Valiant and other clipper circuits: The magic, the mystery, and the scorn

[N1BCG]

"The first thing you need to do it tear out that circuit!"

I've heard this stated over and over to those looking to improve the audio of their Johnson Valiants, Eico 730 Modulators and others that employ such a circuit. Dutifully, these ops drill and blast the circuit out without an understanding of why they were included in the design to begin with.

As is well known, an increase in average modulation will result in an increase in loudness which is vital for making the most of a transmitted signal. This can be done to an extent with audio compression, but it involves a bit of circuitry and audio peaks still restrict the benefit of compression. The problem is that peak energy in the waveform is sampled to determine attenuation even though it doesn't contribute to loudness. Think of audio peaks like static discharge in a dry room when you touch a grounded piece of metal. That spark can be in the tens of thousands of volts but it's of such short duration that you remain alive due to the lack of energy.

In a transmitted waveform, the amplitude of peaks can extend far beyond the average waveform energy, but the modulation has to be restricted to prevent them from cutting off the carrier or damaging an amplifier. Average modulation would rise significantly without these peaks, so they need to go. The engineers at E.F. Johnson and EICO knew this and included a clipper circuit that achieves loudness by clipping off the undesirable peaks thus raising the average energy of the waveform.

So why the universal hatred?

The downside of clipper circuits is that they generate harmonic energy proportional to the amount of waveform clipping, so a filter must follow to pass desired audio but eliminate everything above in frequency. Yep, there's the issue, and "desired audio" is a subjective term.

Hamateur transmitter manufacturers through the years focused on tightly restricted audio for two reasons. First, to do it right, broadband audio requires more expensive components, such as with the modulation transformer. Second, signal to noise ratios are higher if receiver filters are narrow, and there's no sense in wasting modulation on frequencies that won't be received. While they were at it, they rolled off the lows as well since receivers tend to do that also. Yeah! Smaller mod iron! This is how the campaign for "hyellowy audio" became so popular with designers.

Interestingly, clipper circuits have been universally used in AM and FM broadcast processors for decades and their ability to achieve competitive loudness is crucial, so how are they great for Part 73 stations and hated for Part 97 stations? The love/hate relationship with clipping is in the choice of filtering component values.

Most clipper circuits are comprised of two sections: a dual diode that clips both polarities of audio peaks and a low pass filter that is needed to remove harmonic energy generated by the clipped waveform. It's the aggressiveness of the low pass filter that needs attention.

The L/C clipper circuits in the Valiant (L45, C94) and EICO (L1, C6) are nearly identical, and while I could not determine the value of the Valiant's L45, EICO used a 5 Henry reactor for L1.

Valiant clipping circuit:



EICO 730 clipping circuit:



Both employ a 6AL5 dual diode where each section clips a polarity of the waveform. Some have recommended that one diode be bypassed so that only negative peaks are clipped. While this results in unleashed positive peaks, I doubt the consequences to the mod iron and RF components justify the effort, and, you're only letting peaks with no audio energy through. Leave the peak power meter for SSB use.

By the way, a quick tip for Valiant jockeys: If you find that your negative peaks exceed your positive peaks, reversing the plate caps on the 6146 modulator tubes will reverse the modulation polarity. No fuss, no muss, and no soldering!

Using the online calculator for passive low pass 2nd order filters, an L value of 5H paired with a C value of 0.001 uF gives a cornering frequency (Fc) of 2.25 kHz. Y-I-K-E-S! At least EICO gave low frequencies a better deal by using 10 times the capacitance (C4) for coupling from the clipper compared to Johnson's 0.01uf (C84).

So I'll bet you're ready to grab those cutters and dispense with these circuits, but before you do, remember that very similar circuits are used in FM broadcasting. The difference is a significantly higher Fc and very tight control over how much clipping is used. As mentioned, the amount of filtering needed is proportional to the amount of clipping. Broadcast processors use compression *and* limiting prior to the clipping stage, so the amount of clipping and resulting harmonic energy generated is significantly reduced.

