Bob's "convertible" SE SV572 Amp

Preliminary Release - 11/19/97

Last Update - 12/09/97


INTRODUCTION

This project makes use of at least two of the new SV572 series audio power triodes from Svetlana.
This series includes the SV572-3 , SV572-10 , SV572-30 , and SV572-160 .

The circuit topology is easily re-configurable ("convertible") to use either the -30 (in Class A2 - Zero Bias operation) or -3 (in Class A1 operation).
In addition, the -30 can be configured for two different drive schemes (standard grid driven configuration or Cathode Driven . A set of switches are used to reconfigure (with power off). But for clarity, the schematics are re-drawn for each mode as shown below:


MODE 1: Zero Bias SV572-30


Mode 1 is sort of a natural progression from my SE Class A2 SV811-10 amp. The higher mu (30 for the 572-30 vs. about 10 for the SV811-10) means that very little drive voltage is required. This allows the use of a large step-down ratio in the transformer drive. The 10:1 reduction of signal voltage pays for itself by reducing the drive impedance by 100:1.
This is important because the grid impedance of the output tube dips to about 2K on positive signal peaks. The driving impedance of approximately 20 ohms minimizes grid current induced distortion.

I stuck with the zero bias arrangement mainly for its simplicity (although I do think that there are some other interesting features of Zero Bias Class A2 operation that I wish to examine a bit more thoroughly - like near-zero 2nd harmonic distortion...).
This new amp also takes advantage of the higher plate dissipation rating of the SV572 series. The SV572-30 is run at a much higher voltage than was used in the SV811 amp. The operating plate voltage was chosen so as to draw a bit less than 100mA under zero bias conditions. This is near the limitations of the output transformers (UBT-2 max. DC = 110mA).

Calculated Performance

On paper, this amp should give about 12W @ >1%D2 into a 4.8K load, and about 11W at virtually zero D2 into a 9.6K load assuming +/-30V drive (about the limit of the driver at this point).
Almost 20W would be achieveable if the driver could swing +/-40V....

Measured Performance

I measured +18V to -14V peak to peak at clipping. This translates to about 16W. Grid drive was about 35V peak.


MODE 2a: Class A1 SV572-3 (Fixed Bias Option)


This is a neat trick (IMHO). Since the SV572-3 has about 1/10th of the mu of the SV572-30 it needs about 10 times the drive voltage of the '-30. And it just so happens that the step-down ratio of the driver transformer is 10:1. Cool!
This means that if we take the output from the primary of the transformer we have 10 times the drive. The primary of the transformer acts as a "choke-load". It presents the driver tube with a very high impedance(practically constant current) load. This increases the gain of the stage to almost the full mu of the tube. It also minimizes distortion from the stage. Another benefit of the choke load is that it's DC voltage drop is relatively small (determined by the few hundred ohms of winding resistance). The peak-to-peak output possible from such a stage is actually GREATER than the supply voltage (but somewhat smaller than 2 times the supply voltage). The limitation of this type of arrangement is that the choke must have high enough inductance to present a reasonably high load down to the lowest frequencies you wish to reproduce. The choke must also be sized to handle the DC plate current for the stage. Of course, this is all taken care of in this application, because all of the same criteria apply to the primary of the transformer. If it worked in MODE 1, it should work for MODE 2!

A coupling cap is required to couple the driver's plate signal to the output grid.
A grid resistor is required to feed the negative bias supply for fixed bias. These parts and bias supplies are built-in to the amp. A switch network (not shown for clarity) disconnects the transformer secondary drive (from MODE 1) and connects the cap/grid resistor/bias supply.

Calculated Performance

The bias supply is adjusted for a plate current of about 100mA. Ths works out to be about -120V or so.
So the operating point is about Ebo=710V, Eco=-120V, Ibo=100mA.
The calculated Class A1 power output is as follows:
12W @ 5.5% D2 into 3K load
10.5W @ 3.7%D2 into 4.8K load


MODE 2b: Class A1 SV572-3 (Cathode Bias Option)


This is just another way of doing mode 2a, above. Using cathode bias makes the design a little simpler. It does away with the bias supply, and it makes the switching network a little simpler.
(I just leave the grid resistor connected to ground at all times)...
I tend to prefer the sound of cathode bias amps over fixed bias. This is no exception. For the slight loss of output power, and another 10W of dissipation per channel, I think I may stick with cathode bias.

Calculated Performance

The cathode bias voltage is about 100V, the plate (to cathode) voltage is about 610V, and the plate current is about 100mA. The cathode bias resistor works out nicely to 1K.
The calculated Class A1 power output is as follows:
8.9W @ 1.7% D2 into 4.8K load.

Measured Performance

I measured a fairly symmetrical +/-14V peak to peak at clipping.
This translates to about 12W.


