Repairing a ("smoked") MFJ-998 1.5kW automatic antenna tuner

Figure 1:
The (repaired) MFJ-998 front, now working, sitting
atop my Heathkit SA-2060 manual tuner and underneath
my homebrew neon bar-graph VSWR/Power meter.
Click on the image for a larger version.

When I buy some types of ham gear second-hand, I'll treat it like as I would if I were to buy used Heathkit gear that had been put together by someone with "average" kit building experience:  I take the cover off, tighten the screws, reflow suspicious solder joints and do a visual inspection.  Regardless of brand, it's probably a good idea to pop the covers and take a look at gear that you buy second hand before using it.

For some reason when I recently bought a second-hand MFJ-998 on EvilBay, I didn't do that.

When this '998 arrived I did a cursory look to the case, connected it, and tried it out at 100 watts - and it seemed to work OK - but over the course of a few weeks being used at 100 watts, I noticed something odd:  It would occasionally start re-tuning during a transmission for no obvious reason and, perhaps, a faint whiff of something in the air - but I never connected the two and didn't investigate.  Using the same antenna, I'd operated using my old Heathkit SA-2060 (non-"A" version) antenna tuner for several years and hadn't noticed a changing VSWR that might explain this tendency.

Figure 2:
A of relay K2 from the top:  Evidence of
damaged glass-epoxy PC board material
is very clearly evident!
Click on the image for a larger version.

I checked for the usual suspects external to the tuner:  Loose connections, a branch touching the antenna somewhere - but there was nothing obvious.  This "problem" wasn't consistent, either - not happening often frequently enough to cause me to track anything down.

Several weeks ago I had occasion to run a bit more power and threw about 800 watts through it and things seemed OK for a while, but then it started retuning itself again - this time, accompanied by a very distinct burning smell.

I immediately pulled the tuner out of the circuit, going back to the manual tuner and things were fine once again, further indicating that the "instability" was related to the auto-tuner and not the antenna itself.  Later, when I had time, I pulled the cover off the MFJ-998 and immediately saw the problem:  As can be seen in Figures 2 and 3, the PC board was carbonized in the vicinity of relay K2 which is used to select antenna #1 or antenna #2.

Figure 3:
The damage to the PC board as seen from the
bottom side.  Sections of the board have become
carbonized, offering current paths to RF, causing
the degradation to accelerate.
Click on the image for a larger version.

Upon seeing this, I ordered some replacement relays - ten of them, as they weren't particularly expensive - and tuner sat around for several weeks until they arrived.  I did briefly consider just omitting this relay, "hardwiring" it for just one of the antenna outputs, but decided to proceed with the repair:  If this happens again, I'll reconsider doing this - or perhaps changing the way the relay(s) are configured.

What happened?

Clearly, the PC board material had "tracked" at some point:  A bit of leakage between the traces had obviously occurred and with the higher RF voltage resulting from my running higher power, this "slight" leakage had gotten very much worse, heating the board material, decomposing the epoxy in the PC board material and causing it to become conductive - and the it gets worse and worse from there.

I have the suspicion that a carbon tracking was present before I owned this tuner and it likely occurred due to attempting to tune an antenna that wasn't connected (possibly resulting in very high voltage), an intermittent antenna fault, or perhaps even lightning.  At the time it failed,  I had nothing connected to the connector or post of antenna #2 so I'm not sure why it so readily burned across the traces between the antenna connections - but it did.

Figure 4:
Damage to the original relay.  I don't think that
the relay itself failed, but rather that it
was damaged by the intense heat of the glowing,
carbonized board material.
Click on the image for a larger version.

Analysis of the damage

With the "tracking" between the tuner and the antenna "A" and "B" connections, the circuit board was nearly burned-through in a few places and the relay was destroyed - but it looked as though the damage of the relay was caused by the heat from the (burning!) PC board.  Despite looking really bad, the damage was very localized - and it provided an opportunity for improvement.

The fact that the damage was worse on the bottom of the board than the top also indicated that it was likely on the bottom side that the issue first started.

What to do?

In many cases - when high voltages are present across a section of PC board - manufacturers will place an "anti-tracking" slot between the two points:  Rather than rely on the surface of the PC board to withstand high voltages - the ability decreasing if moisture, dust or other contaminants are present - a physical slot is cut in the PC board material between those connections.

Figure 5:
Using a rotary tool to remove ALL carbonized PC board.
All potentially-conductive board material must be removed
or else the same thing will happen again!
Click on the image for a larger version.

This technique is frequently found on mains-powered devices that have human contact - such as phone chargers and other power supplies - where excess leakage between the high voltage from the wall plug and the low voltage output could result in injury or death if someone touched an accessible metal contact.  While this isn't a "life-safety" issue like a power supply, it would have made sense, given the high voltages that are possible, to implement such measures here.  For "reasons", MFJ did not choose to manufacture the PC board with such "anti-tracking" slots in this particular location - on that is very likely to have the highest voltages present across adjacent contacts to be found in the tuner.