I was pretty astonished (huge understatement) to read about setting the clipping control in the Valiant manual where it is recommended that 12dB of clipping "is not at all objectionable, on the contrary, speech sounds as though the speaker is enunciating with special care".

Okay, my fellow broadcast engineers and audiophiles, you're probably making faces right now, I was too when I first read that, but we think of this as clipping a processed waveform, not a raw waveform as would be the case in these devices.

That said, I feel that the real focus should be on the low pass filter and responsible clipping levels.

The value of L is WAY too high, I believe, but audio reactors are hard to come by. However, simply changing the C in the filter to 250pF would increase Fc to 4.5 kHz. Adding a compressor such as the popular Symetrix 528 would allow much lower levels of clipping to be used, and together, would add significant punch to the Valiant or EICO.

None of this addresses the many other components that deserve scrutiny when improving audio, but I felt that someone should stick up for a circuit and function that has drawn much criticism, and mostly undeserved as I hope you will agree.

[n1ps]

Well...interesting Clark.  Didn't Johnson implement this circuit for SSB?  With perhaps the thought that as the TX is communications grade audio, no one will care as far as AM (in 1955).  So what you are saying is the 'Clipper" is really a LP filter and a peak limiter.  From your math....there seems to be no other choice but to drill and blast to make the filter usable to those of us who want something a little better than...er a lot better than, communication grade.  So we cut it out anyway.

Peter

[KD1SH]

Clegg had another approach in their Zeus 6 meter transmitter: sensing the negative peaks at high level, and using the result to apply clipping at low level. Clipping is still clipping, though, and they still had a low-pass at both the low level and high level end. The signal coming off the 6C4 cathode follower in the schematic gets fed back to the twin solid-state diode clipper just upstream of the 12AT7 phase inverter.

[WQ9E]

The biggest problem with the Valiant clipper (also used in the Viking 500) is owners who never read the manual  The clipper control operates opposite to what is expected and minimum clipping occurs with the control set fully clockwise which introduces no clipping.  But the average ham knows too much to need to read so the clipper is quickly set fully counterclockwise which does sound truly terrible with that level of clipping.  My Valiant, Valiant II, and Viking 500 all still have their clipper circuitry; I never use it on the 500 but I have used a little clipping at times on the Valiant back in the mid 1990s when it was the only AM rig I had and a little clipping with its 200 watts input sometimes made the difference.

The Heathkit TX-1 has a similar operator issue because it has two controls like the Valiant with one only accessible via the CW key jack using a long straight blade screwdriver.  If this hidden control is set incorrectly, then you will have the classic scratchy Apache sound from far too much clipping.

Rodger WQ9E

[Steve - K4HX]

The other problem with these lowpass filter is the horrible impulse response. It reduces the effectiveness of the clipping. I never checked the phase response on these filters but I'm guessing it not the best for audio quality.

No need to drill and blast to remove these. They Eico required the change of two leads and can be performed in about a minute.

[W3SLK]

Steve said:

No need to drill and blast to remove these. They Eico required the change of two leads and can be performed in about a minute.

Same with the Apache! I remember how easy that was and left everything intact in case somebody wanted it original.

[N1BCG]

Well...interesting Clark.  Didn't Johnson implement this circuit for SSB?  With perhaps the thought that as the TX is communications grade audio, no one will care as far as AM (in 1955).  So what you are saying is the 'Clipper" is really a LP filter and a peak limiter.  From your math....there seems to be no other choice but to drill and blast to make the filter usable to those of us who want something a little better than...er a lot better than, communication grade.  So we cut it out anyway.

I've heard some stock Valiants sound remarkably good when paired with D-104 mics. The key was that the D-104 elements were in great shape, ie, the crystals were producing the designed response so it effectively countered the Valiant's high end roll-off.

