DCAudioDIY.com

Recently finished a complete rebuild of my nearly 20yr old PP 2A3 monoblocks. Fixing a few nagging problems with the original design, using a different driver topology, a modern and more robust power supply, and much nicer vintage potted power transformers (the old power transformers were cheap Allied by Hammond that buzzed and ran very hot). Still using a Hashimoto HW-60-5 output transformer.
Basic topology is a modified Mullard 5-20 front end, only using 1/2 of a 6CG7 as the input, direct coupled to a 5687 LTP phase splitter that is choke loaded on each section (using the primaries of a Lundahl LL1692A), and a CCS in the tail, which is cap coupled to the 2A3 grids. The 5687 running at 15mA per section and around 180V across the winding has lots of voltage swing so should be nice and linear. I would have preferred to use interstage coupling, but the Lundahl rings like mad, and it complicates negative feedback and compensation (using around 6dB global NFB). So to keep the 10K squarewave response nice and clean, using cap coupling for now (that can always change with further experimentation in damping the secondary). I am using cathode bias for now...fixed bias is an option should I go with 300B power tubes

Tubes are an issue. I started with my JJ 2A3-40s, since it was my original intent to use fixed bias with a B+ of 410V to achieve around 28W output. I was having a rough time with the JJ tubes, the emissions were really low on 3 out of 4 tubes I had, so never could bias them up per the published curves (they were ridiculously weak). I found myself in a situation where I needed to regroup and convert to cathode bias in the middle of the build and use Sovtek 2A3s that can withstand a little more abuse than most 2A3s. So I now have 360V across the tubes and around 51mA current, which runs the tube at 18-19W plate dissipation, over the 15W/300V spec, but within the experience base of others with the tube.

On the test bench, I get 16 watts with no visible distortion of a sine waveform with my generator maxed out (The Picoscope I use has a max 2V P-P output). According to an output stage simulator, I should be able to get 20W Class AB (it is Class A to around 7W).

So using the Sovtek 2A3s the amps have plenty of output for my speakers and sound pretty nice. I did manage to dig up a quad of FullMusic/Sophia 300B/2.5V tubes that biased up fine. Running at the 2A3 operating point they sounded more detailed, however as I increased the bias (reducing the cathode resistor value) to around 70 mA per tube, I did not like the tonal balance...not enough body, and after a day, when back the Sovteks at the old setting.

I plan to explore using 300Bs in the amp (along with fixed bias), but need to convert from the AC heaters to a space constrained DC filament heater solution (no space for Coleman regulator or CLC arrangement). Breadboarding different 5V DC heater supplies and measuring the output ripple and noise of each. I have seen some strong recommendations for Emission Labs against current regulation using 3 pin regulators, particularily low dropout types, better to use a 1.5A current limited 5 volt fixed regulator (LM7805 or similar) with a 5.6V power zener across the output to avoid over voltage during warmup. Interestingly, they also recommended trying a stabilized zener as a 5V regulator. I have the space to put in a pair of 6.3/3A filament transformers in place of the 2.5/3A transformers I am currently using in the 2A3 heater arrangement, and can squeeze in the power supply/regulators.

Well this has gotten to be a long post. I do have a question...

Any experience with using the Chinese (Shuguang, Psvane, Linlai) 2A3C (big bottle). The envelope should be able to take the plate dissipation (same as the 300B), and I am assuming the plate and filament is the same (JJ uses a centertapped/folded 300B filament for its 2A3-40, and wonder if the Chinese tubes are similar. Also, the Russian monoplates can unofficially withstand a voltage up to 400 to 450V, don't know if the Chinese monoplates can do the same. If I find 20W work for me (dont' get too greedy), then it would be nice to have some more choices.

Oh, pictures..

Forgot to add, that I have seperate power tube and driver tube power supplies. Damper diodes/SiC Schottky bridge for the power tube, 6C4 rectifier/SiC Schottky bridge for the input/driver tubes.

_________________
Walt

Looks great! Sovtec 2A3s can take a beating and sound pretty nice, they are a well balanced sounding tube. I have run Sovtec 6A3 at at 300B operating levels (70mA, 450v) with no issues. I always suspected they just stuffed a 300B into a smaller bottle and called it a 2A3.

If they did, it would be a 5A3…..

_________________
I can explain it to you, but I can’t understand it for you.

That was how I ran it!

Even run at 5v, Sovtec 6A3 sounded fine.

Pete, Roscoe,

Well, I will have to try increasing the current a bit on the Sovteks (which really gets me deeper in Class A rather than more power). The tradeoff is when they start to sound hard, which was the case with the TJ 300B/2.5V, they sounded much better around 52 mA than at 70 mA (still not as tonally balanced as the Sovteks). Maybe the sweet spot is closer to 60mA. But I just dug out a working quad of Shuguang 2A3Cs and put them in. Unfortunately, one has a really loose base that I need to fix. But they sound more liquid and musical than the Sovtek tubes on first listen this morning. I will let them stay in for a while. I also have a quad of Shuguang 2A3Bs, which is the smaller envelope like the Sovtek 2A3s. I may just spring for a quad of Psvane or LinLai 2A3C, they are pretty cheap.

However, JC Morrison (who was involved in the design of the Sovtek tube when working at EH) maintained that the 15W max plate dissipation and 300V max voltage is real. Running them hot increases the emissions on the filament causing them to wear out faster. However, I have heard of SO many people pushing these tubes well beyond the voltage or thermal limits and getting good life, so there must be alot of conservatism in the design.

David

Ditto on the superb workmanship. As far as the signal generator output, I mentioned this generator I got last year. It works great and can output 20VPP.

https://www.amazon.com/dp/B08T9H9FBY?ps ... ct_details

Tom,

Thanks, I will have to check that out. My old Heathkit signal generator I built 25-30 years ago is not working so well, tough to get a clean stable waveform, probably needs re-capped. Modern devices should be cleaner, more reliable, and have better features.

David

It also functions as a counter and has two channel output which can be different signals.

Tom