Using a toroidal transformer as a mod transformer Part II

[AB2EZ]

Several months ago I posted my results (which became a very long thread) on using a toroidal filament transformer as a modulation transformer with my Johnson Ranger. This is a follow-up post on my results in scaling that experiment up to full legal limit.

First to recap what I did with the Ranger in the earlier experiments:

Externally modulated
40 watts output at carrier (500 volts plate voltage x 125 mA plate current => 62.5 watts input power)
Modified Heising configuration: 20H choke, 1uF capacitor (essentially no unbalanced current in the toroidal modulation transformer)
Toroid = 25VA Antek filament transformer wired for 12 volts in and 240 volts out (20:1 step up)
Audio source was my regular audio chain as input to a Samson "Servo 120a" audio amplifier (using one of the two channels)

Results: excellent (on-air tests revealed that listeners could not tell the difference between the Antek modulation transformer and a Hammond 1629SEA output transformer).
Frequency response essentially flat between 50 Hz and 7000 Hz (audio input => demodulated off-air monitor output)

Scaling up to legal limit

I'm gathering the components needed to scale the earlier experiments up to legal limit. I plan to plate modulate my homebrew amplifier (currently configured as a linear amplifier), which uses a pair of Russian GS-35b triodes.

Target:

Class C operation
Plate voltage (at carrier): 1600 volts
Plate current (at carrier): 300 mA
Power input: 480 Watts
Power output: ~ 375 watts (78% efficiency)
Modulation impedance: 5333 ohms
RF load impedance: 2667 ohms

Heising reactor (50 H 300 mA Peter Dahl NOS)
Capacitor: 1 uF (must be rated at > 4500 volts)
Required audio power ~ 240 watts (for 100% positive peaks)
Required audio voltage at transformer output: 1600 volts peak

Desired step up ratio to present a load to the audio amplifier of between 4 and 8 ohms: roughly the square root of (5333 ohms / 6 ohms) = ~ 30:1. [For 8 ohms: ~26:1]

Modulation transformer:

I decided to use off-the-shelf Antek transformers, available from his Ebay store. I am using two (2) step up transformers in tandem (which means that any artifacts from the transformers will accumulate). A single, smaller transformer (300VA rating) would be cheaper, but he doesn't have one, off-the-shelf, that also has the output voltage rating I need. [If the output voltage rating is not high enough, then the core will saturate at low modulation frequencies (v ~ dB/dt).]

Transformer 1 (as specified):  63VAC input => 115 VAC output (rated at 800 VA)
Transformer 2 (as specified):  115 VAC input = 1600 VAC output (rated at 800 VA)

Tandem combination: 63 VAC input => 1600 VAC output (rated at 800 VA), implying a step up ratio of 1600/63 ~ 25.4:1*

 *As measured with audio applied at 400 Hz: 2.19 volts in : 55.3 volts out => measured step up is 25.3:1

Initial results:

I decided to try the new transformer with my Ranger... i.e., substituting the new transformer for the 25VA transformer I used previously. Otherwise, everything is the same as my previous experiments.

The results are excellent. The demodulated output waveform from the off-air monitor overlays perfectly (after scaling and offset) on the audio waveform from my audio chain... for frequencies between 50 Hz and 7500 Hz (the Ranger's plate and screen bypass capacitors are the limiting factors at higher modulation frequencies). I.e. the modulation is linear, and has a flat frequency response from 50 Hz (lowest frequency measured) to beyond 7500 Hz.

Obviously, with these transformers, there was no problem reaching 100% modulation.

While I have yet to try this at legal limit power, the fact that it works so well with these transformers in combination with my Ranger implies that there is no problem with the parasitic component values in the larger transformers (leakage inductance, capacitance). This is what I expected from some back-of-the-envelope calculations... but its nice to see it in practice.

Next step (probably next month): plate modulation of my GS-35b amplfier at full legal limit

Best regards
Stu




 

[W1VD]

Interesting info - will be watching for your test results.

I notice that Antek now lists a couple audio transformers that will be available starting September. Assuming your tests work out as well as expected, it might be interesting to see if they would consider entering the mod transformer (and perhaps even reactor) market. The toroidal design would make it relatively easy to offer a 'multimatch' setup.