Using a small routing bit in a rotary tool I ground away ALL of the carbonized (conductive!) material:  Not doing so would have risked additional "tracking" in the future where it could have found a new path.  The result of this work can be seen in the photos - a bit of "Swiss-cheese" of the circuit board - but now, there was only air between the contacts across which there was likely to be high voltage.

Installing the new relay

With all of the carbonized material removed, I vacuumed up the debris and cleaned the area around the relay on both the top and bottom side with alcohol, removing deposits from the soot of the burning PC board.  I repeated this process after the new relay was soldered in.

Figure 6:
After cleaning the board with alcohol, the new.
relay was installed and the connections made
using 16AWG wire.  The "air gap" between
pins should make it more resistant in the future.
Click on the image for a larger version.

On this relay, the "common" (armature) pins are in the center with the normally-open and normally-closed pins on the sides - but since I removed most of the PC board material under the relay, the "common" connection was completely missing.  This problem was easily solved using a piece of tinned wire as can be seen in Figure 6:  The air gap between the other relay contacts was maintained.

I was fortunate that the connections for the coil (the two solder pads near the top of Figure 6) and both the "Normally Open" and "Normally Closed" terminals were still intact (e.g. the board wasn't burned) and this provided a solid mounting for the relay.   It was only the "common" relay contacts - those that connected back to the tuner itself - that were no longer extant so I folded a piece of 16AWG copper wire and made the connection back to the remaining PC board trace as seen in Figure 6.  (Note:  This relay, K2, is a DPDT relay of the same type as all of the other relays and both of its sections are connected in parallel.)

Testing and comments

It worked!

As there was no damage to any other circuitry, the relay properly selected between antennas 1 and 2 as designed and the memory pre-sets (on the bands other than where I was operating when the failure occurred) were just as they were indicating that the matching conditions were identical to before.

In doing research on the relays used in this tuner (all of them were identical - Hui Ke HK14FH-DC12V-SH - which is the same as the American Zettler AZ576-1C-12DE, the Songle SMIH-12VDC-SL-C and many others ) I noted a few things about their specifications.  As expected, they have 12 volt, non-latching coils (e.g. power must be applied for it to hold the tuning configuration) and their contacts are rated for 16 amps (resistive) for DC and mains-frequency AC, and they also have good isolation between the contacts and the coil (rated for 5kVAC at mains frequency).

What did concern me a bit was the fact that they have only a 1kVAC rating between "open" contacts - and a quick check with a "Hi-Pot" tester verified that it did, in fact, break down (arc internally between the armature and the normally-open contacts) at about 1.7KVDC.  It would seem that the designers of this tuner considered this aspect of the relays' limitations to an extent as all of the capacitors in the tuner are switched in/out using two relays in series to accommodate higher voltages - but this was not applied to K2, the antenna switching relay.

A quick check of the ratings of this tuner indicate that it is rated for up to 1600 Ohms at full power (1500 watts).  Knowing the power and impedance - and presuming a resistive load - we can use the following formula:

V = √(P x R) 

Where:

V = Voltage

P = Power in Watts

R = Resistance

and based on this, at 1500 watts and 1600 Ohms, the voltage in a purely resistive load would be about 1550 Volts RMS (or about 2200 peak volts) - well above the known breakdown voltage of the contacts of the relay.  It's worth noting that it's often the case that at radio frequencies, insulation and breakdown ratings are lower than they might be at mains frequencies and DC - something else to consider!

What this implies is that under such conditions there would be enough voltage for the armature of the relay to arc to those of the unused antenna and, perhaps, the designers of this tuner should have put two relays in series for antenna switching, too.

What about the other extreme, where the impedance is low?  The tuner is rated for as low as 12 Ohms where the voltage would be lower and current higher and here we use a different formula:

I = √(P / R) 

Where:

V = Voltage

P = Power in Watts

R = Resistance

 

In this case 1500 watts and 12 Ohms (resistive) yields a current of about 11.2 amps - somewhat lower than the relay's contact rating of 16 amps, plus there's the fact that both sets of contacts of this relay are in parallel which further increases their durability.  At such high currents, the concerns aren't necessarily with the relays, then, but rather potential I*R heating of the inductors and capacitors whenever an extreme match condition is encountered.

 

In short:  If you are running anywhere near 1500 watts and have a high impedance being presented to the tuner by your antenna (e.g. using something like a half-wave end-fed antenna) - you should probably reconsider your arrangement!

 

* * * * *

 

This page stolen from ka7oei.blogspot.com

 

[END]