What do people think of the SETH 2A3 PP?

[theskintman]

After much reading, researching etc., I think I'm [this close] to undertaking building a couple of SETH 2A3 PP mono blocks.

Just wondering if anyone else has been down this path, and what do they think of the results?

I've previously only built from kits, so the very clear instructions in the PDF helped make this decision for me, along with the fact that it does seem like a quality bit of equipment.

Link is here

 

[Miles Prower]

Given that this circuit doesn't include gNFB, the performance is going to be very dependent on the quality of the iron used here, especially that of the splitter autotransformer. That type of splitter is often used for RF amps, where you tune it with a split stator capacitor, and can include a balance trimmer. Not so common with AF amps. I would also ditch L1 and replace it with a grid resistor.

The worst flaw is the 2A3 bias circuit. Filament xfmrs aren't designed to pass audio frequencies, and so I'd replace this arrangement with filament balancing resistors and connect the bias RC network at the center point. (See attached)

 

[arend-jan]

The worst flaw is the 2A3 bias circuit. Filament xfmrs aren't designed to pass audio frequencies

I thought, that with push-pull the audio currents are not going through this transformer, only the PP error currents (input/output) are?

Anyway, I have this exact arrangement on the cathodes of my Amity and I can't say otherwise than this amp sounds fine.

regards,
Arend-Jan

 

[Miles Prower]

^^^^

I still wouldn't do it unless it was something where fidelity didn't count, as in a modulator driver or something like that. It's easy to avoid, only requires a few additional parts that don't cost much, so why not avoid the problem completely?

 

[Bandersnatch]

hey-Hey!!!,
That front end is built that way so as not to look complicated. There are far better ways to amplify and split phase than that circuit. My friend has built a pair and after a bit of experimenting with various front ends, he shelved them for PP 807's with an E-Linear front end and less expensive custom output Iron. I think he's wanting to sell the Seth's now.
cheers,
Douglas

 

[arnoldc]

Hi Miles,

I read with great interest your comment about the bias circuit. My 2A3 SET amp (not PP) has had that bias circuit since I "stole" its TDO hum balance for another project.

1. Does your statement apply to all topologies? (SE/PP)
2. If so, then I have not been getting the potential of my amp?
3. Where do I stick a scope probe to see the differences?

 

[Bandersnatch]

hey-Hey!!!,
The PP 2A3 can be a very nice amp. Keep it class A, and perhaps run a 6k6 a-a output TX. As designed AB1, its plate load dropps to ~1k2...and that's not any fun. For a few more watts, build a different amp; leave the 2A3 to deliver a nice 6W from a conservative operating point.

Also, a single split bobbin filament TX for each of the finals will help. Connect a 50R pot in the usual fashion and one can get hum levels near zero...
cheers,
Douglas

 

[theskintman]

Wow - thanks for the excellent feedback. I envy you guys the ability to look at a circuit and judge improvements - I'm new to this, and whilst learning all the time, do not yet have the experience to judge a circuit from reference alone.

I guess I'm back to the drawing board, then. I'm a rock fan, so I thought PP would be more appropriate. Before I decided on PP and found the SETH, I was looking at tubelab's SimpleSE, but thought I'd quite like to try P2P as all my previous projects have been PCB. I currently have a P2P wired chinese EL34 PP stereo integrated amp, and would like to replace it with something I can call my own that will hopefully outperform it.

Oh, and Miles - an excellent modification to the schematic NO, seth NO!

 

[arend-jan]

^^^^

I still wouldn't do it unless it was something where fidelity didn't count, as in a modulator driver or something like that.

If you agree with me that the audio currents are not going through that transformer, then I don't see how fidelity is compromised. Anyway, it should be easy enough to try so I'll give it a spin in my amp.

regards,
Arend-Jan

 

[Shoog]

the performance is going to be very dependent on the quality of the iron used here, especially that of the splitter autotransformer

Quality is obviously going to be a factor here, but if you stick with the magnaquests this shouldn't be an issue. I have heard people rave about this particular splitter arrangement. The impression I got was it sounds wonderful - but measures less so.
I have used interstage transformer phase splitting and it works well. Again on paper it shouldn't work - but it does.