Bought several big power supply transformers from them for LF and MF transmitters. They are excellent quality, delivered quickly and work exactly as advertised. E mail correspondence indicates that they will do custom work although I haven't gone that route.


 

 

[w8khk]

Hello Stu,

I find your mod transformer testing very interesting.  Since you are also using a mod reactor, isolating the dc current from the transformer secondary, I was wondering if a simple intermediate test would be valid.  Instead of strapping to the GS-35B amplifier, suppose you load the mod transformer with a 5000 ohm resistance, and tap a segment of the load with the scope and distortion analyzer, then sweep the audio chain.  Would this method verify the modulation transformer performance, or are there other issues to consider?  I was considering using this method to test my modulation equipment before fully integrating all the components.  Obviously, this method would not verify the audio chain is immune to RF feedback, but for preliminary testing it would limit the number of unknown variables. 

73, Rick

[w4bfs]

Hi Stu ... just excellent ! ... that is one of the things I really like about this group ... excellent innovation and sharing skills ... reminds me of working at GE in the good ole days ... one question... what are hi-pot ratings for these torroids, particularly winding to winding ?

[AB2EZ]

Jay
Rick
John

Thanks for the feedback and the words of encouragement!

Rick: Right now, I am limited by the output power of my audio amplifier. I'm feeling a little "buyer's remorse" about the cost of the transformers and the choke. When that passes, I'll probably purchase an amplifier that can put out around 600 watts in bridged mode. Then I can try what you suggested.

John: Antek's specification says that their transformers are hi-pot tested at 3500 volts:

"Dielectric Test:
TEST CONDITION: Apply dielectric meter between primary coils and secondary coils; and increase voltage up to 3500VAC. No initiate any spark."

Nevertheless, I intend to be very careful in how I perform the full power test.

Stu

[WA1GFZ]

Stu,
I think you may flash it over. Consider a homebrew primary over the stock windings fc

[WA1HZK]

Just checking around.
What if you had a nice custom KW+ donut.
And you tied one of these amps to it:
http://www.bswusa.com/proditem.asp?item=FR2500
That could be a real cheap modulator when you consider the tubes you do not need to feed. Very efficient also. This X (2) Russian toobs? = SNOT!!
Keith

[AB2EZ]

Keith

Sorry for the confusion.

The audio amp you pointed me to is very nice!

The Russian triodes are the RF tubes in my RF amplifier (currently configured as a linear amplfier, and destined to be converted to plate modulated, Class C operation)... they are not being used to build an audio amplfier.

Stu

[WA1HZK]

I know. I'm thinking of easy ways to plate modulate a large final!

[WD5JKO]

  Stu,

    Years ago I did a similar solid state amp to a tube final. I used a Crown M600 to a Viking I 4D32 RF amplifier. I played with an Ultra Modulation circuit, and even went a step further. I had over 600 watts of audio to play with, so I used some to increase the 4D32 B+ (raise the carrier power) so that the carrier would pump up to 200 watts out prior to modulation. This was simple to do with a variac, a step up transformer (1:4) a FW bridge and a 100 uf capacitor. The DC made from audio was put in series with the unmodulated B+ for the 4D32. The B+ went from 750 to 1250 with a whistle into the mic. The 4D32 is so stout it modulates linearly upward to 800 watts PEP output this way with an unmodulated carrier at 100 watts. The rig was VERY effective and sounded good. Overmod in the negative direction was prevented with a 4-2-1 progressive negative cycle attenuation (not loading) using 3 bias supplies, diodes, and a bunch of power resistors. The system was transparent and did not kick in until 80% negative modulation of a 200 watt carrier.
     I will caution you on one thing. The waveform of your voice will likely be asymmetric and therefore have a DC offset to it. If you use a BIG direct coupled solid state amp into a MOD transformer directly, the DC component in your voice will result in a huge DC current into the mod transformer primary. This may cause the amplifier protection circuits to kick in. To get around this problem with my M600 amp, I added a capacitor in series with the amplifier output. In my case I used two 10,000 MFD 100V capacitors in series (+ to +) to essentially make an AC capacitor. I tried with and without diodes across each cap (cathode to + side) and the diodes made no difference, so I left the diodes off.