Interestingly, when frequency response is limited resulting in a muffled sound, rolling off the lows will greatly improve clarity and the Valiant certainly rolls off lows.

This is the exception, not the rule, so for using a Valiant with "normal sources" I certainly can understand the desire to beef up the coupling caps to improve lows and reduce the value of audio shunting caps to improve highs, but the clipper is still a useful tool, so changing the value of the inductor or cap would go a long way to improving frequency response without losing the benefit of the clipper.

[WBear2GCR]

The filter can in fact be made to work at a higher frequency... I dorked with it in LTSPICE some
years back. Trying to recall my results. It was tricky. The Q with the large inductor and small
caps is difficult to work with, iirc. The value of the caps need to chosen carefully, and the
shape of the resulting filter considered as well.

I seem to recall attempting to change the filter to an elliptical type, in essence to brickwall
above something like 5.5kHz. That would be putting the optimum value of cap in parallel
with the inductor.

My interest was not to filter the clipper, but to do a slick job of bandwidth limiting for the rig.
I would not run the stock circuit, and would not run the clipper. They were put in there to
compete with the oncoming SSB popularity, and because at that time the bands were
hot. One could not get another QSO in edgewise, if you wanted
to! The entire band was a mass of heterodynes from one end of the phone band to the
other, on ALL bands pretty much. ESPECIALLY on 75m and 40m... So, being "narrow" was a
thing at that time.

The busiest contest day today is nothing close to an average day back in the 50s and 60s
when these rigs were "modern".

                        _-_-

[WBear2GCR]

Found what I had worked out... Oh yes, this was for a Globe Champion 350 circuit - quite similar.

The stock filter and the elliptical filter. FYI

[w3jn]

The distortion artifacts of lower-frequency clipped audio can still fall within the lowpass filter bandwidth and still make the audio sound distorted.  A lowpass filter, or modification of one, isn't a panacea here.

[WBear2GCR]

The distortion artifacts of lower-frequency clipped audio can still fall within the lowpass filter bandwidth and still make the audio sound distorted.  A lowpass filter, or modification of one, isn't a panacea here.

JN, The modification I showed above was NOT for use with a "clipper".
It was to limit the high frequency response of a normally operating transmitter to
~5.5kHz. Just to be clear. I agree with the earlier comment that a simple removal of the
clipper tube and bypass of the circuit with a capacitor usually will suffice...

[WQ9E]

I think an important thing to consider is don't buy one of these clipper enabled transmitters planning to run the clipper all of the time.  Even a moderate amount of clipping gets old in a hurry for the listeners but with lower to medium power transmitters, there are times that dealing with a little clipping beats trying to piece together a conversation based upon a few random syllables making it through the noise.

It is the same basic answer I give when photographers ask me about using tele extenders with a lens for more reach.  With a really good lens, there are times when they are quite useful but never buy a lens planning to use it with an extender all of the time because even with great lenses the extender always reduce the image quality. 

I spent last Wednesday morning freezing along the river in Iowa City to photograph eagles.  I was using a Canon 800mm f5.6 on one body and a 400 f2.8 on the other and for a few photos I did use the 1.4X on the 800mm for more reach and because these high end primes produce great image quality as a bare lens the results are acceptable with the 1.4X but I use it only when needed.  With AM, I would rather listen to a legal limit rig than a low power rig with clipping but at times that low power rig will only get through with clipping but it shouldn't be something others have to listen to all of the time.

Rodger

[Opcom]

The filter can in fact be made to work at a higher frequency... I dorked with it in LTSPICE some
years back. Trying to recall my results. It was tricky. The Q with the large inductor and small
caps is difficult to work with, iirc. The value of the caps need to chosen carefully, and the
shape of the resulting filter considered as well.

I seem to recall attempting to change the filter to an elliptical type, in essence to brickwall
above something like 5.5kHz. That would be putting the optimum value of cap in parallel
with the inductor.

