High Quality Push Pull 6B4 Bias-Shift Amplifier

[KM1H]

Its the old spec that keeps getting people in trouble. When it is written as Absolute Maximum Values it does not meant that you can run them all at the same time.

Carl
KM1H

[WBear2GCR]

I was incorrect in what I wrote earlier - with a 6Hz. time constant the time to react is 1/6 sec, not 6 secs as I wrote... still slow. Also the text indicates there is an even slower time constant after the intial bias set circuit, fyi.

                      _-_-bear

[WBear2GCR]

Jim,

There is going to be a reason that they did not "speed it up".

Likely it causes the circuit to have some nasty instability. Perhaps akin to the situation where the time constant in an AB1 amp is NG and the coupling cap can sustain a DC charge due to a LF transient and then cause the tubes to cherry due to the bias shift.

I think you will need to keep the amp DC coupled WRT drive if you choose to try this dynamic bias idea.

Again, why not just bias it for AB2 and be done with it? Fewer parts and less to worry about.

Unless you need or want that "clipping" feature that the original design tried to do in the modulator stage, why bother keeping it in class A, and if you use the dynamic bias trick, that feature is gone anyhow...

                 _-_-bear

[WD5JKO]

Again, why not just bias it for AB2 and be done with it? Fewer parts and less to worry about.

  Bear, with the Gonset G50, the 6L6 modulators are in parallel, class A Heising. I cannot conduct less than a full cycle without severe distortion, so class AB2 is ruled out. What I was thinking however, was driving the modulators with a cathode follower, class A2 (draws some grid current), and some bias shift on voice peaks. Then use about 6 db NFB to linearize the kinked transfer curve as the bias moves about. With the mods already done, I get 100% modulation now on voice peaks, but not when sustained:

http://pages.prodigy.net/jcandela/G-50/YEA_AM.JPG
http://pages.prodigy.net/jcandela/G-50/Overload_mod.jpg

  When overloaded, the cathode bias increases, shifting the bias the wrong way, and also reduces the plate to cathode voltage.

   
    I have a second application for a bias-Shift scheme. I have a modified Central Electronics 20A where I am using a 6S4 to drive an E34L (a penta version of the EL-34) instead of in stock form it has 1/2 12BH7 driving dual 6AG7's. Here is the schematic:

http://pages.prodigy.net/jcandela/CE20AQRO/RF_Schem.jpg

    This transmitter puts out almost 50 watts on 80m, and about 35w on 20m full bore. I had it on the bench doing work, and at one point I had a power supply failure resulting in the E34L screen voltage going from 300v to 400v. I didn't know it when I keyed the rig. The wattmeter read over 100 watts on 20m!! I was shocked, shut it down, and checked my coax hookups to wattmeter and dummy load. I keyed again, to check SWR, 1:1. What the heck? Looked at the E34L, and it was not red, but soon turned cherry red and then I shut it down. I was fooled by the E34L since it has a similar tri-metal plate like a 6L6GC such that it does not show any red at all until the entire plate is hot as hell. The increased screen voltage (300 -> 400v) raised my no drive idle current from 30 ma to almost 200ma, and that was at 600 volts plate voltage!!  Even so, the RF output was 3X normal!!

    So my thinking for this 20a is to run the E34L at 600v plate, 400v G2, and set the G1 bias for about 40 ma plate idle current. Then when I drive it, shift the bias such that the operating point moves towards 200ma quiescent but only with full drive. Then play around with the idea on both SSB and AM linear operation. This is probablly a flawed experiment, but most likely would be a lot of fun. 

Jim
WD5JKO

[WBear2GCR]

I might be missing something on this, but if you've got two tubes and running heising, all that you need to do is to fit in another mod iron - a P-P one - and ur done. You don't need a gapped mod iron... so almost anyone that fits physically and has the right impedances will work nicely.

dunno, but that's the way i'd probably go with it - certainly it will maintain tube life better...

                    _-_-bear

[WD5JKO]

I might be missing something on this, but if you've got two tubes and running heising, all that you need to do is to fit in another mod iron - a P-P one - and ur done. You don't need a gapped mod iron... so almost anyone that fits physically and has the right impedances will work nicely.

  Ok Bear, I actually DO have a P-P transformer in there. I don't know what to call it, but this is a variation of Heising that allows full transformer interleaving, no core saturation and hence, no core gap. Carl was right, no amount of simple diddling with the G50 can fix that audio. The stock mod transformer just has way to little inductance with 1/4 amp flowing through the primary. With my change, I am flat from 50-5000 hz at about 80% modulation (with pre-emphasis feature removed). Before the low end suffered greatly below 500hz. Here is my post about this with links to pictures, schematic, etc. Scroll down to reply 7 at:

http://amfone.net/Amforum/index.php?topic=18961.0

   So call me a crazy, but this class A modulation is really clean, and without the bad stuff from mod transformer leakage inductance. If I had a mod tranny with an voice coil winding, I might have done different. With the approach I have, the part I don't like is the cathode bias that rises as I approach 80% modulation; such that 100% is possible only momentarily on voice peaks before that 1000 uf cathode bypass cap charges up.

