Need advice on a small DSD DAC experimentation platform.

[YashN]

It is!

Yep! And I implemented only the Tube rectification option for now for the B+.

With a small DC offset (a sixteenth of B+).

How did you derive this (I am still learning the basics of tube amps)?

So it's triode and the cathode on ground, or dI'd you set it in speaker feedback mode?

Yes, I configured it in simple triode mode for now, with Electro-Harmonix KT-88s, Sovtek 5AR4 and 12AT7 (EH or Chinese generic).

The full implementation is quite versatile: you can also configure it for ultra-linear operation (not yet implemented), Solid-State rectification (not implemented), use Cathode Feedback (not implemented), and all these can be done with switches if necessary and for listening comparisons.

Pretty high gain input circuit, with that current source. Nice!

With the KT-88s, I also chose to put in big Edcors (XPWR series) and CXSE25-8-5 as OPT (these are big!).

Results: I get a very fast and seemingly rather extended and resolving bass, something which SET Tube amps have as weakness, but not this one in this configuration.

George Anderson from Tubelab designed it so that it can use affordable tubes, so although I wasn't planning on tube-tube rolling, I can do that if I want.

The pages on the Tubelab SSE have a lot of great info on how he designed it.

DSD256 on this SET Tube Amp is flabbergasting.

So now, for me, the next level is bypassing the DAC chip, tube buffering the decoded/filtered DSD straight into the SET Tube Amp.

 

[YashN]

I forgot to add another next step for me:

I want to build high efficiency speakers for the DIY Set Tube Amp. My current plan is to build this Open-Baffle, with a good wide-ranger for that SET magic, but also at least a Ribbon or AMT Tweeter.

 

[mterbekke]

I forgot to add another next step for me:

I want to build high efficiency speakers for the DIY Set Tube Amp. My current plan is to build this Open-Baffle, with a good wide-ranger for that SET magic, but also at least a Ribbon or AMT Tweeter.

Tough combination, because of the low power SET triode and the expected SPL of that open baffle speaker. It depends a lot on what lowest frequency you want to get out of that design of course. The only stunning good bass I heard coming from an open baffle design was the Jamo of about 10 years back. They didn't sell very well (I'm guessing that had more to do with the name than with the Sq, but that's another story). Anyway, that speaker could beat some jbl's in the articulation of bass& mid bass. Nit an easy feat. Having no enclosure really helped it( no boomy, swollen or coloured bass). The other good designs that had a well balanced bottom end are really wide and/or large. What I'm trying to say: without a big open baffle design it's really hard to get much out of it. Changing to UL for the SET is a great option if that's the case.
Do you have a big listening room?

Btw if you'll ever do start rolling tubes, you might want to try the KT120. Great tube and easy to get, not too expensive and long lasting. For 12at7, I always try to get a Philips NOS (JAN), but there are others that sound well also of course.

 

[mterbekke]

Quote:
Originally Posted by mterbekke View Post
With a small DC offset (a sixteenth of B+).
How did you derive this (I am still learning the basics of tube amp

Well, so much for pasting screen dumps here, I'll try and explain it by text then:
On your schematic, the filaments are the 3 pyramid like drawings, just below the 5ar4. Left of it is a filtering capacitor that filters the voltage that is created by the voltage divider R3 & R4. R4 is connected to the power supply, R3 to ground, filtered, by C3, for hum/noise and ultimately attached to one of the transformer windings and the filaments. Total dc voltage on it should be B+ × ( R3÷R total). Im not really sure if it's needed, but it's better safe than sorry: for many tubes it's deadly if the filament voltage to cathode voltage gets bigger than approx. 100 volts. This can be the case when using a shunt regulated push pull ( kind of what you input stage looks like), or e.g. when an output tube malfunctions (high current/shorts) and the cathode gets pulled to the anode. Anyway, it looks like a great amp because of versatility as well as robustness and simplicity. As always: devil is in the details in these low parts amps: changing one part can do miracles ( or make oneself want to run out the door;-), and having no overall feedback can produce a really free and transparent sound. Btw : one of the best(natural, not overly tight or heavy etc) bass I have ever heard was from a modified SET Tube. And I don't have one myself.

 

[mterbekke]

And you're right: putting a tube output stage after a dac can create a good sound stage with lots of colour (the right one!) etc.
Do you use screened rca cables?
One if the issues I have is, depending on the type of cables (nordost and its siblings etc) can pick up some really nasty hum and interference if the tube output stage is resistor based.

 

[YashN]

Tough combination, because of the low power SET triode and the expected SPL of that open baffle speaker. It depends a lot on what lowest frequency you want to get out of that design of course.

No worries, I'll make it work.

The only stunning good bass I heard coming from an open baffle design was the Jamo of about 10 years back. They didn't sell very well (I'm guessing that had more to do with the name than with the Sq, but that's another story). Anyway, that speaker could beat some jbl's in the articulation of bass& mid bass.

I think the commercial underperformance has more to do with people being used to buying little to big boxes rather than drivers mounted on simple baffles. Good to know it's a good design, that can inform my build.