I would also ditch L1 and replace it with a grid resistor.

Other than cost - why ? There are good theoretical reasons why this should work better.

Shoog

 

[rdf]

.... stick with the magnaquests this shouldn't be an issue. I have heard people rave about this particular splitter arrangement. The impression I got was it sounds wonderful - but measures less so.
Shoog

I brought this up in one forum or another, discussing bench test results of common split chokes or unloaded transformer windings used as splitters. The answer came back that Magnequest and such optimize the windings of their splitter autotransformers specifically for this application, compensating for the fact that one winding is effectively unloaded and the other terminated by the driver output impedance. The balance probably still goes awry at ultrasonics, however no common music source (ie. CD) will excite that error.

 

[kevinkr]

The worst flaw is the 2A3 bias circuit. Filament xfmrs aren't designed to pass audio frequencies, and so I'd replace this arrangement with filament balancing resistors and connect the bias RC network at the center point. (See attached)

I have tried both approaches and have not been able to measure any discernible difference in distortion, noise, frequency response or sound quality in a number of amplifiers where I have center tapped filament transformers. The secondary winding inductance is generally quite small and since the audio appears common mode across the winding most of what there is cancels out leaving just the small dcr of the windings to contend with. I prefer omitting unnecessary components where possible, and have done this in a number of 45 SE, 2A3 SE and PP amplifiers with no issues so far.

I have not heard this specific amplifier, but have repaired and modified similar amplifiers using the MQ phase splitting choke, and without exception these amplifiers sounded pretty good.

 

[Miles Prower]

If you agree with me that the audio currents are not going through that transformer, then I don't see how fidelity is compromised. Anyway, it should be easy enough to try so I'll give it a spin in my amp.

I didn't say that. There is no such thing as a perfectly balanced OPT. The electrical neutral is never the half-way point since the coefficient of coupling between windings is never 100%. Nor is it likely that each 2A3 used there will be perfectly matched. There will be AF currents flowing in the heater xfmr. If there weren't, then why bypass the cathode resistor in the first place? That bias method is a quick and dirty that I'd use only for modulators, PAs, or anywhere else fidelity wasn't a high priority.

I don't expect anyone else to agree, and I don't particularly care. It's just not how I design.

My perference would be for fixed bias, with a bias control for each final, so that the currents can be brought into better balance.

 

[Gluca]

I built it ... something pretty similar to seth with the same MQ splitter and a LL1660. I preferred the MQ's iron (wound on a nickel core).

My suggestion is to replace the driver stage with a 6H30 SRPP (or a CCS loaded 6H30). You will not be disappointed.

Gianluca

 

[Shoog]

The use of a grid choke and the split choke suggest that this amp should behave very well under overdrive conditions - and so should produce more watts than you might anticipate.

Shoog

 

[kevinkr]

<snip>
There will be AF currents flowing in the heater xfmr. If there weren't, then why bypass the cathode resistor in the first place? That bias method is a quick and dirty that I'd use only for modulators, PAs, or anywhere else fidelity wasn't a high priority.
<snip>
My perference would be for fixed bias, with a bias control for each final, so that the currents can be brought into better balance.

I never said there weren't audio currents flowing through the filament transformer, just that the effective winding inductance was close to nil because those audio currents are common mode in the winding. Obviously another (unrelated) issue would be the coupling of noise from the primary into the filament supply - an issue addressed only through the inclusion of common mode chokes, but in practice the leakage inductance in EI types seems adequate in my environment at least to prevent problems - such may not be the case with toroids or in noisier electrical environments.

Also as I indicated I have not been able to measure any difference in performance using either technique. Incidentally on dc I always use the technique you mention, on 2.5Vac I use either depending on whether or not the filament transformer has a tap, and how good the actual winding symmetry is.. (Hum)

I totally agree with you on the fixed bias issue, virtually all of my amplifiers use it, imo nothing worse than a lousy cathode bypass capacitor...