Monday, December 19, 2016

On the winding of power chokes and transformers: Part 3 - The plate (high voltage) transformer

This is a follow-up of two previous posts in this series:
  • On the winding of power chokes and transformers: Part 1 - Chokes - link
  • On the winding of power chokes and transformers: Part 2 - A filament transformer- link

Using what we already know:

Figure 1:
Plate transformer with attached wires and end bells installed.
The windings and laminations are yet to be varnished or the end bells painted.
Click on the image for a larger version.
In the previous post of this series I described the design and construction of a filament transformer with dual 11 volt, 11 amp windings and a multi-tapped primary.  Building on the experience gained I felt confident to take it to the next step:  The design and building of the high voltage "plate" transformer for the (yet to be described) tube amplifier.

Based on the characteristics of the tubes to be used, the plate voltage needed to be "around 1 kilovolt" with each amplifier section requiring "about 100 milliamps" of average current, or around 200 milliamps, for the pair of channels.  Because of the experience gained in the winding of the filament transformer, we could use the design of the primary winding as a starting point.  For example, we know that to achieve a target magnetic flux of 1.4 Tesla and have the transformer be capable of at least 253 volt-amps and attain a rather conservative cross-sectional amperage of 0.4 amps/mm2 for the primary winding's we could use:
  • 17 AWG wire
  • Taps at 220, 229 and 239 turns for 115, 120 and 125 volts respectively, at 60 Hz
During the winding of the filament transformer's primary I observed that I could easily fit 41 turns of 17 AWG per layer.  This meant that the 239 turns only partially filled the final (fifth) layer, so we could afford to add a few more turns to the primary if necessary.

Two secondaries needed:

While the main secondary will be for high voltage, we will also need a 6.3 volt secondary to power the filaments of some of the driver tubes.  Because such a secondary will have relatively few turns we will need to calculate it first, for reasons that will become clear.

Using the "5% rule" we calculate that our 6.3 volt secondary will actually need to produce 105% of the desired voltage (6.3 * 1.05) = 6.6 volts to account for the drop under load.  Taking our 229 turn, 120 volt primary as a starting point we determine that the turns ratio to achieve this voltage would be (120 / 6.6) = 18.182:1 turns ratio.  With our 229 turn, 120 volt tap we would need (229 / 18.182) =  12.59 turns to obtain 6.6 volts.  

What this means is that for our secondary we should round the number of turns up (I'll explain why shortly) rather than down and with exactly 13 turns we end up with a primary-secondary turns ratio of (229 / 13) = 17.62:1.  From this we can calculate the actual, unloaded secondary voltage will be (120 / 17.62) = 6.81 volts - a bit higher than we'd like.

How do we fix this?  We should increase the number of turns on the primary to be able to more accurately obtain the desired voltage, but why increase the number of turns when we could also establish an accurate result by rounding down the secondary to, say, 12 turns and decreasing the number of turns on the primary to compensate?

You may recall that when winding a primary, the magnetic flux is has an inverse relationship with the number of turns.  Because the number of turns on the primary of the filament transformer was calculated to achieve the maximum target flux, we would not want to decrease the number of primary winding turns as that would increase that flux.  In other words, the main down side of adding a few turns to the primary is that each winding will need a proportional number of extra turns as well, taking up additional room on the bobbin:  If things are already tight, adding those turns could result in more wire than will fit.

Crunching the numbers:
  • Our voltage ratio:  120 / 6.6 = 18.182:1.  We already saw this number.
  • Since our 6.6 volt secondary should have exactly 13 turns, our 120 volt primary should have (18.182 * 13) =  236.4 turns, rounded down to 236.  This increase in turns reduces the magnetic flux from 1.4 to about 1.3 Tesla.
Clearly, a half a turn on the 120 volt winding has a fraction of the effect (18.182th, to be more precise) as a half turn on the low-voltage primary so we will round this down to 236 turns.  Let us now calculate the 115 and 125 volt taps:
  • 115 volts / 6.6 volts =  17.42:1 ratio.  13 turns * 17.42 = 226.46 turns.  I rounded this down to 226 turns.
  • 125 volts / 6.6 volts = 18.94:1 ratio.  13 turns * 18.94 = 246.22 turns.  This was rounded down to 246 turns.
Since we already know from when we wound the filament transformer that we can safely put 41 turns on a layer, we can see that for 246 turns we would need (246 / 41) = 6.0 layers - so we will go with that!