    If I just went to fixed bias with no cathode resistor, that problem would go away. Problem is however that the two stage cathode current (rcvr audio = LO AND xmit audio = HI) goes away, and I'd need to add a bias supply. So instead I was looking at the bias shift topology along with ability to drive them 6L6's into grid current (class A2) to get a little more power out of that modulator by increasing the plate downward swing closer to the cathode.

  Thanks for the comments..

Jim
WD5JKO 

[WD5JKO]

Brian,

    Wow, what a piece of beauty that modulator deck is. You make really nice looking stuff!

   I been following your posts about the audio driver, scope plots and all. It seems that the plots you show are from one plate to ground AC coupled through an orange-drop isolating cap. Near overload it shows considerably asymmetry, typical of high 2nd harmonic distortion. If I understand your setup right, then much of this will be canceled out in the transformer where the 2nd harmonic distortion is largely canceled out so long as the P-P circuit is balanced.

   I built a P-P 6B4 driver way back in the mid 80's (still got it), that drove 4 X 808's in P-P parallel. That was a 600 watt modulator with 2250v on the 808 plates. While developing the 6B4 driver, I would run it into an open circuit, and then into the rated load based upon the tube specs and transformer turns ratio. The 6A3/6B4 series of tubes has VERY low plate resistance so the output should not drop appreciably when loaded down with a reasonably resistance, typical of a voltage source with low source impedance.

    Remember that with 810's Class B with grid bias, the driver load varies from an open circuit, to being almost half cycle loaded by grid current once the grid swing exceeds the bias voltage. The RL varies moment by moment all over the place. Look at the plate curves for an 810. In there they should show grid current versus grid voltage, then do an R = E/I to figure the RL your driver sees (each side of transformer CT) at a specific period in time when driving the 810's into grid current. Figure the maximum grid swing needed from the 810 Class B audio data sheet, figure the RL the driver sees at maximum + grid voltage. Then use this RL across one side of your driver stage transformer secondary. Can you drive to that peak level without overload? Remember that the grid current is only one side at a time, so one 810 shows an open circuit while the other is pulling grid current.

     At one point I developed a solid state direct coupled driver. With an 808 on the bench as a load (no plate voltage), I could swing the grid to + 100 volts where the 1/2 cycled grid current peaked at about 1 ampere. The unloaded voltage was the same as the loaded voltage, i.e., zero source impedance. With a tube driver this is not possible, but I did try to get my 6B4 driver to do as well as possible. With my 6B4 driver, I used NFB to lower the source impedance using up some reserve surplus audio gain. These tubes can do a very respectable job in this application.

    I am curious how the Brooks bias shift circuit might change the circuit behavior.

Regards,
Jim
WD5JKO

     
     

   

[N3DRB The Derb]

yeah, get spastic wit da fantastic Glastics.    That's the right tool for the job.

[N2DTS]

Yes, very nice work!
I like keeping the heavy iron off the chassis, and use a rack mount shelf just for the mod iron.
For the power iron, it goes on the floor in the bottom of the rack cabinet.
Big chokes are bad enough without adding any more iron to decks...


Be carfeful with the 3 diode peak limiter, there is something with the response time of the diodes that can cause you to have a real wide signal. I used to run it on my 813 rig, and got reports of being wide as a barn door.
Even tho I did not run it high and set the neg limit at 90%.
Have someone look at your signal with it in and out of circuit....

I think its MUCH better to unbalance things at the low level stages, in my 32V3 I did ab1 modulator tubes with direct (no driver trans) hookup and unbalanced the driver for more gain in the positive direction.
I get more positive peaks that way.
Yes, its distortion, but its good distortion....

Brett

[KM1H]

I believe Don says diodes blow mod transformers in his experience?

Carl

[N2DTS]

I use/used the 14kv 1 amp bricks k2aw sold.

Not just a little wide, but real wide, although maybe different diodes would work better....

Brett
 

[WD5JKO]

>> At least the guys back east will be able to hear me better


   Brian, maybe you can teach those guys practicing "East Coast audio" a few tricks.  By tricks I mean to actually modulate to a sufficient depth to be heard and understood when the receiving station is listening beyond one hop of propagation. Here in Texas I all too often find that when listening to the W1's, W2's,W3's, all I hear is S9+20 carriers with little or no audio coming through. Then Timtron comes in, with wall to wall audio...

Jim
WD5JKO