Nit an easy feat. Having no enclosure really helped it( no boomy, swollen or coloured bass). The other good designs that had a well balanced bottom end are really wide and/or large. What I'm trying to say: without a big open baffle design it's really hard to get much out of it. Changing to UL for the SET is a great option if that's the case.
Do you have a big listening room?

Yes, I know about the 'need for size', but there are ways around that. I think some of the (what I call 'hybrid') designs, with a couple of drivers on Z-frames or with slot-loading can sure help. Thing is, with that kind of design, you start bracing the frames + baffle, then you start dealing with the resonances induced by the bracing elements, so you're then partially dealing with some of the issues of the usual boxed implementation (hence 'hybrid')

Yes, we have a large space here, so that's not an issue at all.

Btw if you'll ever do start rolling tubes, you might want to try the KT120. Great tube and easy to get, not too expensive and long lasting. For 12at7, I always try to get a Philips NOS (JAN), but there are others that sound well also of course.

It wasn't part of my plan, but the SSE can use EL34, KT-88, 6L6GC and 6V6GT. When I did the troubleshooting, I had diagnosed the KT-88s and they were fine, so I got a second 12AT7 just in case. So I had the opportunity of swapping the 12AT7EH with the Chinese generic 12AT7. The EH sounds more 'metallic' to me, so I am currently using the Chinese generic one, it's smoother.

If I do swap output tubes, I'll probably try the EL34.

 

[YashN]

Anyway, it looks like a great amp because of versatility as well as robustness and simplicity.

Indeed, loving it! Yesterday was my gf's birthday, so a couple of friends were here and they brought their vinyls and said that it was like 'being there'.

As always: devil is in the details in these low parts amps: changing one part can do miracles ( or make oneself want to run out the door;-),

The magic is in the details of these low part amps! Because of the circuit simplicity, we can reach the ideal amp more easily than with a string of opamps.

and having no overall feedback can produce a really free and transparent sound. Btw : one of the best(natural, not overly tight or heavy etc) bass I have ever heard was from a modified SET Tube. And I don't have one myself.

Yes, and mine is a No Negative Feedback example - it's awesome!

 

[YashN]

Left of it is a filtering capacitor that filters the voltage that is created by the voltage divider R3 & R4. R4 is connected to the power supply, R3 to ground, filtered, by C3, for hum/noise and ultimately attached to one of the transformer windings and the filaments. Total dc voltage on it should be B+ × ( R3÷R total). Im not really sure if it's needed, but it's better safe than sorry: for many tubes it's deadly if the filament voltage to cathode voltage gets bigger than approx. 100 volts. This can be the case when using a shunt regulated push pull ( kind of what you input stage looks like), or e.g. when an output tube malfunctions (high current/shorts) and the cathode gets pulled to the anode.

Thanks a lot for that, learning, learning...

 

[YashN]

And you're right: putting a tube output stage after a dac can create a good sound stage with lots of colour (the right one!) etc.

My goal is more the less colouring possible, certainly much less than a DAC Chip and a string of Opamps, fast attack transients.

Do you use screened rca cables?
One if the issues I have is, depending on the type of cables (nordost and its siblings etc) can pick up some really nasty hum and interference if the tube output stage is resistor based.

I use normal RCA cables and I am very, very disappointed with those: for me the RCA interconnect is a very flawed design, one of the issues being the 'return-as-shield'.

Since I have been building my own AC Filter box and have dedicated audio lines, this RCA/single-ended interconnect is a weakness in my system as it is currently, so I need to fix this with balanced at some point.

 

[mterbekke]

My goal is more the less colouring possible, certainly much less than a DAC Chip and a string of Opamps, fast attack transients.

There's "colour" as in "glowing, warmth, full body etc" as opposed to colouring "lean", "grey", "intermodulation/slew rate/harmonic distortion" etc. Sorry for the fact that English is not my native tongue.

I use normal RCA cables and I am very, very disappointed with those: for me the RCA interconnect is a very flawed design, one of the issues being the 'return-as-shield'.

There are many pseudo differential rca cables on the market. With those you can use the screen as a screen, connected to the receiving end, while using the twisted pair in it as a ground and signal pair. Works like a charm!

Since I have been building my own AC Filter box and have dedicated audio lines, this RCA/single-ended interconnect is a weakness in my system as it is currently, so I need to fix this with balanced at some point.

Good thing is that a dsd dac is easily built balanced: just use the normal dsd data signal and feed it into a differential output flip flop. Also connect the bitclock and you're done

 

[YashN]

There are many pseudo differential rca cables on the market. With those you can use the screen as a screen, connected to the receiving end, while using the twisted pair in it as a ground and signal pair. Works like a charm!

Correct, and I get great results with double-shielded twisted pairs wherever I use them, so one of my intermediate plans it to build my own psuedo-balanced RCA cables.

Ultimately, I am looking for a fully balanced system. That will take some time.

Good thing is that a dsd dac is easily built balanced: just use the normal dsd data signal and feed it into a differential output flip flop. Also connect the bitclock and you're done

We also need the SET amp to have a differential input interface...

I know Neurochrome makes some excellent THAT-based interfaces so I may use those when I get to it.