Modified Heising Modulation Circuit For A Valiant

[w0ng]

After spending many, many hours restoring a 1957 Valiant to operating condition, I've had a problem with the modulation xfmr (a new Peter Dahl unit) saturating at low audio frequencies (150cps and below). 

Scope patterns of the speech amp and audio driver give clean, undistorted single-tone audio sine wave waveforms from 100cps to 7k cps with plenty of voltage available.

After researching all the online info I could find regarding adding an outboard modified heising modulation circuit to my Valiant xmtr, I decided to use the circuit devised by Mike, N1IW, that uses two HV oil-filled caps and a reactor that plugs into the Valiant's rear panel accessory socket.

I ended up using two 3uf, 2kv oil-filled caps and two Hammond 193M, 10Hy 300ma ckokes (wired in series) to get the required 20Hy 300ma reactor.

I also built a modulated audio sampling circuit so I could see what the trapezoid waveforms looked like.

The reactor/capacitor circuit seems to help with the low frequency audio distortion I had before. The very low freq trap patterns look better to me.

My question is how much improvement can I realistically expect from this outboard heising modulation setup? Are my capacitor and reactor "values" correct for the maximum intended benefit?

I'd really appreciate any tips, hints, comments, editorials, etc. from those of you who have either used this setup in a Valiant or have experience with these things.

Thanks in advance.............
Bill, w0ng,
Fountain, CO

[AB2EZ]

Bill

I've been using a modified Heising configuration with my EFJ Ranger for a couple of years now. I also use a pair of current-production Hammond 10 Henry 300ma chokes in series:

Some comments:

1. You said that the waveforms look good up to the driver

Question: Did you check the waveforms at the output of the driver (e.g., on the grids of the modulator tubes? In my EFJ Ranger, the driver transformer was definitely the biggest bottleneck in terms of low frequency distortion and roll off

2. I replaced the driver transformer with an electronic circuit:

http://mysite.verizon.net/sdp2/id13.html

I also use the 9-pin connector on the back of the Ranger to connect the Ranger to a separate small chassis, containing: the two Hammond chokes, the capacitor, and a switch to allow me to easily change the configuration from standard to modified Heising. [DANGER: As you know, there are voltages as high as (or possibly higher than) 3x the B+ present on some of the components (e.g., across the capacitor) in a modified Heising configuration... due to the combination of: the B+, the modulation, and series resonance effects.]

Note: The capacitor that you are using (a pair of 3uF capacitors in series) sounds about right for this purpose. You might want to include some large balancing resistors (e.g., 100k ohms at a suitable wattage to handle the B+) if you haven't already done so... to ensure that the DC part of the voltage across each of the capacitors is roughly half of the total B+.

What I find with my Ranger is that the output rf envelope waveforms look better on my scope (as detected by my off air monitor), below 150Hz, with the modified Heising configuration (as expected)... but the waveforms begin to look distorted, even in the modified Heising configuration, at 50 Hz and below. I.e., even without the DC in the secondary, the modulation transformer still saturates (when you run the Ranger at full power with 100% modulation) at frequencies below 50 Hz.

Of note: In on-air tests, most people notice very little difference in the subjective quality of my audio when I switch between the modified Heising and the conventional configuration. Some people like the modified Heising configuration a little better, and some people actually like the sound of the conventional configuration a little better. [I believe there is a psycho-acoustic effect in which harmonics of low frequencies are perceived as the presence of (missing) low frequencies... even when the receiving system doesn't pass those low frequencies]

Also of note: I often use my Ranger with an external modulator (completely bypassing the Ranger's internal audio chain, including the modulation transformer) consisting of: a modern 60 watt solid state audio amplifier driving a reverse-connected Hammond single-ended output transformer (which serves as the modulation transformer). This produces very clean modulation down to 30 Hz (the lowest I usually measure) with either the plate current flowing through the Hammond transformer (with is rated for 100 ma of unbalanced DC) or in the modified Heising configuration.

Best regards
Stu

[The Slab Bacon]

Just FWIW there is an easy way around the mod iron saturation problem. No one has ever complained about the audio from my Valiant.

The simplest way around the saturation problem is to load the final lighter. Instead of loading it to the 360 mA that the manual says to do, load it for 275 - 300mills of plate current. I load mine to somewhere between 275-280 mils, this gives me right around 100-105w carrier output.

