Single Ended Class A2 SV811-10 Stereo Amp

Last Design Update - 9/25/96

Comments / Suggestions are welcome

Here is the schematic for one channel of my amp as it is built right now:

I finally got around to scanning in a few pictures of this amp (1/11/98):

SV811 #1

SV811 #2

SV811 #3

INTRODUCTION

This project actually began its life as a Single-ended 211 design. I received the Parts Connection Catalog just before Christmas (1995). In the back were the data sheets for various tubes, including the lovely 211. I had been hearing a lot of good things about this tube, so I started pokeing around. Most designs I found used a B+ between 850 and 1200VDC. While I wasn't _too_ put off by these relatively high voltages, I wondered if I couldn't just come up with something a little more manageable. Around the same time I began top hear a lot of noise about Class A2 designs using 801's or 805's, etc. Hmmmm.......!

After playing around with the curves I noticed a nice symmetry about the OV bias curve in the area of 400V plate voltage. It seemed that for Ebo = 420V, Eco=0, Ibo=80mA a 10K SE load would yield about 6W at < 0.5% distortion given a +/- 40V grid swing. Similarly, a 5K load yields 7.5W at 1.5%. The trick, of course, was providing linear drive to the 211's grid realizing that it is going to draw grid current on each positive half-cycle.

Several low impedance driver options came to mind including interstage transformer (step-down), Mu-Follower, SRPP,White-Follower, and Cathode Follower. Also, various tubes were considered for use in the final driver including 6SN7, parallel 6SN7, 5687, 6DJ8, 6BX7, 6BL7, and 6AS7. Also considered were various triode connected pentodes, such as 6L6, 6V6, 6W6, EL84, EL34.

After much deliberation, analysis, and simulation, I began to play with some preliminary layouts for a stereo amp chassis (I don't care much for mono-blok designs) for various circuit topologies. It became apparent that in order to maintain a reasonable footprint I'd have to keep the parts count down to a minimum (a good thing, anyway!). I figured I'd start with the simplest approach and work my way up the complexity web as neccessary.

Having firmed up my plans, it was now time to buy some TOOOBS! Unfortunately, Antique Electronic Supply no longer carried the cheap Chinese 211's, and NOS units were about $150.
(....oh!)
Somewhere, I had heard that the New Svetlana SV811-10 and SV811-3 were supposed to be similar to the 211 and 845, respectively, so I picked up a couple of -10's (ya never kno'...).

I developed a model for the SV811-10 based upon the limited info provided by Svetlana's data sheet. I extrapolated the curves to include positive grid bias and plate voltages over 500V. Just recently (9/96) New data sheets for the SV811-10 have been published. The curves now extend to plate voltages above 1200V and they now include positive grid bias curves. (Hooraaaay!!!)
As it turned out, the design didn't really change that much at all. The new data sheets confirmed this (but by that time I'd already built the amp - on FAITH!).

Design Details

Input/Driver Stage

The input stage is comprised of a one half of a 6SL7 octal dual hi-mu triode configured as a conventional common cathode gain stage. This is then cap coupled to one half of a 6SN7 octal dual medium-mu triode, also configured as a common cathode gain stage. Cathode resistors are un-bypassed, for now. These two circuits are powered by B1+ supply (about +400V, or so) which is decoupled from B+ (about +420V) via an RC filter.
The 6SN7 plate output is then cap coupled to a cathode follower using one half of a 6AS7, low mu dual triode. This stage is powered by a separate split supply (+/- 130V, or so) in order to DC couple to the 211 (oops!...I mean SV811-10) grid with the appropriate DC bias. The DC output is adjusted using the pot, R8. The intended useful range is between -20V and +10V. Any more positive and the quiescent plate dissipation limit (65W) will be exceeded. The nominal bias point will be 0VDC.
The cathode resistor is sized to pull about 26mADC thru the cathode follower. This value was chosen to exceed the expected peak grid current draw for the original 211 design (20mA). I have a feeling that the SV811-10 draws LESS grid current than the 211, but I have not confirmed this.

The output impedance of the driver without feedback is about 90 ohms. With feedback (Rfb installed) this value could be reduced to about 20 ohms, but this would probably involve bypassing the cathode resistor(s) in order to boost the loop gain. So far, it seems to work well enough without this added complexity.

Output stage

Couldn't be simpler...
Single ended, Fixed/Zero Bias. No Cathode resistor, grid resistor, etc. Just the SV811-10 and the output transformer.
Oh yeah.... about the output transformer....
I'm using one of the new "One-Electron OPT-2's". These have a Primary Impedance of 4.8K, which works out just fine. However, using the 4 Ohm output tap instead of the 8 ohm tap gives me an effective load of 9.6K. This loading (as was the case for the 211 design) promises to give much lower distortion at the expense of a little less power output. Depending upon the actual DC bias selected I could even get away with using the 16 Ohm tap for an effective load of 2.4K (more power - more distortion). The possibilities are endless...The ears shall be the final judge. More to come in the following weeks as I find time to work on it.

Power Supplies

The power supply section for the stereo amp is given below:

The Main Power supply is designed around a Hammond 600VCT/250mA tranny that I had laying around. Using silicon diodes and cap input to minimize voltage drop (and save space) I end up with about +425VDC under load at the first cap. The first RC filter provides a relatively clean B+ (approx. 415VDC) for the output stage. The second RC filter cleans the supply further for the more sensitive input gain stages. B1+ is approximately +400V.

A separate 200VCT tranny is used with a bridge rectifier circuit to generate the +/- 130V split supplies, as shown. No additional filtering is provided since this stage is not so sensitive to power supply noise.

