First attempts at some DOA designing

[etheory]

I am generally not enthusiastic about the project with or without trimpots

What would you be enthusiastic about?

I'm sure I am more of an amusement to the more experienced members on the board, and that's fine, I'm totally cool with that.  I am young, inexperienced, but also extremely capable.

There's a lot to learn, and there are many different places I could start, but I've chosen to start here, for better or for worse.

I'll be implementing some modifications based on your feedback over the weekend.  Thanks for taking the time to respond with your suggestions it's greatly appreciated.

As for testing for component tolerances, it's rather easy.  I'm a computer programmer by trade, so that particular testing regime consists of LTSpice user scripts that modify the parameters of the parts stochastically, based on any tolerance information I can gather from the manufacturer, to create a stochastic distribution of simulations that I can evaluate for the ranges of currents and voltage modifications that they produce.  This makes it relatively simple (though for this design this testing is as yet undone) to evaluate the likely performance over a user-definable range of random variations and components.

 

[ricardo]

What would you be enthusiastic about?

Luke, some of us old fogeys predate the 741 OPA when nearly EVERYTHING was done DOAs so we've done ALOT of DOA design.  Old Neves & Calrecs may exemplify the state of art in DOAs.  But when the 5532/4 came out, most of us lost interest.

Some of us went through an intermediate stage when we tried to couple a BJT, FET or 2 with a 741 to incorporate the best of all worlds but the 5532/4 put paid to all of that except for VERY specialised applications; eg a transformerless mike preamps with less than 1.5dB NF.

Today, I find it hard to respond to anything which is easily beaten on all counts by 1/2 a 5532 and a couple of resistors.  Even the new uber OPAs hardly give any real advantage.  I would only comment if I thought the Holy Smoke might escape.

The BIG question is WHAT are you trying to achieve?

1.) 24V rails - cause that's what I have

No problem. DONE

2.) Differential inputs for use in differential input stages and EQ's

ditto

3.) I need to get rid of a large number of BD140/BD139/BC550C/BC560C and 2SK170's, so it has to use those

easy.

4.) I want something warm sounding and thick - i.e. even with no transformers, I want lots of harmonic "thickness" - hopefully not distortion per-se, but some girth in the lower order harmonics - if I want clean I'll use a VST - I want something that thickens and makes it's presence known

This is the ONLY interesting requirement.  But if you want some euphonic ideal (EI), why are you trying to reduce distortion?  I'm with PRR in that you don't want this effect in ALL your EQs etc.  I assume, huge size, complexity and loadsa Watts dissipated are all irrelevant to you.

But if the EI is the only thing that matters, why not forget distortion etc and just dream up some way to listen to the differences in a consistent, repeatable and reliable manner.  In my previous life, I was a Double Blind Listening Test guru, ala Lipsh*tz et al so I can help.  But I warn you it is not easy or cheap to do this type of test properly.  Cheaper to buy the latest Audio Precision ...  but that doesn't tell you what really sounds best.

I'll now go back to my hole.

PS    Maybe check the Holy Smoke dun escape. 

PPS  If this is an exercise in 'DOA design 101', may I recommend you get D Self's book on Power Amps.  A DOA is just one of these but smaller.

PPPS  For a Class A DOA with a distinct but GOOD sound, consider John Linsley-Hood's little Class A amp.  Its not differential but this is part of it's harmonic structure.  Most other 'Class A' amps (including Self's) are just Class AB with the bias turned up.  Contrary to popular myth, this is NOT the same as real Class A.  That's not to say I consider Class A 'better'.

 

[JohnRoberts]

I am generally not enthusiastic about the project with or without trimpots

What would you be enthusiastic about?

You might be surprised. I joined this web forum because I was enthusiastic about a a project of Wayne's that I found here (DC coupled mic preamp). I had noodled with trying to find an elegant solution to that problem (evil DC blocking capacitors)  for decades.  Wayne has moved on since then, but like a bad penny I keep turning up here. 

I'm sure I am more of an amusement to the more experienced members on the board, and that's fine, I'm totally cool with that.  I am young, inexperienced, but also extremely capable.