Regards and good luck,
Jim
WD5JKO

[AB2EZ]

Update:

I have everything looked together... and with the audio amplifier I have on hand (120 watts peak into 8 ohms in bridged mode) I was able to comfortably modulate about 30-40% with 400 watts of input power (1600 volts and 250 mA) and 350 watts of output carrier power. See the attached photo

Note: 1600 volts / 0.25A = 6400 ohms. My step up transformer ratio is ~ 25.3 :1... so the impedance that the audio amplifier is driving into is  6400 / (25.3 x 25.3) ohms ~ 10 ohms. As a result, the amplifier (which acts roughly like a voltage source) probably can only put out 120 watts x 8/10 = 96 watts into the load it is looking at.

It looks good on the scope (The output of my off air envelope monitor is tracking the audio input to the audio amplifier)... and it sounds like my transmitters normally sound when I listen to it.

I gave a lot of thought to the issue of dumping the energy stored in the Heising choke when switching from transmit to standby. I finally decided that the energy would, in fact, be dissipated in a safe way (no big voltage spikes on the choke or the step-up transformer) in the audio amplifier. The DC choke current is diverted into the secondary of the step up transformer... which, in turn, produces an opposite current in the primary*... which flows through the output resistance presented by the audio amplifier. Since the toroidal step up transformer has very low leakage inductance... the transition of the current in the Heising choke from flowing into the plates of the rf tubes to flowing into the secondary of the modulation (step-up) transformer should be very fast. There is a brief period (probably less less than a few microseconds or so) when the current flows into the .0044uF bypass capacitor (to ground) at the modulated B+ side of the plate rf choke. The current can also flow briefly through the output rf coupling capacitor (.0066 uF) to ground via the tank coil and the safety choke. When that is occurring, the resistance of the rf choke also provides some absorption of the stored energy in the Heising choke. I tried to measure the voltage across the secondary of step up transformer when I switched the transmitter from transmit (~300 mA through the Heising choke) to standby. I didn't see anything larger than about 1300 volts... and the associated transient waveform was rather complicated... including ringing at around the resonant frequency of the tank circuit . The way I measured this was to measure the transient waveform across one of the 6.3 volt windings on the secondary of the step up transformer, rather than measuring across the 1600 volt winding of the secondary. I scaled the waveform I observed on the 6.3 volt winding by 1600/6.3

[Come to think of it, the combination of the 0.0044 uF r.f. bypass capacitor and the 0.0066uF rf coupling capacitor are providing a total of a .011 uF path to ground in parallel with the tube's 5400 ohm modulation resistance. These are diverting significant modulation current for frequencies above 2680 Hz... so I need to reduce both of them to .001 uF.]


*The primary current is larger than the secondary current by a factor equal to the turns ratio of the transformer.

In the picture (don't do this at home with high voltage circuits):

A. the larger toroid is the Antec 800VA, 115VAC input : 1600 VAC output transformer

B. the smaller toroid is the Antec 800VA, 63VAC input : 115 VAC output transformer

C. the cardboard file box contains the plate transformer I use with this linear amplifier (now running as a Class C plate modulated transmitter... with about 120 VAC across the primary of the plate transformer, instead of 240 VAC)

D. the 50 Henry, 300 m.a. Heising choke is sitting on top the the cardboard box (covering up the "warning, high voltage" sign)

E. to the left of the Heising choke, on the wooden base, is a row of ten (10) 18 uF 450V electrolytic capacitors, in series; each with a 100k ohm 5 watt balancing resistor across it. The resulting Heising capacitance, 1.8uF, is optimal in conjunction with the 50 Henry choke... given my target of 1600 volts and 300 mA at carrier. I.e. 5400 ohm modulation resistance. This capacitance resonates with the Heising choke at 16.8 Hz. At 16.8 Hz, both the capacitor and the choke have reactances of ~ 5270 ohms.

[w4bfs]

hmmmm   hmmmmm  doughnuts

[w1vtp]

Stu

I am personally excited about this approach.  The biggest frustration of getting "iron" for an AM station is the lack thereof and now (especially in Ebay) Iron has somehow been touched by the Loadstone and have been turned into gold or some such precious metal.

