diyAudio (
-   Tubes / Valves (
-   -   Unity-Coupled amp project (

SpreadSpectrum 18th February 2012 05:07 PM

Unity-Coupled amp project
1 Attachment(s)
I started this a while back. I have all of the physical construction done, everything is mounted to the chassis. I am doing everything on PCBs so things are going slowly one step at a time. Supplies are all to be regulated and bias is to be controlled by circuits inspired by some Broskie posts. Should require little to no adjustments. Should be a pretty complex/fun project. Output transformers are PAT1070UC.

Anyway, I thought I would throw my basic circuit up and let you guys throw rocks at it. Don't hold back.

SpreadSpectrum 20th February 2012 03:53 PM

Nobody? I guess I'm mostly looking for feedback on dropping screen voltage through the Zeners. Anybody ever tried anything like that? Should I bypass the zeners with caps?

These output transformers are about 6.4k:8 impedance ratio, so if I kept screens at 450V (which the basic McIntosh circuit would do) the load line would cross way below the knee of the pentode curves. I was afraid this would be hard on the screen if this thing were overdriven. Options for dropping the screen supply are limited in this output configuration, hence the zeners.

SemperFi 20th February 2012 10:25 PM

Cool circuit, but the use of 841 as gain devices seem a bit over kill, but Im sure they'll add some awesomeness to the amp. I've never used zeners in the signal path b/c I assume they add harshness. :-). Since u plan to use regulated supplies, why not for the screens? (remember tho, fixed bias all around begges for reduced tube life).
Other than those uneducated comments, I think ur design look good. (is global feedback back?)

Edit: I see its triode connected now, so of course now I see ur point in dropping the screen voltage with series zeners. Perhaps a cap in parallel with zener?

SpreadSpectrum 20th February 2012 11:30 PM

Actually, it is not triode connected. It is the old unity-coupled circuit used in all of the McIntosh amps. Screen goes to opposite plate. The strong cathode feedback results in full pentode power with characteristics of a triode with mu=2.

I am very excited about the 841s. It was a requirement for the amp to have a thoriated tungsten filament somewhere:D. Seriously, it is not easy to find a linear tube that can idle at 400V plate and deliver 200V+ rms volts signal. They will require a little grid current drive, but it as also a requirement that this amp have A2 drive somewhere in it as well:D.

I think I will put a cap in parallel with the zener. I'm thinking there could be some voltage variation across the zener as tubes cut off and screen current disappears/reappears.

I don't know if global feedback will even necessary for this thing. Open loop damping factor should be better than lots of amps closed loop. I just threw it in there as an option. Should be about 6dB if I calculated correctly.

This should be a fun project. I just got my regulator boards in, and I can't wait to finish this power supply.

Chris Hornbeck 21st February 2012 12:45 AM

A variation that might be fun to try is "Ultralinear", which for these amps means just bringing G2 's back to a good power supply. It's 50% taps, but close enough. Also a convenient way to have differing anode and G2 voltages.

All good fortune,

bear 21st February 2012 02:22 AM

Will a zener just going to the screen actually drop any voltage, or will it float?

Or will it drop voltage only when the screen draws current...


SpreadSpectrum 21st February 2012 03:03 AM

When there is no screen current there will be no voltage drop across the zener. That's not a big deal since the tube is cut off and is effectively not there. The problem would be as it turns back on (or as it shuts off), there would be some small variation in voltage across the zener. If we assume that it is always best to keep screen voltage constant in a beam tube then we may want to bypass the zener with a cap to decrease these voltage variations.

The question is, how big of a cap do we need? I'll have to think about that a bit.

bear 21st February 2012 03:20 AM

The screen can be kept free of AC if you wish. Afaik. :D

You might want to drop a highish value resistor to ground from the screen, keeping the zener alive unless the AC drives the whole side "off"...

The size of the proposed cap will depend on the effective impedance the screen presents, which if drawing current will look fairly low... otoh you do not want a cap that can hold a charge and float the DC high at the wrong time...

I think you might do better with a straight voltage divider to supply a % of the total voltage. Then the question is (again) do you want to drive the screen with AC or not?



SemperFi 21st February 2012 07:34 AM

Ah, I see now, was using my iphone last night and didn't get the whole picture. I agree you should ensure the zener never cuts off, and a resistor to negative voltage rail may help.
Looking at the voltage swings on plates and screens I am a little puzzled this doesnt destroy the tube...Lets say you are close to max and swing 350volts on the plates. If mu is 2 you swing 175volts on the grids. So V7 has almost -200v on grid, -100v on screen, and +800v on plate. No harm here, it is completely shut off. V9 has +150v on grid, +600v on screen, and +150v on plate. Wont the screen current in V9 be a tad high?

Interesting connection tho, I need to ponder this.

If no global NFB are u going to include local feedback (cathode resistance)? Gain will be a little much as is? Perhaps go from 6SN7 to a lower gain DHT on the input as well?

Kay Pirinha 21st February 2012 11:13 AM

Please have a look at the MC275 schematic. It uses the same power tubes as yours, running both screens and plates at 450 volts. As this well reputed design doesn't feature zeners for screen supply, I don't see any reason for them.

Best regards!

All times are GMT. The time now is 06:53 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright 1999-2015 diyAudio

Content Relevant URLs by vBSEO 3.3.2