Designing the high voltage secondary:

If you are familiar with tube-type amplifiers you may have already have guessed from the voltage and current requirements that the plate impedance of the amplifier would be quite high:  10k ohms, to be precise.  The output transformers themselves are designed for single-ended triode operation with 8 ohm secondaries, rated for 25 watts (maximum) output.  Going through the math one can see that the turns ratio of this transformer is approximately √(10000/8) = 35.36:1.  If 25 watts RMS were being produced into 8 ohms, this implies that the RMS output voltage is around 14.14 volts, or almost exactly 500 volts RMS on the 10k primary which translates to 707 volts peak.

According to the specifications gleaned from the Edcor support forum (a link to the message thread may be found here) the maximum "safe" voltage across the primary and secondary windings would be 1000 volts.  Clearly, assuming a 10k primary impedance, 25 watts RMS of power and any reasonable plate voltage to achieve anywhere near this output power one will have to exceed this maximum voltage rating - unless a bipolar power supply is used where the high voltage is split - that is, the standing DC voltage between the primary and secondary is reduced to half.  To do this a full wave "bridge" rectifier is used with our choke-input filter network with the centertap of the transformer being grounded.

A final (loaded) DC voltage of around 970 volts for the plate voltage was (somewhat arbitrarily) decided as the target for the tubes that will be used - a reasonable compromise between the constraints of the output audio transformer voltage rating and the efficiency of the tube.  With this in mind let us calculate the actual, unloaded voltage for the secondary.

We know from when we designed our choke that at 200 mA there will be a 60 volt drop, so we will need to increase the output of 970 volts by this amount, which means that we will need (970 + 60) =  1030 volts.  Because the power supply will use a choke input we know that the loaded voltage of such a power supply is typically around 110% of the RMS voltage which means that for 1030 volts DC we will need approximately (1030 / 1.1) =  936 volts RMS.

Using the "5%" rule of thumb to take into account resistive loading of the primary itself we can calculate the actual, loaded voltage for the secondary, as in (936 * 1.05) =  982 volts, unloaded.  Using the 120 volt tap from the reference design we can now calculate our turns ratio and the number of turns, as in:
  • 982 volts / 120 volts = An 8.183:1 turns ratio.
  • 236 turns (at 120 volts) * 8.183 = 1931 turns which will be rounded down to an even 1930 turns so that the center-tap will be made at the 965th turn.
Based on the recommendations from the Turner Audio and Homo-Ludens web pages (see previous articles for the links) we can use a general rule of thumb of 0.33-0.35mm2/amp and since our current is to be 0.2 amps, we need a wire with the size of at least (0.2 amps * 0.33 mm2/amp) = 0.066 mm2.  Consulting our wire chart we see that 29 AWG has a cross-sectional area of 0.0642 mm2 resulting in a density of 0.321 mm2/amp - pretty close to our design goal.  As noted in the previous installment, Edcor seems to use a value of around 0.253 mm2/amp for their transformers and if this is applied our primary would be capable of (0.0642 mm2 / 0.253 mm2/amp) = 0.25 amps.

As it happens I had 29 AWG wire available when the choke was wound (it, too, was designed for 200mA) so this is the wire that I used.

Will it fit? 

At this point the question must be asked:  Will all of these windings fit on the bobbin?

We know from when we wound the choke that approximately 161 turns of 29 AWG wire will fit per layer, and with 1930 turns total, we'll need 12 layers.  With 29 AWG wire having an outside diameter (with insulation) of 0.33mm and the tape from each layer adding 0.05mm of thickness, each layer will occupy 0.38mm or, with 12 layers, 4.56mm of of bobbin "height". 

We also know from our winding of the filament transformer that one layer of 17 AWG wire plus 0.05mm of insulating tape has a total height of 1.274mm and with 6 layers (yes, I know that there is actually five full layers) that comes to 7.644mm.  Put together, the combined height of both sets of windings is 12.204mm - approximately 73% of the 16.5mm available bobbin height.

