforced ABS plastic:  No annoying metal
  RF shielding here!  The potentiometer in the lower-left
  corner of the picture sets the radio's reference frequency.
  Click on the image for a larger version.

• By strictly following the manual's instructions it does not seem possible to save a current frequency and its settings (tone, offset, etc.) into a memory.  Perhaps there is some permutation of buttons that allows this, but the manual is not helpful on this point:  Perhaps I've missed something, or a later software revision fixed this?
  

In short:  If you insist on using this radio (if you read on, you'll see why you probably won't want to!) you are best-off using a program like Chirp to set it and its memories up.

Receive sensitivity and "desense":

Two radios were tested - we'll call them #188 and #198 - and the results were very consistent.  The sensitivity of this radio on the amateur bands was very good (probably "too good"):  At under 0.15 microvolts the received signal was at least 12dB SINAD - but this comes at a cost:  The receiver is easily overloaded by strong signals on the same "band".  Badly, as it so-happens.

This radio was put on the test bench and it was given the "two tone" test in which a signal, modulated with a 1 kHz tone with +/-3 kHz deviation, was fed from a signal generator, through a hybrid combiner and into the receiver at approximately 0.5 microvolts ("almost" full-quieting - approximately 20dB SINAD) and another signal, unmodulated (into the other port of the combiner) was made variable.  In this test the "other" (unmodulated) signal was increased until the SINAD of the desired signal dropped below 12dB SINAD - a very obvious degradation.  Before we started this test, the "other" signal was checked with a spectrum analyzer to make sure that it was the signal itself and not its noise floor that caused the degradation.

This testing was done with the "other" signal separated from the desired one by 40 kHz (approximately 2 "channels" away), 100 kHz, and 1 MHz - and checked again at 10 MHz.  In all cases it was observed that the 1 MHz and 10 MHz "desense" values were pretty much the same, likely indicative of the inherent dynamic range of the signal path.

The results of this testing are as follows:

2 meters, test signal at 146.5 MHz:

• @40kHz separation:  >= -60dBm caused noticeable degradation
• @100kHz separation:  >= -60dBm caused noticeable degradation
• >=1 MHz separation:  >= -44dBm caused noticeable degradation

222 MHz, test signal at 223.9 MHz:

• @40kHz separation:  >= -65dBm caused noticeable degradation
• @100kHz separation:  >= -60dBm caused noticeable degradation 
• >=1 MHz separation:  >= -40dBm caused noticeable degradation 

70cm, test signal at 445.5 MHz:

• @40kHz separation:  >= -70dBm caused noticeable degradation
• @100kHz separation:  >= -65dBm caused noticeable degradation
• >=1 MHz separation:  >= -40dBm caused noticeable degradation

Note:  The above values will likely vary +/- several dB from unit-to-unit.

Interpretations:

In this area, there are quite a few mountaintop repeaters and if such a hypothetical 2-meter repeater were to have an EIRP of 100 watts, it would yield a signal greater than -60dBm within a distance of approximately 20 miles (about 32km) line-of-sight when a receiver was connected to a unity-gain antenna.  At 222 MHz and 70cm, the signal levels are similar, the typical repeater antenna's gain compensating for frequency effects.  What this means is that if you are listening on a frequency to a weak signal and a repeater a couple of "channels" away were to key up, it is possible that the signal to which you were listening would "disappear" due to receiver desense.

Figure 3:
Another view inside the radio.  Note the blockage of
the cooling fan - and the lack of something else...
The small potentiometer above and to the right of the large
chip (near the center of the image) sets the "Low"
transmit power.
Click on the image for a larger version.

In the case of the ">=1 MHz separation" case, these radios have a very broad receive input filter for each "band" meaning that a 2 meter signal won't particularly bother a 222 MHz or 70cm signal (unless it is very strong) - but any signal in that receiver's "band" coverage can cause issues.  For example, if you are listening to a weak-ish signal on 2 meters and a nearby transmitter on 159 MHz were to key up with a strong signal (stronger than -44dBm or so) it would likely cause degradation.