The SV811-10 filaments are both AC powered from the 6.3VCT@8A winding off the Hammond tranny. (The 5V rectifier winding is not used.) The 6SL7, 6SN7, and 6AS7 filaments are DC powered using a 12VCT tranny and 5A schottky diodes. I will probably go to DC on the SV filaments at some point. I do have a noticeable hum in my 90dB/1W/1m speakers. still have a little power supply ripple to kill, too.

Parts Selection

I'm currently using aSovtek 6SL7, a NOS RCA 6SN7 and a NOS RCA 6AS7. The SV811-10's are obviously from Svetlana. Resistors are mostly Rat Shack 1/2 Watters, for now. The Cathode Follower Cathode resistors are overkill Dale 25W chassis mount units. The coupling caps are cheap (?) polyprops from All Electronics. All HV filter caps are 200u/300V photo flash caps (stacked as neccessary) due to availability and real estate concerns. Eventually, maybe I'll upgrade in some areas...

Construction Details

I built the entire amp on a 17" (D) x 13" (W) aluminum plate. This plate will drop into a nice burly maple box that my brother, Rich, will make for me. I cut the tube socket holes using a hole-saw on a low speed power drill. I use WD-40 as a "cutting-oil". I mounted all components between the socket terminals and strategically placed terminal strips (from Radio-Shack).
The 6SL7, 6SN7, and 6AS7 are placed in a line down the center of the plate with the SV811-10 off to each side. Main power tranny is centered at the back of the plate with the OPT's on each side. 200VCT tranny, 12VCT tranny, filter cap bank and all other components mounted underneath.

Performance

Calculated Performance ... (updated 9/25/96)

The following performance was calculated using the New SV811-10 curves from Svetlana:

0V bias, 5k load => 8.3W @ 0.9%D for +/-50V grid swing

0V bias, 10K load => 4.5W @ 0%D for +/-40V grid swing

0V bias, 10K load => 7.6W @ >5%D for +/-50V grid swing

0V bias, 2.5K load => 9.7W @ 1.1%D for +/-50V grid swing

0V bias, 2.5K load => 16.4W @ 1.8%D for +/-70V grid swing

Not too much different from my original computer simulation results.

Overall gain is about 25dB.
Input sensitivty is about 500mVrms.

Measured Performance:

With zero bias & 4.8K load, 0.5Vrms input provides max clean output of almost 9W (24 vpp into 8ohm). At this point the bottom half of the sinusoid starts to flatten (due to the increase in grid current). Max peak to peak output (top and bottom clipping is approx 30vpp into 8 ohm (14W peak).

At the max clean output 24vpp/9W, sv811-10 grid swing is +/-50V and looks fairly clean and symmetrical. This would suggest that the sv811's grid impedance isn't as bad as I would have thought. I figured that my driver's output impedance (without feedback) would be on the order of 100 ohms.

I tried fiddling with the grid bias (from -20 to +5V). negative bias shrunk the output amplitude and seemed to increase distortion compared to 0 bias. positive bias actually did about the same. 0 bias gave the most symmetrical output when on approaching max output. will experiment some mo'.

Tweaks

Listening Impressions

I was able to spend some time listening to these amps while I was working on a couple of other amps (futzing around with an interstage tranny driver for my pp2a3 amp and building a new guitar amp). I tried the 16 ohm, 8 ohm, and 4 ohm taps on my 8 ohm speakers. this gave me 2.4K, 4.8K and 9.6K loads, respectively. From my previous calculations it seemed that the 4.8k load should give the best trade-off in performance. About 8.3W @ <1% D2 Later I tried 2.4K on the new SV curves and found that I could get 9.7W @ 1.1% D2 with the same +/-50V swing and maybe even 16W @1.8% with +/-70V. 10K was only good for 7.6W with +/-50V but probably should be limited to +/-40V grid swing due to distortion at the positive extreme. The 2.4K load certainly gave the amp some authority. Very punchy, and loud. Good for rock-n-roll, I guess. I was initially impressed with the amount of clean power I was getting but soon had to turn the volume down, as I was getting a headache. At that point I changed back to the 8 ohm (4.8k load) taps. This arrangement seemed to sound much better. It sounded more "alive" - at much lower volumes too. Plenty of bass, and hi-end, too. I turned off the -3dB pad on my speaker to verify. (on my very rolled off pp2a3 amp there is only a slight difference with the pad in or out). Figuring that this was going to be as good as it gets, I decided to try the 9.6K load in order to hear the opposite extreme. Instead, I found that I liked this arrangement better! Not as punchy in the bass as either 2.4K or 4.8k but a lot tighter. Not so boomy. The midrange and highs seemed to be better blended. At higher levels, nearing distortion it didn't emphasize the 'esses' as much. The 'image' also seemed to rise well above the speakers. I tested this phenomenon with my brother and with my wife, independently. I told them each to point to the guy singing as I clip leaded to different OPT taps. Both of them raised their arms from about 15 deg angle to 30 degrees, cool... So here we go again. Another amp with about 4W output - only turned a quarter of the way up!?! Next chance I get I'll put it up against my pp2a3 amp and SE 2a3 amp to see how it compares. They each put out less than 5W. On the surface, the 9.6K loaded design does seem a waste considering I could have gotten 3.5 to 4W out of a simpler 2a3 design. I guess if i put in a switch or multiple 4/8/16 ohm binding posts on the outputs I could have a more versatile amp, since my present 2a3 amp(s) don't handle heavy rock too well ;) .

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

This page, and all contents, are Copyright (C) 1996 by Bob. D.