The building  of an AM modulator based on currently "off the shelf" components open up new possibilities for the AMer.  Good luck on the venture and I'll be following your progress carefully.

Another great approach is the use of commercial reasonably inexpensive AF amplifiers.  The line of AMP you are using offers power all the way up to over 300 watts.  So more exotic amplifier design, should one want to avoid that exercise, could be avoided.

73, Al

[WA1GFZ]

Seems to me Steve QIX built a couple solid state modulators using FETs and transformers. I would think it easier to do than the class H.
All you need is a phase splitter FET driving two banks of output devices.
A bias supply and your in business. This way you can put the stereo amp back in the living room. gfz

[AB2EZ]

Frank

I wonder if I opened up the Samson 2-channel amp...

Would I find:

"a phase splitter FET driving two banks of output devices" 
 
Stu

[AB2EZ]

It works 

I was able to get a hold of a Samson Servo 600 (600 watts into 8 ohms, bridged) audio amplifier to complete these experiments.

With that amplifier, I could easily obtain 100% modulation, with a nice clean sine wave, at full legal limit power ... with a flat modulation frequency response and no visible distortion from 30 Hz to 9.5 kHz (as far as I measured at each end).

As a "bonus", the pair of GS-35b's run in a Class C plate modulated configuration with very high efficiency:

Input electrical power: [1700 volts of plate B+ x 220 mA of average plate current] + 30 watts of drive power + 50 volts of peak grid-to-cathode voltage* x 225 mA of average grid current]  = 415 watts of electrical input power

Output power (Bird power meter: 360 watts, "Powermaster" power meter: 314 watts ): 337 watts

Efficiency = 337/415 x 100% = 81%**

*The grid current flows when the grid is positive with respect to the cathode. The negative bias is 40 volts, and I am estimating that the average input power contributed by the grid current is ~50 volts x the average grid current ~ 11 watts.

** This number seems high, but the amplifier is biased heavily into Class C at carrier... with -40 volts of grid-to-cathode bias and only 1700 volts on the plates of the GS-35b's ... which would produce very high efficiency. I.e., the plate current flows during a small percentage of the rf cycle... and during that part of the cycle, the plate voltage is at its minimum value.

I am using a Johnson Ranger (running at 30 watts of carrier) to drive the plate modulated rf power tubes (a pair of GS-35b's) in grounded grid configuration***. I found that I could obtain significantly improved linearity by also applying some plate modulation to the Ranger. I.e. about 20% plate modulation of the Ranger.... just enough to get equal positive and negative peaks on the output of the rf power amplifier, with a nice, symmetrical sine wave on the output of my off-air envelope monitor.

*** In grounded grid configuration, the load impedance seen by the Ranger varies (as the modulation of the GS-35b rf power amplfier goes from 100% negative to 100% positive) from infinite -to- one half the load impedance seen by the Ranger at carrier. On positive modulation peaks of the GS-35b rf power amplfier... the reduced rf load impedance seen by the Ranger is translated by the Ranger's pi output network into a doubling of the plate load resistance on the Ranger's 6146. I.e. the load resistance on the Ranger's 6146 r.f. output tube is proportional to Q**2 x r, which is proportional to L**2/r, where Q is the Q of the tank circuit, r is the load resistance at the output of the tank circuit, and L is the inductance of the tank coil.  In principle, if the Ranger had more power output capability (at carrier), and was lightly loaded (rf output voltage swing << B+)... then the Ranger's r.f. plate current waveform would remain approximately the same... and the output power of the Ranger (looking into a higher load resistance) would double. However, with the Ranger fully loaded, the opposite is true... i.e., the Ranger's output power actually decreases (or, at best, remains approximately the same) on positive peaks of the modulation of the rf power amplifier. This results in a reduction, by a factor of more than 40%, of the peak rf drive voltage the Ranger delivers to the input of the GS-35b power rf amplifier (i.e. less power into a load resistance that is roughly twice as large).

Needless to say... I am pleased with the results.

This completes my experiments with the use of toroidal power transformers as modulation transformers.