What's worse, there may be several such signals within the radio's currently-selected "band" that could combine their energy.  In other words, several such signals anywhere in the 137-174 MHz range would add cumulatively for a total power that could be significantly higher than any single signal.

To be sure, many radios made by the "Big Three" overload easily, but this radio is particularly prone to doing so in an "RF busy" environment where there may be other transmitters within a few 10s of MHz - such as a parade or other public service event.

Finally, one will notice that the "@40kHz" specs degrade with frequency.  The reason for this is unclear, but it is suspected that this may be due to limitations in the "all in one" receiver chip related to local oscillator phase noise and/or differences in the dynamic signal handling of this chip's on-board circuitry with respect to frequency.

Remember that the entire receiver ("IF" filtering, amplification, limiting, demodulation) is all done in the digital domain, on the chip with the received signal being digitized at some point:  It is likely that signals in close proximity with each other are being handled by different hardware filter types on the chip than widely-spaced signals.  I suspect that this chip uses onboard 8-bit A/D converters preceded with hardware filtering:  The hardware filtering probably helps at greater frequency separation, but is not effected in closer quarters.

Figure 4:
Almost the smallest fan that I've ever seen - but
does it do any good?
Click on the image for a larger version.

So, how good is this receiver overall?  In terms of absolute sensitivity it is fine, but in terms of handling "other" signals it is rather poor.  It is likely that this receiver would actually perform better in the real world if it were NOT quite as sensitive.  In other words, there is too much gain in front of the receiver section causing even moderately-weak signals to be strong enough to degrade performance:  Losing 5-10dB of gain in the signal path (after its RF preamplifier) would likely improve receiver performance in congested areas.

Interestingly, rather than seeming to generate "intermod" with strong signals within the radio's RF passband where a "new" signal is created out of the combination of several, when overloaded this type of receiver (an all-in-one chip using DSP techniques) seems to just go deaf, so the casual user may not be aware that there is a problem at first.  If there are multiple transmitters nearby (say, a number of operators at a public service event or at a shared site with other users) it is very likely that a radio like this will seem to randomly go deaf when other, strong signals go on the air - likely frustrating the user.

Output power:

The power output was also pretty close to what it should be.  The specifications oddly states "<=25 watts" - and this seems to be true:  At 15 volts, the output power was, in fact, a bit over 25 watts, dropping to 10-15 watts at 10 volts, depending on frequency.  Aside from the obvious problem with harmonics (mentioned below) that makes the legal use of this radio rather dubious, this wide voltage range (possibly) makes it a useful candidate for battery-powered portable operation - again, if it were actually legal to use on most of its "bands" as-is.

The "cooling fan":  Ineffective, but loud.

Mentioned several times now is the "cooling fan" - but it may not do much good.  Not only is this fan very tiny and incapable of moving much air (and amazingly loud for its size) - but there are some other problems:

Figure 5:
The tiny little fan is mostly blocked by the
aluminum casting and parts of the power amplifier that
are in the way.  With the fan's blades mere millimeters
away from solid objects, it gets amazingly loud for
something so small!
Click on the image for a larger version.

• The fan sucks - which is to say that it is set up to draw air through the case and exhaust it out the back - if that were possible:  There aren't any vents or holes in the case to allow this.  Even if there were holes, drawing air from inside the radio is not the most efficient way to cool nearby components unless airflow is carefully regulated. 

• As can be seen in figure 5, most of the hole for the fan is blocked by the aluminum casting and some of the power amplifier components.

In other words, the fan is largely ineffective (except at making lots of noise) and is probably there as much to make the user feel good when their radio gets too hot to touch.  There will be a very slight amount of air movement around the components at the back of the board near the fan opening, but this will more likely be due to blade turbulence than actual fan-induced air flow, something that contributes to the amount of noise that it makes. Clearly, the efficacy of the cooling fan would be better if air flow were directed over the fins of the heat sink rather than into an aluminum wall.