No one on the other end is gonna see that much of a difference in their S meter, but they will definately hear the difference in the audio. I have been running mine that way for years. It is also a lot easier on the mod tranny as well. Keeping in mind that the Valiant mod tranny is a bit undersized to begin with.

                                                             The Slab Bacon

[WA1GFZ]

I do the same thing with the V2

[Steve - WB3HUZ]

Hookup and component info on Heising for the Valiant here.

http://www.amwindow.org/tech/htm/valiant/heising.htm


Complete article at

http://www.amwindow.org/tech/htm/valiant/valiantmods.htm

[WBear2GCR]

Fwiw, the V2 I have here was modified by KE1AV... I got it third party in a swap/sale.

I load it to 225ma.
It has the 3 diode limiter circuit.
A pair of strapping sweep tubes for the modulators.
A very strapping driver transformer.
Modulates the needle right against the pin when looking at the mod current.

Big positive peaks.

I think it is similar to the circuit in the amwindow, but is using stock voltage on the plates of the mod tubes...

Never tested the freq response so far.

I'd be suspicious of the little interstage.
My preference is for either going with a better xfmr there, going modified Heising for the driver iron, going push pull there, or changing it to a cathode follower driver (requires possibly an extra dual triode tube in there...)

              _-_-bear

[stevef]

Don't take your panel meter reading for granted.  Verify with a known accurate milliameter.   If it's off, you can fiddle with the meter shunt length - )while it's off though )

[w0ng]

WOW... Really appreciate all the replies to my post.

I do have clean audio waveforms at the grids of the 6146 modulator tubes. Have been loading the finals to 250ma and consistently get 100-110 watts out.

Looked at the audio waveforms with a 2000 ohm, 100 watt resistive load connected to the modulation xfmr secondary (secondary leads disconnected from the circuit) to eliminate any chance of rf pickup by the scope probes and the audio oscillator sine waveforms were clean from 100cps to 7000cps.

With the reactor/capacitor circuit plugged in, I get considerably less distortion at 50-100cps.

BTW... I'm using the circuit that has two "separate" 3uf 2kv oil-filled caps and a pair of 10Hy 300ma reactors (series connected to get 20Hy) as shown in the drawing posted by Steve, WB3HUZ.

John, WA5BXO, has been helping me understand the theory and practice behind all this. He has had tremendous patience with my endless questions.

Bottom line seems to be that my Valiant is going to have some low frequency distortion and some high freq phase shift no matter what I do. Time to get it on-the-air to see how it sounds.

Bill, w0ng

[AB2EZ]

Bill

Yes... sounds like on-the-air subjective testing is the "next order of business".

Of note:

Putting one capacitor on the top of the mod transformer secondary, and one capacitor on the bottom of the mod transformer secondary is a fine compromise with respect to how the B+ is distributed among the components. This arrangement will, if working properly, put 1/2 of the B+ on the secondary of the mod transformer.

Some of us like to put the capacitor(s) entirely on the bottom of the secondary... i.e., between the secondary and ground... so that the secondary is at the full B+ potential. That approach keeps the primary of the mod transformer (which is at the full B+ potential, in the case of the Valiant) and the secondary of the mod transformer at the same DC potential.

However, the need for balancing resistors is still there.

Each of the capacitors has a certain amount of leakage current. The leakage current of each capacitor corresponds to a resistor in parallel with each capacitor. Since the capacitors are unlikely to be identical (either when first manufactured, or over time) these "built-in" resistors form a voltage divider for the DC (i.e., the B+) that splits the DC unevenly between the two capacitors. [The secondary of the mod transformer, which sits between the two capacitors, is essentially invisible with respect to the issue of how the B+ splits between the two capacitors.] For example, if the parallel leakage resistance of one of the capacitors is 5 megohms, and the parallel leakage resistance of the other capacitor is 15 megohms... then (after the B+ has been on for a few seconds) the B+ will split as follows: 1/4 of the B+ across the capacitor with the 5 megohm parallel leakage resistance, and 3/4 of the B+ across the capacitor with the 15 megohm leakage resistance.

The purpose of the balancing resistors is to "swamp out" the differences between the leakage resistances by placing a physical resistor in parallel with each capacitor... whose value is much less than the leakage resistance of any (normal) capacitor.

I suggest 100k ohms as the value for each balancing resistor... but you can determine the proper value by measuring the leakage resistance of each capacitor. One way to do this would be to (very carefully) see how long each capacitor takes to discharge (its RC time constant) after the HV is removed.

