I became infatuated with Valiant when I first heard W8YAT on 160 meters in 1962. Bob was using a Valiant with a D-104. The big thump when that carrier came on and such audio!

I bought my first Valiant in 1970.

Chassis View

There are a lot of Internet modifications to the Valiant. My take on the mods is people are getting a little too carried away. The Valiant isn't all that bad. As time permits I'll add my mods here. All of my mods are verified with test gear, including spectrum and distortion analyzers.

HV Power Supply

Some people claim the power supply is weak. Factually, the supply is pretty well engineered unless you plan on tearing the Valiant all up and making it into a totally different transmitter.

I tested the supply on CW and AM. CW is easy. On CW, strings of dots at various speeds would show power supply defects. CW is also the worse case loading for the supply, since the dynamic load varies abruptly and to extremes. The unkeyed current is a few dozen mA, and the full load current is around 500mA.

AM is a little tricky to test, since there are both syllabic and tone variations. The supply has to be reasonably stable for slow dynamic variations and not have excessive ripple induced at audio frequencies by the peak load caused by peak modulator current. Let's look at just how bad a Valiant is.

CW Test

Running 475mA (ouch) and adjusting dot duty cycle and speed for maximum ripple, here is what my stock Valiant does:

The scope is scaled for 100v/div, or 20x scale reading. 800 volts is the top line. Multiply delta V1 by 20, and we have 184 volts p-p dc fluctuation with load. This fluctuation occurs because the power supply choke cannot recover fast enough to impede current when load suddenly stops, and likewise impedance cannot fall fast enough to hold voltage when load starts. The mean supply voltage in my Valiant is 760 volts with no load. It is 620 volts under normal CW load. Most of that sag is because the filter choke is marginal on inductance for the bleeder current, and allows the supply to move higher than the .9*RMS of a typical choke supply that does meet critical bleeder current/choke impedance values.

By no means should a supply voltage no-load soar from having less than critical inductance (or bleeder current) be interpreted to mean a supply is "weak". Extremely well designed unregulated single phase supplies sag 10-15% from no load to full load. The Valiant sag is less than 20%, and that could easily be improved to 12% or better with an increase in bleeder current or effective choke reactance.

As for the bounce, any choke input supply behaves this way. Even a commercial BC transmitter will do this when the carrier is keyed at the right rate. This is NOT a sign of a weak supply, it more an indication of how choke input supplies behave under CW keying loads. Try CW keying a 900 pound kilowatt AM BC transmitter, and see how the supply behaves!

How would I fix that? The first thing I'd try would be to resonate the choke with a non-polarized capacitor. You'd have to be sure the capacitor-choke combination was resonant slightly above 120 Hz. It is often possible to find a "sweet value" of capacitor that takes much of the no-load voltage increase out. A second option would be to load the supply heavier with a lower value bleeder resistor, but that would waste power and make undesirable heat. There should be about 400 volts or so peak AC across the filter choke. From appearances and my experiences with other choke input supplies, I would say the addition of a single component (a capacitor across the choke) would pull no-load voltage down to 700 volts or less without changing full-load voltage. The no-load to full-load bounce would be greatly reduced.

Maybe someday I'll try tuning the Valiant supply choke to reduce the no-load voltage peak. For now, it does not appear to be that big of a deal. Remember the scope capture above shows the WORSE CASE dynamic load variation.

We can see the effect of the sag and overshoot on the CW envelope. Look at the hump in the leading edge of the CW envelope in the following shot:

Notice the leading edge overshoot. The envelope overshoot at the leading edge is created by HV supply dynamics. Since the envelope is rounded and occupies 7.5 ms, it does not affect bandwidth. I opted to not do anything about it. (By the way, I reshaped the CW waveform from stock, more on that later.)

AM Dynamics

To test AM, I used a pulsed tone. I set the tone for maximum ripple and maximum dynamic variation. This wound up being about 300 Hz AF, and a pulse rate of 5 Hz. This test was running 350 mA and slamming the modulator current against the pin. you will never have it this bad in real life. Here is the worse case power supply error on AM:

The 300 Hz audio ripple is down in the noise of power supply hum, and the total pulse dynamic variation is only 64 volts p-p, or 10%. This variation is at a syllabic rate, and is only this bad under worse-case pulse audio conditions. It is not measurable as audio distortion on the output. The supply in the Valiant is strong enough for AM. Without any question, the Valiant does not need power supply work for AM operation. (By the way, this is with old stock electrolytics.)

Modifications

Audio

Here is the 3kHz envelope of my Valiant (after minor) modification at 100% modulation:

The shot below shows modulation at 350Hz. Notice the rolling off after the peak. The Valiant is running out of iron, the modulation transformer (the grids of the 6146's are fine) at low frequencies.

If you compare the two envelopes, you'll see very little amplitude change. For a fixed generator input, my Valiant is almost perfectly flat from about 300 Hz to 3500 Hz. The only thing required to flatten response were the addition of a few capacitors. I never removed or changed the value of RF bypass capacitors, and I never removed the low-pass audio filter that follows the audio clipper.

As a matter of fact if I remove the audio low pass the response actually sags above 2500Hz! The audio filter, besides starting rolloff at 3500Hz or so, adds an upward slope as cutoff is approached. This upward slope is just right to compensate downward slope caused by bypass capacitors. Take that filter out, and you'll also have to remove bypass caps. The result of removing the bypass caps is distortion products will blow right through the system and modulate the PA for 10kHz and further away. No need to do that!