Initially, I thought that the fan was thermostatically controlled, but after testing on the work bench I realized that this may not actually be the case:  I'm thinking that the radio's computer simply winds up and down the fan speed slowly, depending on how long one transmits with it, irrespective of transmit level or temperature - but whether or not this is true remains to be seen.

Too much microphone gain?

While many inexpensive Chinese radios seem to have low transmit audio, this radio has quite the opposite problem:  Even holding the microphone about 2 feet (50cm) away from one's mouth and talking in a normal speaking voice caused the modulation to smash into the clipper pretty hard in both sample radios we tried.  (Clipping seems to be set to +/-4kHz when in the default FM "wide" mode.)

To be sure, having a bit too much audio is usually better than having too little, but the mic gain is so "hot" that your voice will sound a bit harsh and compressed - and everyone listening to you on the air will not only be able to hear everything that is going on in the room that you are in, but likely the bodily noises of any creature in your house as well.  In any but the quietest vehicle, road noise will be competing strongly with your voice causing challenges with intelligibility.

Unfortunately, there is no menu item to adjust microphone gain, but it should be possible to make a change in the microphone itself to reset the gain to something more sane.

The FM Broadcast receiver:

Like many of these Chinese radios it will also receive FM broadcast stations.  For this radio, connecting it to a typical amateur antenna in an area with fairly strong mountaintop transmitters located 15-20 miles (20-30km) away caused the (separate) FM broadcast receiver chip to be overloaded very badly, making it impossible to hear weaker "local" stations in the resulting muck:  It took about 30dB of RF attenuation to prevent the "FM broadcast" receiver from being clobbered and for the weaker signals to become audible leading us to believe that, like the main communications-band receivers, the designers likely put way too much gain in the front end.

Modes/situations in which this radio may not work:

This radio appears to be based on the same type of  "everything-in-one" chips that the Baofengs are based on - quirks and all.  Unfortunately, the nature of these chips - or at least the way that they are configured by the radio's processor -  preclude their use in a few situations/modes, such as:

• In areas of very strong adjacent-frequency signals or multiple transmitters - due to easy receiver overload.
• For packet operation - because of slow transmit/receive turn-around time.
• For any sort of DF (Direction Finding) system that uses switched antennas - due to the variable audio phase/delay properties of the receiver - a known problem with the "all in one" receiver chips found on many inexpensive Chinese-made radios.
• Any sort of high-duty cycle operation - due to the tiny heat sink which will get plenty hot, even at "low" power.

A "calibrated" S-Meter?

Interestingly, the "S" meter on the front panel  seems to indicate a 7-bit binary number - possibly from one of the chip's registers - that is proportional to the signal strength, each count being very close to 1dB, making it (potentially) more useful than a typical radios' S-meter.

Unfortunately its range (a bit more than 60dB) may not be entirely usable:  It doesn't start indicating meaningful values until the signal is about full quieting (between 0.5 and 1.0 microvolts) and it "pegs" at signal levels that would be commensurate with a transmitter several blocks away - and its update rate is fairly slow.

In other words, it's not useful for weak signals (you'd have to use your ear and listen for quieting for those) or moderately strong signals (e.g. a nearby transmitter) when it comes to direction-finding with a beam - although the latter could be mitigated with an outboard step attenuator.

(I am surprised that this thing doesn't also have a flashlight!)

* * * * * * * * * * * * * *

The real problem with this radio

Having buried the lead, the real problem with this radio is when you use it on-air:

If you transmit with this radio as-is on 2 meters or 222 MHz,  

you are breaking the law!

This is (literally!) the worst commercially-made radio I have ever seen in terms of harmonic/spurious output!

With any inexpensive Chinese radio (or any radio, for that matter) my first inclination is to throw it on the workbench and see how it really performs - which includes checking things like its sensitivity, power output, microphone gain, and spectral purity - and it is this latter point that made us catch our breath - this lesson having been learned when very cheap Chinese radios first appeared on the U.S. market about a decade ago.