My interest was not to filter the clipper, but to do a slick job of bandwidth limiting for the rig.
I would not run the stock circuit, and would not run the clipper. They were put in there to
compete with the oncoming SSB popularity, and because at that time the bands were
hot. One could not get another QSO in edgewise, if you wanted
to! The entire band was a mass of heterodynes from one end of the phone band to the
other, on ALL bands pretty much. ESPECIALLY on 75m and 40m... So, being "narrow" was a
thing at that time.

The busiest contest day today is nothing close to an average day back in the 50s and 60s
when these rigs were "modern".

                        _-_-

15KHz filter, probably brick wall-ish. This kind can change frequency pretty easily by proportionally changing the cap values in an evenly proportional way, such as multiply all x1.3, or x2, or whatever gets you close. 15K is high starting point but there are those who like Fc higher than 5KC, soo.. it's only $6.50.

[WBear2GCR]

Sorry, What is only $6.50??

[Steve - K4HX]

If conditions require the use of a clipper on AM, it's probably time to sign off.

[Opcom]

Sorry, What is only $6.50??

this filter, sorry
https://fairradio.com/product/fo69215/

[w1jto]

Clark,

Thank you for the very interesting post.

I’m  new to boatanchor AM and have a refurbed Valiant 1 + unamplified D-104.  Recently I purchased one of Steve's REA mod monitors and noticed that my negative peaks were equal to or slightly higher than positive peaks.

WRT

By the way, a quick tip for Valiant jockeys: If you find that your negative peaks exceed your positive peaks, reversing the plate caps on the 6146 modulator tubes will reverse the modulation polarity. No fuss, no muss, and no soldering!

A few questions:

1.   Your comment about switching plate caps caught my attention.  Why does this work ?

2.   I understand that the fidelity of a D-104 depends on it’s condition.   Is there any way to test the frequency response of a D-104 either alone or thru the Valiant audio chain ?  for instance, could you record audio received remotely with a websdr and check the recording against what was transmitted ?.

3.   Any good resources on adjusting the audio/clipping settings for the valiant/D104 or is the Valiant manual sufficient ?

My previous experience was all modern stuff and SSB.  I do have access to scopes and function generators, if needed.

Thanks again,

73's  John W1JTO

[N1BCG]

A few questions:

1.   Your comment about switching plate caps caught my attention.  Why does this work ?

The modulator, like many, uses two tubes in a "push-pull" configuration where one tube amplifies the downward audio cycle while the other amplifies the upward. Reversing the caps changes their connection to the modulation transformer. It could also be done by changing the jumpers in the accessory jack, but this is MUCH easier!

2.   I understand that the fidelity of a D-104 depends on it’s condition.   Is there any way to test the frequency response of a D-104 either alone or thru the Valiant audio chain ?  for instance, could you record audio received remotely with a websdr and check the recording against what was transmitted ?.

Using a web SDR is my highly recommended means of determining on-air sound since you can make a recording and hear how you compare to others. Unfortunately, it's not easy to evaluate microphones without a standard to compare them to. I've heard a few D-104s that sound remarkable, which reminds me of how few there are.

If you're ambitious, take audio from various points in the circuit by pulling out tubes and sampling the audio at grid pin connections in the sockets. This should be done through a D.C. blocking cap and series resistor to play it safe. Run that audio into an amplifier and see how it sounds. Consult the manual for this! You need to be sure you're testing at the right pin and socket.

3.   Any good resources on adjusting the audio/clipping settings for the valiant/D104 or is the Valiant manual sufficient ?

The Johnson Valiant manual gives a very basic outline for setting the clipping level. But, since you have a modulation monitor (which is indispensable for AM use) I would  increase your mic gain while speaking at a normal level and watching the modulation on the monitor. Set the gain to the point where you see a frequent, but not steady, negative 100% indication then dial in the clipper to reduce the peaks below 100%. After that, the clipper should pretty much prevent overmodulation. If not, use slightly more clipping.