MODE 3: Zero Bias SV572-30 - CATHODE DRIVE


This may be an even neater trick!
One of the difficulties associated with a Class A2 design is the grid current drawn by the output tube. The worst part is that it only occurs on the positive half of the grid voltage cycle. During the negative portion of the grid voltage signal the grid impedance is VERY high. Then during the positive part it drops to about 2k. This causes distortion. The lower the output impedance of the driver, the lower the distortion.
Switching to CATHODE DRIVE helps solve part of the problem. By connecting the grid directly to ground we achieve a low impedance path for the grid current. The input impedance at the cathode, although it is relatively low, is at least (fairly) constant.
I believe it is something like 1/gm + Rl
where gm is the transconductance of the tube 1800uS
and Rl is the LOAD impedance.
For the SV572-30 gm is about 1800uS, so 1/gm=556ohms. Adding to this the typical load of 4.8K we have a total input impedance of 5.3K. This is not too bad as far as the driver is concerned. We could even use a 9.6K load if we wanted to improve matters even further. Of course, this neglects the fact that the speaker itself is not exactly a constant load. In practice, the actual speaker load reflected to the primary will result in a variable Rl. So once again, we have some distortion due to varying input impedance. In this case, it will probably be more subtle than in the grid driven case of MODE 1. However, it is likely to be of a more complex nature as opposed to the predominantly 2nd harmonic distortion caused by grid current draw.
To be honest, in brief listenting tests I really couldn't HEAR much of a difference. Eventually I'll get around to some firm measurements....

Another novelty/feature of this configuration is that the DC cathode current from the output tube flows through the secondary of the driver transformer. The way it is wired, this secondary current effectively cancels the DC flux induced in the primary by the driver tube. With careful selection of operating points, the net DC flux in the driver transformer can be reduced to zero.
As it turns out, the transformer I am using (actually a surplus vertical output transformer from a tube TV) is designed to handle a fair amount of DC (at least 20mA). But cancelling the DC in the core would simplify the design of this transformer if we were to start with a clean sheet of paper. I've also been looking into more readily available substitutes for this item. If the net DC can be cancelled, "70V" or "25V" line matching transformers (used in PA distribution systems) could be a possibility....

Calculated Performance

On paper, this mode should do about the same as in MODE 1:
about 12W @ >1%D2 into a 4.8K load,
and about 11W at virtually zero D2 into a 9.6K load assuming +/-30V drive .
The actual output power might even be marginally higher since the drive power is actually ADDED to the output... another benefit of cathode drive.

Measured Performance

I measured +/-15V peak to peak at clipping. This translates to about 14W. I was really suprised to see how straight and symmetrical the X-Y trace was on the scope (suggesting good linearity). This apparently comes from the negative feedback effect of having the plate (cathode) current modulate the input signal. The effect is similar to the degeneration achieved when a cathode resistor is left un-bypassed in a common cathode amplifier. Linearity is improved at the expense of reduced gain in the output stage. This was also borne out by the fact that I now had to turn the volume to near maximum to reach clipping, whereas with the grid driven topology it only took about a 50% setting.



Listening Impressions

I've only briefly listened to this amp in each mode. So far, here are some of my initial impressions.

The SV572-30 in Class A2 Grid Drive (Mode 1) sounds really powerful. There seems to be way more than 12W on tap! And I'm pretty sure I'm not getting all I can out of it.
I plan to boost the driver power supply so I can get closer to 40V peak to the output grid. The amp overload is fairly gradual, and appears to occur at the driver way before the output tube. This is especially true with a 9.6K load. Again, the driver supply voltage boost should help out here...

The SV572-30 in Class A2 Cathode Drive (Mode 3) sounds very similar to Mode 1, as described above. Perhaps once the driver stage overload is taken care of, we will start to see how cathode drive is helping us (!?). Have not seen the effects of grid current yet in Mode 1. Mode 3 should not show these effects, so it should be cleaner. (It certainly does look better on the scope...)

The SV572-3 in Class A1 (Modes 2a & 2b) sounds very different from the others. This arrangement brings me closer to the "Single-Ended" sound usually associated with really nice 300B and 2A3 amps. There is a "life-like" quality to the sound and the image rises to about 3ft above my speakers.
The interesting part is that the sound is Super-Clean. For those who don't care for the harmonic-rich "tubey" 300B-type sound, this gives you the best of both worlds!
Switching to cathode bias (Mode 2b) warms up the sound a bit (the way I prefer it). But this is certainly a unique sound. Perhaps I need to listen to an 845 amplifier. I suspect that this is the sound that I've heard all the raving about...


A quick summary:

The SV572-30 simply ROCKS! Warm and loud. Perfect for Rock-n-Roll. Does quieter stuff real well too, but it's a shame to "waste" the power ...
...

The SV572-3 gives a solid, clean, punch to "Single-End" sound. For clinical reproduction freaks who still want to keep the "life" in the sound, a fixed-bias -3 is the way to go. So far it does all kinds of music really well. It really shines with traditional or electric jazz. You oughta hear the "SNAP" of the drums on Chick Correa's "Light Years"... Cathode bias increases the "FUN" factor for me. This amp breathes along with Sarah Vaughn. Then it gets up and dances to Bob Marley...!

MUCH MORE TO COME --------- STAY TUNED!!!!!!

Email your comments/suggestions/critiques/questions to Bob. D. at robert.m.danielak@lmco.com


Back to Bob. D.'s Music/Audio Page
Back to Bob. D.'s Home Page

This page, and all contents, are Copyright (C) 1997 by Bob. D.