The real problem was the actual transmitter specifications:  The literature states that spurious and harmonic energy is ">60dB" down - but it is not!

For 70cm, this radio seems to be "Okay" - but for 2 meters the results for this particular radio (let's call it "#198") were terrifying:  The 2^nd and 3^rd harmonics measured both as being as high as -23dBc - the precise values varying quite a bit with supply voltage.

Figure 6:
What appears to be the main CPU clock crystal:  "If it
doesn't fit, just cram it in there!" - but hey, what do
you expect for around $70?
Click on the image for a larger version.

Putting this into other numbers:  For an output power of 25 watts at 2 meters, this means that the harmonics are approximately 125 milliwatts each - roughly as much power as many handie-talkies produce when set to low power!

As an experiment we did something that we probably should not have done with this radio:  We connected this radio to an antenna that is designed for both 2 meters and 70cm and transmitted.

In this case we transmitted on a 2 meter frequency that was 1/3^rd of a local UHF amateur repeater located about 20 miles (30km) away on a mountaintop.  The result was that the 3^rd harmonic was full quieting into that repeater!

If one peruses the FCC rules you will spot FCC §97.307(e).  According to that rule, on a 2 meter transmitter of this power class we are allowed no more than 25 microwatts of spurious emission:  This radio exceeds that by a factor of approximately 5000 (about 37dB).

Remember:  This was the 3^rd harmonic of 2 meters which, if you are operating within the 2 meter amateur band, will always land somewhere in the 70cm band - but what about the 2^nd harmonic - which is just as strong as the 3^rd?  This would land somewhere in the 288-296 MHz range which is used for military communications - including aeronautical mobile.  What this means is that it is possible that your 2 meter transmissions made with this radio could be heard from, perhaps up to 100 miles away by an aircraft in line-of-sight.

What about 222 MHz?

The situation there isn't quite as bleak as the second harmonic was between 42 and 50dB down - the precise level varying wildly with power supply voltage.  Fortunately, any harmonic due to operation in the U.S. 222 MHz band (which covers 222-225 MHz) will land in the 70cm band (between 444 and 450 MHz), but its level will also be a bit high:  With 25 watts out on, say, 224.0 MHz the signal at twice this (448.0 MHz) will be around 1.5 milliwatts.

This may not sound like much, but this signal would be easily audible via line-of-sight at a distance of 10-20 miles (15-20km) - and it still can exceed the FCC rules by a factor of 63 (about 18 dB).

* * * * * * * * * * *

"Are they all this way?"

(Updated 4 February, 2019)

Without testing each unit as it comes from the factory, this question is impossible to answer, but I was able to obtain another unit (we'll call it "#188") and it was almost as horrifying (e.g. "slightly less terrible") in terms of its spurious output.  Because of minor component variations, one can expect an (essentially) unfiltered RF power amplifier to exhibit different properties in terms of spurious output - and these also vary based on temperature and power supply voltage.

I was able to put this radio on the RF bench, connecting it via a 40dB power attenuator (known to be flat within +/-1dB from <1 MHz to 1 GHz) and record spectrum analyzer plots, shown below with comments.

The spectrum analyzer had been calibrated to take the attenuator into account and in the plots below the power of the signals can be read directly as dBm, with "50dBm" (100 watts) being the top line with 10dB vertical divisions.

Testing on the 2 meter band:

Figure 7:
Radio #188 - which isn't as bad as #198, transmitting at 144.625 MHz, and 289.25 and 433.875 MHz - all with enough RF energy to be heard over line-of-sight distances of 10s of miles!
In this plot, the level indicated in the upper-right corner is that of the 4th harmonic.