My previous experience was all modern stuff and SSB.  I do have access to scopes and function generators, if needed.

You can certainly use a scope to monitor modulation as has been done for decades, but a decent modulation monitor will be a more precise method.

[KD6VXI]

You should also be able to change the wiring of the microphone.

Reverse the leads on the mic, in other words.


--Shane
KD6VXI

[w1jto]

Clark, 

Thanks for clarifying these points, follow up questions in blue

1.   Your comment about switching plate caps caught my attention.  Why does this work ?

The modulator, like many, uses two tubes in a "push-pull" configuration where one tube amplifies the downward audio cycle while the other amplifies the upward. Reversing the caps changes their connection to the modulation transformer. It could also be done by changing the jumpers in the accessory jack, but this is MUCH easier!


Does this also suggest that the tubes should (ideally) be matched ?


2.   I understand that the fidelity of a D-104 depends on it’s condition.   Is there any way to test the frequency response of a D-104 either alone or thru the Valiant audio chain ?  for instance, could you record audio received remotely with a websdr and check the recording against what was transmitted ?.

Using a web SDR is my highly recommended means of determining on-air sound since you can make a recording and hear how you compare to others. Unfortunately, it's not easy to evaluate microphones without a standard to compare them to. I've heard a few D-104s that sound remarkable, which reminds me of how few there are.


Is it possible/not crazy to use a tone generator input to the D-104 so as to create known freq. inputs to compare with measurable freq. outputs from the websdr recorded audio ?

or is this crazy-overboard... i.e.  just do it by ear ?


73 John O'Hara, W1JTO

[N1BCG]

Does this also suggest that the tubes should (ideally) be matched ?

Ideally, yes. If you find that they are mismatched, use the better tube for positive peaks. But, asymmetry will change even with different vocal sounds. Our voices, particularly male, are asymmetric but the polarity is usually opposite at lower vocal frequencies ("Aaaaaaahhh") than at higher frequencies ("Eeeeeeeee"). Some ops roll off frequencies below 120 Hz to allow the higher frequency polarity to dominate. Basically, microphones can only be connected in the proper polarity based on most speech.

Is it possible/not crazy to use a tone generator input to the D-104 so as to create known freq. inputs to compare with measurable freq. outputs from the websdr recorded audio ?

or is this crazy-overboard... i.e.  just do it by ear ?

A proper measurement of a microphone's performance would require a calibrated audio source and a specially designed, sound-dead room. Even with that information, your on-air sound will be a product of everything from the mic's windscreen through the antenna (yes, high-Q antennas and tuned circuits can attenuate higher audio frequencies). Of course if you're being heard on a receiver with a narrowed up filter, very little you do will improve the fidelity.

I think you should just get on the air and have fun with it!

[WBear2GCR]

Clark, 

Thanks for clarifying these points, follow up questions in blue

1.   Your comment about switching plate caps caught my attention.  Why does this work ?

The modulator, like many, uses two tubes in a "push-pull" configuration where one tube amplifies the downward audio cycle while the other amplifies the upward. Reversing the caps changes their connection to the modulation transformer. It could also be done by changing the jumpers in the accessory jack, but this is MUCH easier!


Does this also suggest that the tubes should (ideally) be matched ?

Imo, not in practice. Unless they are substantially different in terms of gain, or bias point
it makes no difference.

However, IF the modulator stock or modified uses negative feedback around the modulator
tubes or modulation transformer, then you can not swap the plate leads. That will cause positive
feedback, aka oscillation.

2.   I understand that the fidelity of a D-104 depends on it’s condition.   Is there any way to test the frequency response of a D-104 either alone or thru the Valiant audio chain ?  for instance, could you record audio received remotely with a websdr and check the recording against what was transmitted ?.