Figure 2:
Center tap of high voltage plate winding located in the middle of the winding
before Nomex insulation was added.
Click on the image for a larger version.
This figure does not include the low voltage secondary winding (one layer of 17 AWG, adding another 1.274mm) or the extra insulation that must be added between windings (approximately 0.5mm for each of the three) all of which adds another 2.774mm, taking us up to 13.704mm - about 83% of the available space.

While this will be kind of a tight fit, we ended up with the same sort of numbers when we built designed and successfully built the filament transformer so we can have good confidence that this, too, will work.

The winding:

While it may seem customary to wind the primary first, that may be just because most transformers that are seen these days are step-down, with the secondary winding handling more current than the primary and thus using larger wire.  It usual to place the smallest wire on the inner-most winding since it is more flexible and  easier to handle on the smaller-diameter "inner" layers of a bobbin, going around the square-ish corners and leaving the larger wire for later when the bobbin diameter is larger and the corners more rounded.

Following this convention a hole was "drilled" in the side of the nylon bobbin with a hot soldering iron and a piece of Teflon™ insulated wire was pulled through, attached to the start of the winding and then insulated with several layers of polyimide tape and a layer of Nomex™ paper insulation.  With that task completed the winding proceeded with care being taken on the first layer to assure both neatness and tight packing - the latter being done by pausing every few turns to slide the wire over to minimize the gap between adjacent conductors.

Figure 3:
End of the high voltage secondary winding, insulated with both
polyimide tape and Nomex ™ paper.  A loop was made in the wire which
brought out at a right angle from the other turns so that the tap would
not interrupt the continued neat, side-by-side windings.  Taps are
always made on the two sides of the bobbin that face the end bells
rather than the sides inside the core where the height is
is more limited.
Click on the image for a larger version.
The first layer done, a single layer of 0.05mm polyimide tape was placed over the top.  When I wound the choke I had only a single width of this tape available, but this time I had a selection of widths so as I proceeded with the layers, the location of the overlap and widths of this tape was changed with each layer to minimize "piling" of the turns where the tape overlaps which would later make it difficult to keep the layers even.

After a few hours of intermittent winding over several days - with each layer individually insulated with 0.05mm polyimide tape - the center tap was reached and for this a loop of wire was made in the conductor at right angles to the lay to which another piece of Teflon wire was soldered which was brought through the side via a hole made in the side of the bobbin with a hot soldering iron.  This joint was carefully placed in the middle of the flat side of the bobbin that would face outward from the core and insulated it with a few layers of polyimide insulation and Nomex paper to prevent it from damaging or being damaged by the pressure of turns in the layers above and below.
Figure 4:
Overlay of Nomex ™ insulating paper atop the finished high

voltage secondary winding before the top layer of polyimide
tape and its "creepage" insulation along
the sides of the bobin was added.
Click on the image for a larger version.

After a few more days of occasional winding the last turn was laid down, nearly filling the 13th and final layer.  I soldered to this a piece of Teflon wire and insulated it and the wire was brought out through the side of the bobbin and the entire secondary was covered with several layers of polyimide tape and 0.05mm Nomex paper.  As a final covering over the Nomex, another layer of polyimide tape was laid down, this time with the tape slightly going up the sides to increase the "creepage" distance between the primary and secondary - a sensible safety precaution, particularly with a high-voltage transformer!

Now, the primary...

The conductors of the primary were now laid down atop the insulated secondary.  As with the filament transformer the 17 AWG wire was brought directly out through the side of the bobbin and tucked out of the way:  The connection to flexible wire would be done later.
Figure 5:
The three "end" taps of the primary winding:  Top-left is the 115 volt tap,
below it is the 120 volt tap with the 125 volt finish on the left.  After
this picture was taken small pieces of Nomex paper and additional
tape were placed below and above the taps.
Click on the image for a larger version.

As with the start of any new winding the first layer of the 17 AWG primary was done with special care to make it neat and tight and each layer was individually insulated with 0.05mm polyimide tape.  When the 220th and 229th turns (for the 115 and 120 volt taps, respctively) were reached, loops of wire were put in the conductor, which was brought out through marked holes in the bobbin at right angles to the conductor.