Figure 7 shows the output of this sample transmitting at its out-of-the-box default frequency of 144.625 MHz with a transmit power of 25 watts (approximately +44dBm) with markers 2, 3 and 4 on the 2^nd, 3^rd and 4^th harmonics, respectively.  The measured output level of these spurs are:

• 2^nd harmonic @ 289.25 MHz:  +16 dBm (40 milliwatts)  41dB above FCC §97.307(e)
• 3^rd harmonic @ 433.875 MHz:  +11 dBm (13 milliwatts)  36dB above FCC §97.307(e)
• 4^th harmonic @ 578.5 MHz:  -2dBm (0.63 milliwatts)  23dB above FCC §97.307(e) (This frequency falls within off-air TV channel 32)

While these numbers aren't as bad as those of the first radio tested, the levels of the 3^rd harmonic are still capable of bringing up a line-of-sight UHF repeater from 10s of miles away!

Not mentioned previously is the 4^th harmonic which, in this case, lands in the UHF TV band.  What this means is that transmitting with this radio will likely disrupt nearby off-air viewing of whatever digital TV channel is on that frequency. (Affected off-air TV channel frequencies include channels 31-34 depending on the 2 meter frequency being used.)

How about the top end of the 2 meter band at 148 MHz?

Figure 8:
Radio #188, transmitting at 148.0 MHz as well as 296.0 and 444.0 MHz.
As mentioned previously, a signal of the amplitude shown in the UHF range is more than enough to key up a repeater!
In this plot, the level indicated is that of the 4th harmonic.

Translating the above:

• 2^nd harmonic @ 296.0 MHz:  +16 dBm (40 milliwatts)  41dB above FCC §97.307(e)
• 3^rd harmonic @ 444.0 MHz:  +15 dBm (32 milliwatts)  40dB above FCC §97.307(e)
• 4^th harmonic @ 592.0 MHz:  -5dBm (0.32 milliwatts)  20dB above FCC §97.307(e)  (This frequency falls within off-air channel 34.)

In this case the 3^rd harmonic is actually worse than at 144.825 MHz while the 4^th harmonic is 3dB weaker - but these differences are insubstantial in terms of legality.  It's worth noting that the 4^th harmonic is at a frequency where the low-pass filter is just starting to have its effect - which is why the higher-order harmonics are not really visible.

Again, we have significant energy in the UHF TV spectrum.

Testing on the 222 MHz band:

Figure 9:
Radio #188 being tested at 224.0 MHz.  The second harmonic is quite high, but at least it lands in an amateur band!  The 3^rd harmonic - which lands in the UHF TV U.S. "first responder" band - is probably strong enough to "blank out" reception on that channel in the immediate vicinity.  Even though the marker says "450 MHz", it is reading the power of the 2nd harmonic at 448:  The 3 MHz RBW and granularity of the wide sweep account for the offset.

Translating the above:

• 2^nd harmonic @ 448.0 MHz:  -2 dBm (32 milliwatts)  23dB above FCC §97.307(e)
• 3^rd harmonic @ 672.0 MHz:  -20 dBm (0.01 milliwatts)  5dB above FCC §97.307(e)

As with the other radio, the harmonics are lower than they were on 2 meters with the low-pass filter having a significant effect at the 3^rd harmonic - but we can see that it is still 5dB above where it should (legally) be.  In this case the 3^rd harmonic lands in the middle of the public safety band - but its "not terribly far from being legal" level is not likely to cause much of a problem.

On 70cm:

Here we have 430 MHz:

Figure 10:
Radio #188 again:  As you might expect, the harmonics from 70cm are farther down - but still "there".

Translating the above:

• 2^nd harmonic @ 860.0 MHz:  -16 dBm (25 microwatts)

Oddly,  FCC §97.307(e) doesn't list spurious/harmonic requirements for transmitters operating above 225 MHz so this is technically not illegal, but it's worth noticing that this power level is about 9dB above where it would be were the rules for the other bands to apply - and it also lands in the "800 MHz" cell/mobile band.

Going to the top of the 70cm band we see this:

Figure 11:
Radio #188 operating at 450 MHz.  Interestingly enough, the 2^nd harmonic is worse here than at 430 MHz.