If the B+ value is (for example) 600 volts... then, with roughly half of the B+ across each capacitor, the power dissipated (just from the B+, not including effects of modulation) by each resistor will be (300 x 300) / (100,000) watts = 0.9 watts. Thus, I would suggest using, as a minimum, 2 watt resistors for this purpose. [Wirewound is fine]

A separate benefit of doing this is that these resistors will also serve as bleeder resistors for each of the capacitors. With 100,000 ohms across each 3uF capacitor, the discharge time constant will be 0.3 seconds.

Best regards
Stu

[w0ng]

Stu,

Thanks for the advice on adding the 100k resistor in parallel with each oil-filled capacitor. I'll definitely do that. If memory serves me right, I've got a couple of 5 watt, 100k metal oxide jobbies in my parts bin that I'll use. I definitely like the added benefit of having a bleeder across each cap.

73
Bill

[W2XR]

Hi Stu,

Not really germane to this thread;

I currently use a 30 hy choke in series with the screen supply for my 2x 4-400A plate-modulated final. The choke, of course, is to provide self-modulation of the screens.

My modulation xfmr (from a Gates BC-1H 1KW BC rig) includes a tertiary winding for plate & screen  modulation of the 2x 807 drivers in this rig. I have been thinking about connecting this tertiary winding in series with the screens of the 4-400As to directly modulate them, instead of using the screen reactor. This mod xfmr is used in conjunction with a 40 hy modulation reactor and a 2 uf DC blocking cap.

The tertiary winding provides about 155 VAC across a 3000 ohm load. The screen Z presented by the 2x 4-400As (based upon the  way I load the final) is about 5600 ohms.

Do you see any advantages or issues with this topology?

Thanks & 73,

Bruce

[AB2EZ]

Bruce

Hi!

I did some experiments a few weeks ago (see my post on separate plate and screen modulation of a DX-20 a few pages back... last reply on January 23) to demonstrate the effect of modulating the screen more pro-actively.

I was able to get improved linearity... of the modulated rf output envelope vs. the audio input signal... by driving the screen of the r.f. tube with a voltage waveform that was (for that particular transmitter) about 20% of the plate modulating voltage waveform... and with a source impedance that was significantly lower than the impedance looking into the screen (i.e., screen voltage / screen current) at carrier.

I believe that my experiments showed (probably confirming what many others had determined, long ago) that forcing the screen voltage to accurately follow the modulating waveform is the critical factor for obtaining good modulation linearity... and that the accuracy with which the plate voltage follows the modulating waveform is much less important (actually unimportant, unless you run out of headroom).

Now, having demonstrated that on my DX-20 (having a resale value of less than $100.00, and operating a a B+ value of around 550 volts)... the question that comes to mind is whether it is worth the trouble and the risk to "mess around" with your 2 x 4-400A rig.

My KW-1 also uses a reactor to achieve self-modulation of the screen voltage of a pair of 4-400A finals... and people who hear it on the air think it sounds pretty good (including me) ... with my Class E rig as a baseline for comparison.

When I modulate the KW-1 with a 400 Hz sine wave, at 90% modulation (a few minutes ago... just to check)... the second and third harmonics (800 Hz and 1200 Hz) of the modulated rf envelope, as observed on my off-air monitor, are more than 30 dB down from the fundamental... and the rest of the harmonics are much more than 30 dB down. The waveform itself looks very nice (i.e., it looks like a sine wave).

So, while my experiment with the DX-20 showed some objective advantages to pro-actively modulating the screen voltage with the modulating waveform... it appears that the self-modulation approach also works very well... at least in my KW-1.

Note: the KW-1's modulation transformer, and the associated 2 x 810 push-pull modulator are doing a good job of providing linearity... at least above around 50 Hz.

If the modulation transformer and/or the modulator were (hypothetically) not providing good linearity... then pro actively modulating the screen voltage with a separate modulation signal (not derived from the output of the plate modulation transformer) would (in my opinion) significantly improve the linearity of the modulation, as well as the low frequency performance of the modulated transmitter.

I would (of course) never experiment with the screen modulation architecture of my KW-1.

However, if I were building a plate modulated 2 x 4-400A transmitter from scratch... or any other tetrode-based plate modulated transmitter... I would pro-actively modulate the screen, using either a low voltage tap on the mod transformer (e.g., the "ultralinear" tap on a backward-connected output transformer) or a separate, screen modulation transformer whose primary is driven by a separate driver stage (so that non-linearities in the plate modulator do not disturb the linearity of the screen modulator).