With each tap being insulated with polyimide tape and Nomex paper where they crossed over other windings, the entire primary was then covered with several layers of polyimide tape and Nomex paper.  Again, a bit of insulation was brought up along the sides of the bobbin to provide extra "creepage" distance to provide good insulation for the 6.3 volt secondary to maximize both safety and reliability.

More about the 6.3 volt secondary winding:

Because it was on-hand, 17 AWG wire was used for the "6.3 volt" additional secondary.  With a cross-sectional area of 1.04mm2, we can calculate its current-handling ability:
  • Using the 0.33 amps/mm2 recommendation from the Turner Audio site, a safe current is:  (1.04mm2 / 0.33 amps/mm2) = 3.15 amps
  • Using the 0.253 amps/mm2 design Edcor guidelines a safe current is:  (1.04mm2 / 0.253 amps/mm2) =  4.11 amps.
Figure 6:
The completed winding - including the 13 turn, low-voltage secondary -
with the just-started core stacking.
Click on the image for a larger version.
Even in the worst-case scenario the addition of a 4.11 amp secondary would add only another 28 volt-amps of load to the transformer - well within its capacity.  Because this winding is on the outside of the bobbin and "exposed", it has good opportunity for cooling by convection and thus the Edcor rating would seem to be applicable - and 3-4 amps is plenty of current for several 6.3 volt tubes.

Comment:  If more current is needed it would be easy to add another parallel 17 AWG conductor to double its capacity.

As with the primary winding - which also used the same 17 AWG conductor - the ends of this 13 turn secondary were brought straight out the sides of the Nylon bobbin for later connection to flexible conductors and this additional secondary was overcoated with polyimide and polyester tape.

Finishing and initial testing:

With the addition of the low voltage secondary, all layers were over-wrapped with another layer of polyester tape to both secure and insulate the windings.  The transformer was almost ready to be tested!

Figure 7:
The stacked transformer undergoing initial testing with a
a variable transformer.
Click on the image for a larger version.
Although there are approximately 111 pieces of iron to be inserted into the core, the process is pretty easy:  Simply lay the bobbin on the table on one of the "outer" faces (where the taps are made and wires are attached) and alternately place the "E" sections atop each other.  With the "E" sections done, the transformer is then set on end to provide access to the vacant slots between every other "E" section into which the "I" sections were dropped.  Once these sections were added to one side, the bolts were slid through the laminations with the "I" sections to prevent them from falling out as I turned the transformer over and the "I" pieces were added to the other side.

With all E and I sections installed, a block of wood and a small hammer were used to abut the pieces of laminations against each other, a process that required several passes on all four sides.  With this done some nylon shoulder washers were installed (visible under the screw heads in Figure 7) to prevent the effect of eddy currents that might be caused by the "shorted turn" effect of the screw, and the bolts tightened.

Using a variable transformer the unit was then tested, first noting that the unloaded (magnetization) current was comparable to that of the previously-tested filament transformer indicating that nothing seemed to be amiss.  Very carefully, the high voltage secondary's voltage was then tested on each side of center tap and I noted that they were within a fraction of a volt of each other, and exactly at the calculated value with 120.0 volts applied:  491 volts on average.  I could not directly measure the 982 (unloaded) volts across the entire secondary since I have no voltmeter that is "officially" rated above 750 VAC.

After a test of the low voltage secondary, which was also measured to be at its designed voltage, I attached permanent wires and the end bells as seen in Figure 1 at the top of this page.  At this point the transformer  only awaits being dipped in insulating varnish - something that will happen after inital testing of the (yet to be described) amplifier prototype.

A future post in this series will describe the final steps in finishing these transformers:  Impregnation in "insulating varnish" and the final painting of the end bells.


This page stolen from "".

No comments:

Post a Comment


While I DO appreciate comments, those comments that are just vehicles to other web sites without substantial content in their own right WILL NOT be posted!

If you include a link in your comment that simply points to advertisements or a commercial web page, it WILL be rejected as SPAM!