Translating the above:

• 2^nd harmonic @ 900.0 MHz:  -13 dBm (50 microwatts)

It's interesting that the 2^nd harmonic is twice as strong on this frequency.

If we were to presume that the power amplifier's "natural" 2^nd harmonic energy is -30dBc (and that's being generous!) this tells us that the radio's low-pass filter is attenuating this harmonic by roughly 30dB - not really a very good filter.

But wait - there's more!

While testing the radio at 70cm, I noticed something else:  When the radio was keyed up, it would briefly output a wide spectrum of spurious signals all over the place.  This lasted, perhaps, 50 milliseconds - but it was definitely observable, as this "max hold" plot shows:

Figure 12:
Radio #188 transmitting at 450 MHz, this "flash" of spurious signals was briefly output at key-up by the ST-7900D.
This plot was captured using the "max hold" feature of the analyzer with several, repeated "key-ups".  This "feature" did not seem to be present on the 2 meter or 222 MHz bands.

The spectrum plot in Figure 12 looks frightening - particularly the brief "spur" at marker #2 which landed in the middle of the UHF TV spectrum (or in the public safety band in some parts of the U.S.) with a power level of nearly 100 milliwatts!  Also visible is a sprinkling of other signals - including a rather strong-ish signal in the 360 MHz area that has a power output of roughly +8dBm (approximately 6 milliwatts) - in the middle of the military comms band.

Practically speaking, such a brief "burst" isn't likely to cause much of a problem and quite a few older VHF/UHF transceivers made by "reputable" companies did this - but it is interesting nonetheless.

Comment:  No transmit testing was done in the "300 MHz" range as we have no intention of using it there.

* * * * * * * * * * *

"But there's an FCC logo on the radio!"

If you look at the radio and its packaging, you will find an FCC logo:  If you think that this automatically means that the radio is "OK to use" - you would be wrong.

No matter what the radio's specs say, what the reviews say, or what others say, if you are using a radio that, for some reason, does not meet the legal requirements - YOU are responsible, even if you didn't know that it doesn't pass muster!

Why is this?  Because you agreed to this when you got your license.

In other words, the onus is ultimately on you to make sure your gear is working properly - not the manufacturer - and if you happen to buy something that doesn't meet specs and get into trouble, it's ultimately your fault.  Now that you have read this, if you use one of these radios on 2 meters or 222 MHz, you have no excuse at all.

Practically speaking, the FCC certification does not mean that amateur gear is actually checked to see if its transmitter has spurious outputs or not:  If a piece of gear is checked at all it's usually just to see if it meets FCC Part 15 rules which typically cover spurious radiation caused by the receiver, its computer, or other circuitry - but not the transmitter, which is covered by Part 97, and being that amateurs are licensed under part 97, you are ultimately responsible for making sure that the gear that you are using is in compliance.

Having said that, anyone could make something and simply slap an FCC logo on it!

After pointing out the terrible harmonics produced by this radio someone commented to me: "Wow!  It's a tri-bander any time you key up on 2 meters!"


Why did they do this?

It's cheaper, of course!

In "older" radios it was common to have a separate power amplifier for each band - each with its own filter - but with today's inexpensive power RF MOSFETs a single amplifier like the one in this radio can work over a very wide range of frequencies - but this means that you must switch the appropriate filter inline for the band being used.

This switching is typically done with RF PIN diodes and/or relays - but either one of these options (particularly PIN diodes) gets to be pretty expensive (adding a couple of dollars to the bill of material) when you get into the 10s of watts at UHF frequencies.  Because this radio was "built to a price" it is almost inevitable that something was left out - and among those things that was omitted was proper low-pass filtering of the transmitter!

* * * * * * * * *

"I got one of these radios - can I modify it to make it legal"

The quick answer is NO, not if you use it anywhere other than the 70cm band.

