No doubt's they are realMaybe Papa has been tweaking us all along and the XA25 is not really a product.
------------------------------
Thank you Mr Colburn.
I feel there is a real emphasis (and interest) on bias stability here.
It is not difficult to achieve very good stability with a thermally-coupled and "tweaked" Vgs-Multiplier.
Knowing now that the output stage is a complementary source follower, the most
straight forward design would be the Vgs-Multiplier.
However, this would lead to the inevitable headroom loss (from 3 to 10 V total - depending on devices used), due to
the Vgs threshold voltage requirements of the output FETs.
It might be possible Pass Labs has implemented a Bias-Voltage-Source that - at least partially - minimizes this loss.
The XA-25 is a very new product so the jury is still out on this one.
I am sure there will be more info on this later on.
----
PS:
These are VERY SIMPLE linear circuits. There is nothing magical about them.
Last edited:
----------------------------aha
that's why is xxx Labs production in every village ..... everyone is able to produce great simple linear circuit
we all know that the whole is never more than the sum of its parts
I am not sure I understand what you mean, but I am referring to the Vgs-Multiplier.
Cheers...
Last edited:
-------------------------
Interesting.
There are of course many ways to compensate for the bias drift of the
complementary source follower stage.
The multiplier can be tweaked to perform very well indeed, but from your comments I take
it that this is not the implemented solution in XA25.
Being a little familiar with Passlabs designs, the solution might be very simple and elegant.
It is mentioned in the XA literature that the amplifier is DC coupled.
Hopefully that means the output-stage is directly coupled to the VAS-stage. Not like the
examples in the F4 Beast thread.
Cheers.
Last edited:
historically , PL had Vgs multiplier as bias generator in their earlier products
later , it's replaced with TL431 based circuit , temperature compensated
there must be a reason for that ...... besides TempCo - I believe - lesser current involved , thus biasing net having greater impedances , so less garbage from rail reaching middle of the amp
later , it's replaced with TL431 based circuit , temperature compensated
there must be a reason for that ...... besides TempCo - I believe - lesser current involved , thus biasing net having greater impedances , so less garbage from rail reaching middle of the amp
We have done that for years - we don't bias up the Vgs generator with
the front end bias current like the competition, so we don't lose the 5 volts
or so of Vgs on the output.
Thank you Zen Mod and Mr. Pass for your help.
Ok, so the bias generators are floating with respect to the Front End AC output.
In this fashion. it is indeed possible to get amplitude of the output signal to swing very
close to the PSU supplies with very little headroom loss.
Is the Front End AC- or DC-coupled to the Followers?
Ok, so the bias generators are floating with respect to the Front End AC output.
In this fashion. it is indeed possible to get amplitude of the output signal to swing very
close to the PSU supplies with very little headroom loss.
Is the Front End AC- or DC-coupled to the Followers?
OK, Interesting info from Mr Pass and Zen Mod.
Please look at the schematics below:
- The TL431 is used as a differential variable Voltage Reference, biased by resistors.
- The two resistors connected in parallel with the signal capacitors (unavoidably electrolytic) make sure Front End is DC coupled to the Output Stage.
- This is the most simplified arrangement I can think of that probably works well.
- In this fashion, the headroom loss, due to the combined Vgs-threshold voltages of the output complementary-pair transistors are eliminated.
What do you all think??
Cheers
Please look at the schematics below:
- The TL431 is used as a differential variable Voltage Reference, biased by resistors.
- The two resistors connected in parallel with the signal capacitors (unavoidably electrolytic) make sure Front End is DC coupled to the Output Stage.
- This is the most simplified arrangement I can think of that probably works well.
- In this fashion, the headroom loss, due to the combined Vgs-threshold voltages of the output complementary-pair transistors are eliminated.
What do you all think??
Cheers
Last edited:
understanding of all published info is that there is no source degeneration anywhere
so - toss R25 and R26
also - with that simple biasing mechanismus , prepare plenty of heatsinking and expect pretty big difference between cold and temp equilibrium
I believe that in other thread you'll find few interesting implementations of small sized NTC around TL431
though , I'm almost 100% positive that there is no single TL431 in XA25
so - toss R25 and R26
also - with that simple biasing mechanismus , prepare plenty of heatsinking and expect pretty big difference between cold and temp equilibrium
I believe that in other thread you'll find few interesting implementations of small sized NTC around TL431
though , I'm almost 100% positive that there is no single TL431 in XA25
Well, I have thought about it, although there is no engineering justification for it - specially in reference to the XA25 design, but I guess it would be possible.
Yes, this is just a "conceptual" schematics. The details might not be there.
If more than one Voltage References are used, then it would be possible to adjust and tweak the n- and p-channel bias voltages independently.
Last edited:
On the contrary... I have looked at this for sometime and in some detail. Degeneration in a gnfb amp is sometimes a necessary evil, but it's evil all the same. If you can avoid it, so much the better.
-----------------------------------
I hear you, but please elaborate.
Otherwise I stand firm on the fact that there are no engineering justifications for it, as there are
no advantages in a complementary VAS stage design, particularly in a commercial amplifier, as in Pass XA25.
Unless someone from Pass Labs comments otherwise, I am willing to bet that degeneration resistors ARE USED in
the XA25 VAS stage (2nd Voltage Gain stage with the MosFets).
I believe by stating in the XA25 literature that degeneration is not used, Pass Labs is referring to the fact that the OUTPUT pair Q's are not degenerated.
That is of course a possible design decision since there are two important advantages of not using degeneration here.
So my guess is that in the XA25:
- the JFETs (input pair): are NOT individually degenerated
- the MosFet pair (VAS): are indeed degenerated
- the Ouput MosFets (follower stage): are NOT degenerated.
Last edited:
You can read about it here when I last played with this 4 years ago.
http://www.diyaudio.com/forums/pass-labs/132491-distortion-negative-feedback-8.html
I don't think it's that big a deal esp if you use relatively little degeneration, but the 'engineering" effect is there for sure, it's not voodoo
And other than output fets, smaller ones don't tend to run away (thermally) and explode when you don't use a source r.
Anyway like I said, it's not that big a deal and sometimes I have to use them to get to a convenient Dc operating point. But I always try and do without if I can.
http://www.diyaudio.com/forums/pass-labs/132491-distortion-negative-feedback-8.html
I don't think it's that big a deal esp if you use relatively little degeneration, but the 'engineering" effect is there for sure, it's not voodoo
And other than output fets, smaller ones don't tend to run away (thermally) and explode when you don't use a source r.
Anyway like I said, it's not that big a deal and sometimes I have to use them to get to a convenient Dc operating point. But I always try and do without if I can.
The justification for going to the trouble to avoid degeneration is that
it sounds better. In the case of the output stage, it also increases the
Class A envelope for a given bias.
Actually it is not difficult to work the VAS without degeneration if you want
to use the right parts, heat sink them well, and adjust them carefully. It
takes more time and money, things that most manufacturers avoid.
Keep in mind that if the output stage bias networks are driven by VAS
current, variations can create issues with the output stage bias.
it sounds better. In the case of the output stage, it also increases the
Class A envelope for a given bias.
Actually it is not difficult to work the VAS without degeneration if you want
to use the right parts, heat sink them well, and adjust them carefully. It
takes more time and money, things that most manufacturers avoid.
Keep in mind that if the output stage bias networks are driven by VAS
current, variations can create issues with the output stage bias.