Best regards
Stu

[WA1GFZ]

Bear,
Pegging your mod current meter doesn't prove anything it the mod transformer is saturated. I used to be able to peg the mod current running 807s but switched over to AB2 dropping the screens to 250 volts. I can't peg it anymore but it sure sounds better with a bigger driver transformer. The monkey swings the 807 grids to almost plus 20 volts and it is limited by drive to the new 6AQ5 driver.

[k4kyv]

I modified the Eico 730 modulator that I use with my 10m rig.  After changing some coupling caps in the low level audio stages and removing the clipper circuit, I shunt fed the mod xfmr using a 40Hy 120 MA modulation reactor (actually manufactured for the purpose) that I had in my junk collection.  The mod reactor fit nicely into the space formerly occupied by the clipper stage.  However, I can't put the top cover back on the modulator because the mod reactor is too tall.

Those mods improved the audio quality tremendously, but I still wasn't satisfied.  So I disassembled the mod xfmr and eliminated the core gap by re-stacking the laminations so that the each adjacent set of E and I laminations is oriented in the opposite direction, just like a power transformer.  This is sometimes called "cross laminating" a transformer.  The result was that the modulator is now flat from about 20~ to well beyond 10k.  I used to get frequent reports of "excellent audio" with it, and am looking forward to when 10m will open again on a regular basis.

Instead of the normal low-level microphone input to the amplifier, I use the "phone patch" input, to feed audio from the same 500-ohm balanced line output from the audio rack that supplies audio to all my other transmitters.

Of course some purists would say that I "ruined" the 730 because it is no longer stock and I destroyed its "collector's value".  But this wasn't exactly a "Hammy Hambone" modification, and the modulator is worth much more to me now than it ever would have been stock.

[WA1QHQ]

Bill,

I believe your question was how good will the modulated waveform be if the mod transformer does not have to pass unbalanced DC current through its secondary. To get the answer just disconnect the secondary windings from the Valiant circuitry, terminate the secondary in a resistance equal to your loaded final plate impedance and then put a scope across the secondary and look at the waveform while cranking up the modulation to a modulator current equivalent to 100 percent modulation. This gives you the best case condition for modulator quality but removes any question of contribution from core saturation of the modulation transformer.

Mark  WA1QHQ

[W2XR]

Bruce


However, if I were building a plate modulated 2 x 4-400A transmitter from scratch... or any other tetrode-based plate modulated transmitter... I would pro-actively modulate the screen, using either a low voltage tap on the mod transformer (e.g., the "ultralinear" tap on a backward-connected output transformer) or a separate, screen modulation transformer whose primary is driven by a separate driver stage (so that non-linearities in the plate modulator do not disturb the linearity of the screen modulator).

Best regards
Stu

Hi Stu,

Thank you very much for the usual thoughtful and detailed reply!

If I read your reply correctly, assuming the modulator & mod xfmr do indeed provide good linearity, then you believe that connecting the tertiary winding of the mod xfmr to the 2x 4-400A screens could provide an improvement in the modulation linearity of the transmitter, vs. using the existing topology of self-modulating the screens with a screen reactor.

Am I correct in interpreting your opinion?

Note that I homebrewed this rig myself, and I've never been fearful of going in there and experimenting, so long as the potential risk of damaging the rig is small!

Thanks & 73,

Bruce

[AB2EZ]

Bruce

If you have measured the modulation linearity (input audio -to -envelope of modulated rf output), and you find that it is not satisfactory (e.g., modulation distortion products less than 30 dB down)... then you might want to consider each of the following.

1. First, check the unmodulated screen voltage to see that it is reasonably close to 500 volts (i.e., somewhere between 500-600 volts). I assume that you have already checked this... but I just wanted to be sure.

In your initial post, you said that the screen Z is around 5600 ohms... which I assume is calculated by dividing the screen voltage by the screen current, in the absence of modulation. For two tubes in parallel, and assuming 500 volts on the screens, this implies 45 mA of screen current for each tube (22.5 watts of screen dissipation for each tube). While still within the 35 watt maximum screen dissipation rating... the screen current seems a little high. Eimac suggests 26-30 mA of screen current for ICAS operation... depending upon the plate voltage.

1A. Check to be sure that the r.f. bypass capacitors from the screens to ground are less than .001 uF. If they are too large, you will roll off the high frequencies in the screen modulating waveform (even when using self-modulation of the screen).