The problem with this radio is that it seems to have only one low-pass filter after its (single) power amplifier.  On other radios (e.g. Yaesu, Kenwood, Icom) there would be a separate filter for each amateur band after the power amplifier to remove the harmonics for that band - but this radio seems to have just one - and it doesn't seem to have too much of an effect below roughly 550 MHz.

What this means is that this low-pass filter does absolutely nothing for any harmonics or spurious signals below roughly 500 MHz - and this is why the 2^nd and 3^rd harmonics of the 2 meter band and the 2^nd harmonic of the 222 MHz band is way out of compliance!

What if you continue to use this radio, anyway?  At least on 2 meters, the 2nd and 3rd harmonic signals are quite potent and may be heard from a great distance line-of-sight.  There is good news:  Because this is an FM radio, when someone using this radio IDs, their callsign will be clearly heard on these same spurious signals, so they should be easy to identify.

Work-arounds:

If you have one of these radios and wish to operate it legally on 2 meters or 222 MHz, you would need to do the following:

• For 2 meters, you use an outboard low-pass filter that will attenuate the 2 meter 2^nd and 3^rd harmonics by at least 40dB.

• For 222 MHz, you use an outboard low-pass filter that will attenuate the 222 MHz 2^nd harmonic by least 30dB.

What this means is that you would not be able to use this radio for transmitting on more than one band without having to swap out low-pass filters.

Using a triplexer as a low-pass filter for 2 meters and 222 MHz

While a low-pass filter specifically designed for 2 meters or 222 MHz or 70cm isn't a common item, you may find one disguised as an antenna "diplexer" or "triplexer" - a device used to combine/split different bands between radios/antennas.

• The 2 meter port of a 2 meter/70cm diplexer will work well as a low-pass for 2 meters.  The 70cm side - being a high pass - will not filter harmonics.
  
• A "222 MHz/70cm" diplexer will do the same there - functioning as a low-pass filter for 222 MHz.  Again, the 70cm side will not filter harmonics.
  
• If you are worried about 70cm, you will need a 70cm/23cm diplexer.
• As noted below, there is also such thing as a "triplexer" which will have a "middle" band.  If you use a 2 meter/222 MHz/70cm triplexer, you will get low-pass filter for 2 meters and 222 MHz on the separate ports.
  

Again, if you do use a diplexer or triplexer, note that you cannot it's highest band to filter harmonics:  For example, the 70cm port of a 2 meter/70cm diplexer - which is the highest band on the unit - is actually a high pass filter and will do nothing to attenuate harmonics.

Example of units that may be useful for both 2 meters and 222 MHz are the Comet CF-142, CFX-324A and CFX-514J triplexers which have a low-pass port for HF through 2 meters and a "mid-band" port - typically for 200-320 MHz.  The UHF port - being a high-pass - is not useful for filtering in this instance.  As I don't have these radios or triplexers available (I own neither) I'm unable to present measurements of the harmonics using them, but the "numbers" indicate that their levels should be more or less within legality and I seriously doubt that you would be able to key up a distant repeater with them!

As you would expect, you will have to manually move the radio between the "low" port for 2 meters and the "mid" port for 222 MHz. PLEASE NOTE:  There are different variants of the these devices - usually with different letter suffixes indicating different connector options (e.g. UHF, "N" and with/without pig-tails.)  If you get one, be certain that you get the correct connectors for your application.

Note:  

There is a also follow-up to this article describing a 2 meter/222 MHz low-pass filter linked here.

In short:

You cannot legally transmit with this radio "as is" on
the 2 meter or 222 MHz bands.

* * * * * * * * *

"I got one of these radios - what should I do?"

I would suggest that you not use it on other than 70cm without the use of an outboard low-pass filter.

If this isn't what you had in mind when you got the radio I suggest that you consider getting a refund from the seller as it is "not suitable for its intended use."

* * * * * * * * *

Answer to the question in Figure 1:  Actually, two of the frequencies shown - 245.625 and 350.025 MHz are not amateur frequencies!

 * * * * * * * * *

This page stolen from ka7oei.blogspot.com


[End]