Best regards
Stu

[w4bfs]

looks really good Stu .... I am still considering winding to winding voltages ... if some way could be figured out how to establish a equal common mode gradient between the transformers, it might be possible to have some 'comfort zone' ...would you mind to measure winding to winding capacity on both cores and see what you get? thanks ...John

[w1vtp]

It works 

I was able to get a hold of a Samson Servo 600 (600 watts into 8 ohms, bridged) audio amplifier to complete these experiments.

With that amplifier, I could easily obtain 100% modulation, with a nice clean sine wave, at full legal limit power ... with a flat modulation frequency response and no visible distortion from 30 Hz to 9.5 kHz (as far as I measured at each end)...... [text deleted]

.....Needless to say... I am pleased with the results.

This completes my experiments with the use of toroidal power transformers as modulation transformers.

Best regards
Stu

Excellent!  This certainly has changed my approach to my future legal limit HB AM rig.  I will be scrapping the idea of using tubes as a high power modulation source.  My plan would be to connect a solid state high power AF amp to an intermediate transformer to get the 8 ohms to ~ 500 ohms.  From there my UTC CVM-4 will take over and match just about anything I build by way of RF deck.  Another nice outcome of your experiment is the demonstrated use of a GG triode amp.  No more neutralization issues.

Are you going to have this setup on this weekend?  I wanna here it

73, Al

[Steve - WB3HUZ]

I am using a Johnson Ranger (running at 30 watts of carrier) to drive the plate modulated rf power tubes (a pair of GS-35b's) in grounded grid configuration***. I found that I could obtain significantly improved linearity by also applying some plate modulation to the Ranger. I.e. about 20% plate modulation of the Ranger.... just enough to get equal positive and negative peaks on the output of the rf power amplifier, with a nice, symmetrical sine wave on the output of my off-air envelope monitor.

I would think this would be required since the drive power feeds through to the output. With no modulation of the drive power you could never achieve 100% modulation (there would always be unmodulated power in the output).

[WA1GFZ]

Very cool Stu and I like the test set up on the floor.
my old 4-1000A had a stack of Zeners for bias and I could bias it well into cutoff. I also measured high efficiency in class C but it takes more drive.
I wonder if a 2  or 3 dB pad between them would help the Ranger. 

[AB2EZ]

Frank

I was thinking the same thing. If I hadn't been able to get such good results by modulating the Ranger's output... my next approach would have been to use my base station transceiver to drive the grounded grid plate modulated amplifier and a 50 ohm dummy load in parallel... i.e., seeing if I could partially "swamp" the input of the grounded grid rig to reduce or eliminate the undesirable effects of the varying (with modulation) input impedance presented by the grounded grid configuration. The Ranger doesn't have quite enough output power to drive the grounded grid rig with a 50 ohm dummy load in parallel with its input... although it would come close to doing so. I probably would have done some experiments using this approach with the Ranger as well.

Steve

You raise an interesting issue regarding the effect, on the modulation of the output envelope, of the contribution of the input drive power to the output power.

The waveform of the output of my off air monitor (envelope detector) showed good negative peaks (up to 100%) but compressed positive peaks (significantly smaller than the negative peaks)... suggesting that the problem was associated with the drive voltage dropping on positive peaks.

I agree that when you load up a grounded grid amplifier... you can load it with a larger output tank circuit impedance at the operating frequency (for a given amplitude of the component of the current at the operating frequency that is flowing through the tank circuit)... before the voltage between the plate and the cathode goes to zero on rf peaks (because the drive voltage adds to the modulated B+). In that sense, the drive power can add to the maximum rf output power that is obtainable, and not just add to the the total input power (which it always does). 

However, I'm not so sure about the impact the above has on ability to modulate the rf output. Since the voltage between plate and cathode is increased by the presence of the drive signal (between cathode and ground), i.e. the cathode is negative with respect to ground, I believe that you can simply modulate the B+ with an audio signal whose amplitude is larger than the B+. On 100% positive modulation swings, the added modulation (voltage) would equal the sum of the B+ and the peak value (amplitude) of the rf drive voltage. On negative swings, the added modulation (negative voltage) would be equal to minus this same sum. It is a little non-intuitive for the plate to go negative with respect to ground... but the cathode is also negative with respect to ground ... and both are sine waves, in phase at the operating frequency.

