Viking II Mod Supply

[ad5st]

I picked up a Vik 2 recently, and the bleeder/mod-screen-divider resistor was open as I expected it would.

I'm looking at a couple of possible solutions:
- regulated screen voltage from a dropping resistor from the B+ and a pair of OA2s
- like the Apache (and DX-100?) put low voltage supply directly to mod screens

I like the simplicity of simply putting the low voltage supply on the modulator screens: that means I simply need to locate a suitable bleeder resistor, which can be of much higher value and significantly lower power rating. That of course presumes that the low voltage supply is suitable, with these higher primary voltages these days, for the 807 screens, and that the supply's clean enough.

What do you guys think?

--Wayne AD5ST

[WA1GFZ]

I just did this modification on my V2 CDC. I removed the old resistor and mounting brackets. I took a hunk of 1 1/2 inch aluminum angle and drilled holes in one side to take a 5 lug terminal strip. Then I used the holes as a templet to drill the chassis for 2 #6 SS self tapping screws. On the other side of the L I mounted two 170 volt 10 watt zeners with insulating hardware. I mounted a 25 K 20 watt resistor across the terminal strip and wired the resistor and zeners in series. I put a 200 K 1 watt resistor across each zener to act as a voltage balance and bleeder. I put bypass caps caps around it. Then mounted the assembly in the chassis with a little thermal compound for good measure. I wired the screen tap at the junction of the resistor and top zener. The high voltage to the free end of the resistor and the return to the bottom zener. Result solid screen voltages with atomic yealows.
KISS

[Bacon, WA3WDR]

Regulated screen voltage is good, but then you might want to make the control grid bias adjustable, because the screen voltage adjustment is how you set the resting current in a stock Vilking II.

If you can get a bit more control grid bias voltage, or use lower-bias modulator tubes like 6550 or 8417, I recommend adding one resistor from the screen regulator to the screen of one modulator tube, and another resistor from the screen regulator to the screen of other modulator tube.  (One lower-value series resistor to both screens in parallel doesn't make good audio). Something around 2K will probably be right, but the regulator needs to be about 50 volts high, and the grid bias needs to be extra negative. The idea is that the screen voltage will not be low on audio peaks.  This makes the tubes more linear, and still provides full output power.

See "IMPROVING THE LINEARITY OF SCREEN GRID TUBES"
http://amfone.net/AMPX/110.htm

[WA1GFZ]

Your mod will allow the screen voltage to drop as the screens draw current.
Why bother regulating it. I plan a Zener regulate my bias a bit lower because the screen voltage is a bit lower 335VDC  so resting current is down a bit. I will do this when I decide on an audio driver. considering 3 FETS splitter and 2 source followers.
The gain of the tube will drop if the screen voltage drops....peak compression??

[W2JBL]

i have had good results on my Viking-II with running the 807 screens off the low voltage supply and clamping the modulator bias with a string of zeners to keep me at 60MA resting current.

[Bacon, WA3WDR]

As long as you get enough plate current on peaks, you can lower the screen voltage a bit, and reduce the bias.

The idea of the series screen resisitors is to let the screen voltage dynamically pull down a little bit on peaks.  The right amount of series resistance linearizes the curvature (mostly concave upward) of the plate transfer characteristic that is typical of power tetrodes and pentodes.  Too much series resistance overdoes it and makes a sort of soft clipper.

But the screens draw current at different times.  If they are tied together in this circuit, they will pull each other's voltage around.  This is why a single resistor, or a badly regulated supply, is not the way to go.  A regulated screen supply with the right amount of series resistance to each screen grid is best.

On top of this effect, the screen current also varies inversely with plate voltage, and this introduces some negative feedback.

[WA1GFZ]

I see your point but my screen voltage is right on solid at about 335 volts. I would prefer running lower screen voltage so there is wider plate swing before you draw screen current. Soft clipping at the peak does sound interesting though. It is hard to prevent distortion when the screen conducts. A tube like the 6DQ5 runs low screen voltage which would allow even wider plate swing without screen current.
Screens off the low voltage is less heat in the box but remember you need a HV bleeder.

[Bacon, WA3WDR]

The 820 ohm resistors I used in my 8417 Viking-II modulator were 2W units that were lying around at the time, and my test was by ear, but the improvement was obvious.  The only place where I tried an X-Y scope comparison of input versus output was on a push-pull 6L6 amplifier I made out of an old Bogen PA amp.  In that unit the plate and screen are at about 350VDC, and resistors of 5.1K were about optimum to each screen grid.

Since I wanted to maintain maximum power in the modulator, I raised the screen voltage a little bit so that the sag on peaks would pull it down to about the original value.  This meant that I needed a little more negative bias on the control grids, so I lost a little bit of voltage gain (not much).  But the screen voltage was the same on peaks, where the plate voltage gets low and screen current rises, as it was before.

I think that this technique would be especially good in 6DQ5 amplifiers, because the 6DQ5 has a lot of curvature in its dynamic characteristic.

[John K5PRO]

After finding how unstable the screen voltage and idling current drift was on my DX100 1625 modulators, I converted to using the 448 VDC LV supply, with a relay that is keyed by the HV switch. That supply would fluctuate thusly:
417 VDC with no modulation
412 VDC when talking, no audio processing

I converted to 6CA7/EL34 audio pentodes, since matching 1625s was nearly impossible - resulting in one tube always glowing red while the other didn't. The EL34s were used in the Apache. Using a separate substitute lab supply I found that EL34s liked having 400 VDC on the screen, when the plate is at 770-800 as in my DX100. I built the circuit shown, after trying to use stud mount 50W zeners and a resistor. It uses an IRF820 mosfet as a series device, regulated with a 400V small leaded zener. The entire circuit is slightly larger than the TO220 case, and is mounted on a vector board off a standoff, with a small heatsink on the FET. Its only dropping 20 volts or so. This worked to 'regulate' the widely varying LV B+ without such massive power dissipation that a zener string requires. Also it eliminated most modulator idling current drift, and the tubes were more stable with tones or just idling, without trying to run away.
Test results with -38.2 V grid bias, set with cathode zeners. Modulator idling current was ~30 mA per tube, with 20 mA of screen current per tube:
 
audio condx        Vin            Vout
no mod              420            398
50% 1 KHz         413            397
95% 1 KHz         408            394
90% 400 Hz        406            393

With the higher cathode emission of EL34s, the DX100 had no trouble making 115% positive modulation when processed assymetrically, at 130 watts carrier.

[WA1GFZ]

I was about to use a FET but found a pair of 170 volt zeners. The effect is the same zeners fets linear regulator is a linear regulator.
Use the stronger 1625 on the positive peak side. Cool idea on the cathode zener. I was considering that.
I'm running about 40 ma resting and it is pretty stable. 807s no color. fc