WebSDR, yes, just listen to the recording and decide if you like the way it sounds.

The D-104 requires a super high impedance as a load in order to have good low frequency
response. The stock rigs do NOT have the required load at the mic input (nor does outboard gear).

There are two solutions:
A) change the input resistor to ground to a 5 to 10megohm resistor.
B) place a JFet follower circuit (with a 9v battery - or get the power off the rig) in the base
of the D-104. This give the D-104 element a JFet's input Z to look at, and a low Z output.

Either one works very nicely. The (B) lets you use the same mic with other rigs that have not
been modified

Using a web SDR is my highly recommended means of determining on-air sound since you can make a recording and hear how you compare to others. Unfortunately, it's not easy to evaluate microphones without a standard to compare them to. I've heard a few D-104s that sound remarkable, which reminds me of how few there are.


Is it possible/not crazy to use a tone generator input to the D-104 so as to create known freq. inputs to compare with measurable freq. outputs from the websdr recorded audio ?

or is this crazy-overboard... i.e.  just do it by ear ?

Not sure what is being said here.
One can not input a signal generator into a D-104 and get anything out.
This can be done with a dynamic mic, but it's a good way to blow out mic elements.

If you are suggesting playing tones from a speaker and picking that up with the
mic? That is doable, but probably not the best thing to do on the air, and not tremendously
useful either.

Think of the mic like the tone of a guitar player's guitar + amplifier.
It's a question of the way it sounds not how it measures?

A mic that is just great for Clark, might be not so great for me.
Different voices.

Also, you can't get the rig to pass what it can't or won't pass anyhow.
That being lows past a certain point and high past a certain point.

73 John O'Hara, W1JTO

[WBear2GCR]

Note: the stock clipper circuits have a LP filter to prevent "splatter" on the clipped
audio products. The frequency of this filter, as previously discussed is somewhat
lower than we typically prefer to run AM rigs these days.

Aka: another reason to bypass the clipper circuit?
Or,
Modify the "splatter" filter. (worst case)

Imho, ymmv. As always.

                       _-_-bear

[Tim WA1HnyLR]

One thing that people seem to be unaware of is that the low frequency response of the modulator has to be much better that what the Valiant offers. Rip the clipper out. For some unknown reason in the plate circuit of the 6C4 the two resistors are bass ackwards. The 47 k from the B+ should be the 4.7 K rather than the way it is. Being that it is the last resistance coupled coupled stage there is no need for the .1mfd cap. Simply break out the dykes and surgically remove it. This is one example of some of the engineering defects in the Valiant. If one wishes to use clipping get rid of the driver transformer. Go with resistance coupling .The 6C4 becomes a phase inverter and the 12AU7 becomes a push pull audio voltage amplifier. After solid stating the HV supply and relocating the screen dropping resistor there is plenty of real estate on the chassis for a substantial piece of iron with good low end performance. The 6146 are not great modulator tubes but they will work OK Proper bypass capacity is a must for the screen voltage feed. A 10-20 Mfd electrolytic from the screen voltage feed to ground. The VR tubes present a non linear source impedance of screen voltage. Get another low pass filter reactor to build a low pass cut off at 5Khz rather than 3Khz. This will give a buzz cut to the speech audio before the clipper. Basically a preparation filter . Follow up the clipper with the usual clipper filter but pare down on the bypass caps to roll off @ 5Khz.
I have never tried any of this but it is my thought. Myself I don't use clippers in these transmitters. I will say tht the clipper in the Eico 730 works more effectively do to the potential low frequency response of the back end of the modulator. If one is to employ clipping to be heard on AM it is time to take up stamp collecting.
Tim WA1HnyLR

[KD6VXI]

If one is to employ clipping to be heard on AM it is time to take up stamp collecting.
Tim WA1HnyLR

If someone is using clipping to be heard on AM, it's time to build a triode with handles for an amplifier.


--Shane
KD6VXI