2. If the unmodulated screen voltage is okay... then I agree that it would be worth a try to change the screen modulation methodology from self-modulated to pro-actively modulated.

Things to keep in mind / watch out for:

A.  The tertiary winding of the modulation transformer may have B+ on it... depending upon whether it is a totally separate winding or a tap on the secondary winding. It sounds like, in your case, this is a totally separate winding.

B.  To modulate the screens of the 4-400's you will need a peak screen modulating voltage of roughly 350 volts (i.e., 70% of the screen bias voltage). It doesn't sound to me like the tertiary winding you have available is producing a sufficient voltage. If not, you may have to either use a supplementary 1:3 step up transformer in conjunction with this tertiary winding, or a separate screen modulation transformer in parallel with the plate modulation transformer. This is not quite as bad as it sounds... because the audio power required to modulate the screens is much lower than the audio power required to modulate the plates. Since the differential (small signal AC) impedance looking into the screens (dv/di)  goes down when the screen voltage goes up, and vice-versa, the source impedance of the output of the screen modulator has to be around 3000 ohms or less (preferably less than 2500 ohms), in order to swamp out the effect of this non-linear load... which means that the separate screen modulation transformer has to be about the same size as the plate modulation transformer that you would use with a Johnson Valiant.

C. You need some way to adjust the level of the screen modulating signal... relative to the level of the plate modulating signal. One way to do this would be to use a multi-tap screen modulation transformer. Another way would be to use a separate audio driver for the screen modulator... with an adjustable gain control. If the screen modulating signal is too small... then you won't achieve 100% modulation. If it is too large, then you will experience distortion. [Remember, it is really the screen voltage that is modulating the plate current... and it is the plate current that needs to faithfully follow the modulation].

D. The phase difference between the screen modulating voltage and the plate modulating voltage is not very critical. Obviously you don't want to accidentally put the screen modulation 180 degrees out of phase with the plate modulation. At frequencies below a few kHz, phase differences between the plate modulating signal and the screen modulating signal will be small. At higher frequencies, the phase differences may be larger... but that is true even for traditional plate modulated rigs, which derive their screen modulation with a dropping resistor from the plate or via self-modulation of the screen.

Best regards
Stu

[AB2EZ]

Bruce

One other thought:

If you put the tertiary winding (assuming it is a totally separate winding) from screen to ground, via a carefully selected resistor (my initial guess would be 5k ohms), and you keep the existing modulation choke in place... it might actually help. Then again, it might not. It might be worth trying.

The audio voltage across the tertiary winding will "add" to the self modulation effect, in the following sense.

The self modulation occurs because... when you reduce the plate voltage (for example), electrons flowing from the cathode to the plate are more likely to stick to the screen (instead of passing through). This causes negative charge to accumulate on the screen... and, if the AC impedance from screen to ground is high enough, they tend to pile up on the screen... which reduces the screen voltage.

If you put the tertiary winding (in the correct phase, of course) between the screens and ground via a 5k ohm resistor (one side of the tertiary winding connects to the screens, and the other side connects to one side of the 5k ohm resistor; the other side of the 5k ohm resistor connects to ground)... then the audio frequency voltage on the screen will be the sum of the voltage across the tertiary winding plus some self-modulation voltage. This self modulation voltage is, again, caused by electrons sticking to the screen, and piling up on the screen because they can't flow to ground quickly enough through the 5k ohm resistor.

It might work!

Note that the DC screen voltage supply will have to provide additional average current through the modulation choke because of the presence of the extra 5k ohm path from screen to ground. You could eliminate this DC path by including a 1uF capacitor (or larger) between the 5k ohm resistor and ground.

Stu

[Steve - WB3HUZ]

With the reactor/capacitor circuit plugged in, I get considerably less distortion at 50-100cps.

This is to be expected. Most amateur grade iron begins to fall apart below 100 Hz.

Bottom line seems to be that my Valiant is going to have some low frequency distortion and some high freq phase shift no matter what I do. Time to get it on-the-air to see how it sounds.

You got it. There's this old saying about polishing things.    Anyway, if the thing is pretty clean from 100 Hz to 7 kHz, you will sound good on the air, given your audio input the the transmitter is appropriate (mic, proper eq, and such). Most people's voice don't have much energy below 100 Hz and most people's receiver aren't going to hear or pass audio above 4-5 kHz. Anything beyond this is either gravy or unnecessary, depending on your POV.

Let's hear that thing on the air!