Al

Thanks for the kind words.

Remember, the key issue I was exploring here is the ability to use a modern, off-the-shelf, ferrite toroidal power transformer as modulation transformer... because of the associated physical characteristics of those particular types of power transformers. I've never tried using other types of power transformers as modulation transformers... but I understand that there are performance problems (saturation, frequency response, etc.) associated with using traditional (i.e. not ferrite) power transformers as modulation transformers. I know that Timmy (WA1HLR) and many others have been able to use modified, traditional power transformers as modulation transformers... with varying degrees of resulting performance.

Also, these findings are not limited to applications involving solid state amplifiers or 8 ohm (output) audio amplifiers. Using one of these toroidal transformers with a vacuum tube modulator ought to work just fine... provided you don't put unbalanced DC through it. The issues are: to get the turns ratio right, and to use a toroidal transformer with a large enough VA rating (heating) and a large enough output voltage rating (saturation of the core). Most toroidal power transformers have a pair of secondary windings which can be used in series or in parallel. Therefore a push-pull audio modulator (no unbalanced DC in the primary) in a modified Heising configuration (no unbalanced DC in the secondary) should work.

John

I don't think I understand what you are asking me to measure. I could try to measure the capacitance between the primary and the secondary of one of these transformers with both windings not connected to anything... but I'm not sure what that measurement result would be useful for. Since the ferrite core has a very high permeability, leakage inductance is not a big problem... so the thickness of the  insulation on the wires, and the thickness of any insulation between the layers/windings can be larger.  Perhaps we can correspond privately. My E-mail is sdp2@verizon.net


Best regards
Stu

[N3DRB The Derb]

D. the 50 Henry, 300 m.a. Heising choke is sitting on top the the cardboard box (covering up the "warning, high voltage" sign)

same as it ever was..... same as it ever was........

[WA1GFZ]

Stu,
Most low frequency toroid cores are tape cores. From my early days of switchers I remember the thinner the tape the better the high frequency response. So this could cause a run on variac cores. Imagine winding a homebrew transformer or choke  on a 240 volt 30 amp variac core....or a 20 amp 115 VAC core.....bring out your dead.

[w4bfs]

Hi Stu ... after reconsidering construction techniques, I am dropping the capacity meas request .... I am assuming both primary and secondary are essentially scramble wound ... I wonder if there is extra insulation between windings ... if not, it may be necessary to wind your own primary ... if you don't mind, try winding a known # of turns say 25 and exciting it with fil xfmr on a variac to estimate # of secondary turns, etc .... I'm sure you know the drill ... may can have extra insulation, use of one core, and even a tertiary winding with old primary ...most interesting ...John

[AB2EZ]

John

I made the measurements you requested on the toroidal transformer which has the following specifications

Input: 120 VAC (using both 120 VAC primaries in parallel)

Outputs:

1600 VAC (using both 800 VAC secondaries in series)
6.3 VAC 
6.3 VAC

I can see the 6.3 VAC windings... which are visible as having been wound on the toroid just prior to application of the transparent, outer plastic wrapper.

Each 6.3 VAC winding has 9 turns, is bifilar-wound with the other 6.3 volt winding, and is uniformly distributed around the toroid.

Just to be sure, I added a 2-turn winding of insulated hookup wire, and verified the the output measured across my extra winding was 2/9 times the output measured across one of the 6.3 volt windings. [For the test I was doing...  with a 400 Hz sine wave applied to the primary, and the "1600 volt" secondary having about 4000 ohms of load resistance across it... the rms voltage measured across the open "6.3 volt" winding was 1.344 volts, and the rms voltage measured across the open, added 2-turn winding was 0.294 volts.]

Therefore, the numbers of turns on this transformer are

6.3 volts windings: 9 turns (each)
120 volt windings: 171 turns (each)
800 volt windings: 1,143 turns (each)

Other notes:

As I understand it... Antek purchases various sized toroidal cores, with primaries already wound on them; and subsequently: Antek winds the appropriate secondaries on top of the existing primaries.

As noted above, the separate 6.3 windings were bifilar-wound.

Best regards
Stu