I am not amused... I am just honestly unenthusiastic... I designed my share of DOA back in the '70s when I couldn't buy off the shelf  opamps that would do the j-o-b. Today off the shelf opamps are very very good.

There's a lot to learn, and there are many different places I could start, but I've chosen to start here, for better or for worse.

and I have tried to give you useful advice, more about how to keep the smoke inside the components since I suspect we may diverge on what sound characteristics to pursue.

I'll be implementing some modifications based on your feedback over the weekend.  Thanks for taking the time to respond with your suggestions it's greatly appreciated.

As for testing for component tolerances, it's rather easy.  I'm a computer programmer by trade, so that particular testing regime consists of LTSpice user scripts that modify the parameters of the parts stochastically, based on any tolerance information I can gather from the manufacturer, to create a stochastic distribution of simulations that I can evaluate for the ranges of currents and voltage modifications that they produce.  This makes it relatively simple (though for this design this testing is as yet undone) to evaluate the likely performance over a user-definable range of random variations and components.

Gee I never found it all that easy to use JFETs as "repeatable" current sources. In fact I did one design back in the early '80s where i wrapped a DC servo around the front end JFET to make it behave.

I have written a few lines of code (actually a lot), but don't find that any consolation when dealing with JFET pinch off variance, unless i add a microprocessor to the design.

JR

PS: I did not attempt to make light of your design efforts, and I'm sorry if others found humor in my words, perhaps it was my dry understatement. I repeat that the jfet current source feeding the LTP looks like a potential problem to me.

 

[steveh]

"how about a 100%  class A DOA?" -ruffrecords

"Most other 'Class A' amps (including Self's) are just Class AB with the bias turned up.  Contrary to popular myth, this is NOT the same as real Class A." -ricardo


Is the definition of Class A restricted to single-ended on this forum, at odds with general understanding?  Or... can we say that a single-ended amplifier is just a type of Class AB with very poor performance outside the Class A limit? 

 

[ricardo]

Is the definition of Class A restricted to single-ended on this forum, at odds with general understanding?  Or... can we say that a single-ended amplifier is just a type of Class AB with very poor performance outside the Class A limit? 

I wish I hadn't opened this can of worms  :

IM very HO, a single ended power amplifier is a very badly designed Class A amplifier ..  unless the main design constrain was a total cost not exceeding 10 cents.  Most 'Class A' amplifiers fall into the vbdcA category.  However JLH's is certainly not single ended.  I suggested it as a basis for 'valve' like sound.  In his original articles, JLH said it sounded closest to a Williamson amp and thought it was due to the small THD  being almost all 2nd.

I've just noticed Luke wants a transformer, not a valve or tube sound.  So belay my suggestion and let me crawl back into my hole.    Belay my nonsense about single ended too cos most tube stages are single ended anyway.

 

[Jean Clochet]

PPS  If this is an exercise in 'DOA design 101', may I recommend you get D Self's book on Power Amps.  A DOA is just one of these but smaller.

+1  Was about to post the same recommendation.  The title is:
Audio Power Amplifier Design Handbook by Douglas Self. 

Regarding a stage with colour (euphonics) that you can switch or blend into the signal path, maybe take a look at an AES paper by Dimitri Danyuk: "Triode Emulator" from 2004.
I've had decent luck getting close to, but not exact, triode-like curves out of a J-Fet by using single-ended stages with drain-to-gate feedback, much like  O.H. Schade of RCA documented for Pentode valves in 1938. 

P.S.  On a different note for clean stages: Theoretically, a "perfectly" matched dual differential-complimentary J-Fet stage can cancel most distortions.  In practice, you'll not achieve perfect enough matches. 

 

[steveh]

However JLH's is certainly not single ended.  I suggested it as a basis for 'valve' like sound.  In his original articles, JLH said it sounded closest to a Williamson amp and thought it was due to the small THD  being almost all 2nd.

If etheory wants to experiment along these lines, can I suggest making R22 & R23 (the output emitter resistors) unequal so that one transistor pushes more than the other pulls.  I expect this is the reason that JLH's class A amplifier distorts the way it does.  A quasi-complementary class A stage might also do the same, even with equal emitter resistors.