Far as the OP project and what most would like to see
likely this. Agree which is what I usually used when first
playing around and doing AC and DC analysis of quasi
output stages.
far as the cap compensation, also have used it before.
This case adding the minimal degen in the diff stage
removes all ringing. With a cap on the output stage
it reduces the ringing at 22 K square wave.
With diff degen goes away completely at 22k no cap needed
often will cause instability. this case very
minimal R23 /24 100 ohms worked fine for degen.
And again removed all ringing on negative rail.
when cap compensation usually only reduces ringing.
and can slew on time at high frequency
likely this. Agree which is what I usually used when first
playing around and doing AC and DC analysis of quasi
output stages.
far as the cap compensation, also have used it before.
This case adding the minimal degen in the diff stage
removes all ringing. With a cap on the output stage
it reduces the ringing at 22 K square wave.
With diff degen goes away completely at 22k no cap needed
often will cause instability. this case very
minimal R23 /24 100 ohms worked fine for degen.
And again removed all ringing on negative rail.
when cap compensation usually only reduces ringing.
and can slew on time at high frequency
Last edited:
Hi, do you have the complete article? I mean Feb & Nov(1973) issue of hi-fi news magazine? Found nothin in Google.
I guess because it's a very simple way to make a current source, one JFET and a resistor (trimming potmeter actually) do the job. A current source with bipolar transistors usually requires at least four components. But that's just a guess, he may have had very different reasons (and as he's dead, unless he wrote about it in his article, we will never know).
Some comments on JLH's Hifi news circuit - I believe JLH liked to design with JFETs, even if it meant having a potentiometer to set up compared to a bipolar design which would not.
In some respects it is a fore-runner of the Blameless architecture, but has a few issues.
First, the input pair currents are unbalanced. That is not useful in a differential circuit. You can see the difference from the 330k base resistor on the input side and only 22k on the other.
Second, a 220pF capacitor effectively across the feedback resistor gives an upper bandwidth of 34kHz. Not really hi-fi.
Simulations show something in the order of 0.25% THD at 20kHz (though whether this is important is another question as we cannot hear distoriton components. What I would be concerned with is whether intermodulation products from the distortion components could be.
The steps towards making this a blameless circuit would be to add emitter resistors in the input pair; a current mirror instead of just a 4.7k resistor, and a dominant Miller capacitor rather than a collection of capacitors and especially eliminating the feedback RC pair across the feedback grounding resistor.
I think JLH agreed there were problems with the original design later on in his Wireless World review of amplifiers some time later, but don't recall if he described cures.
And a current source in the VAS stage.
In some respects it is a fore-runner of the Blameless architecture, but has a few issues.
First, the input pair currents are unbalanced. That is not useful in a differential circuit. You can see the difference from the 330k base resistor on the input side and only 22k on the other.
Second, a 220pF capacitor effectively across the feedback resistor gives an upper bandwidth of 34kHz. Not really hi-fi.
Simulations show something in the order of 0.25% THD at 20kHz (though whether this is important is another question as we cannot hear distoriton components. What I would be concerned with is whether intermodulation products from the distortion components could be.
The steps towards making this a blameless circuit would be to add emitter resistors in the input pair; a current mirror instead of just a 4.7k resistor, and a dominant Miller capacitor rather than a collection of capacitors and especially eliminating the feedback RC pair across the feedback grounding resistor.
I think JLH agreed there were problems with the original design later on in his Wireless World review of amplifiers some time later, but don't recall if he described cures.
And a current source in the VAS stage.
These quasi amps are 45 years old when they made perfect sense, but today I'd go with a fully complementary amp, unless you have some attachment to quasi amps. Power PNP transistors are now available and cheap. You should remember that a "Blameless" amp is the refinement of these old amps, and Douglas Self explains the refinements in detail on his web site.
The Baxandall diode and the resistor in the collector of Q2 are "misguided" ideas from a time when unproven theories were common, and unchallenged by simulation and measurement results. The lack of LTP degeneration puts it before the understanding of TIM / slew distortion. Today we have CFAs and MOSFETs but that doesn't mean amps that use them are without their own mistakes and bad ideas. I do think that the lack of any output protection is a booby trap waiting to blow up with the first whoops speaker short.
The Baxandall diode and the resistor in the collector of Q2 are "misguided" ideas from a time when unproven theories were common, and unchallenged by simulation and measurement results. The lack of LTP degeneration puts it before the understanding of TIM / slew distortion. Today we have CFAs and MOSFETs but that doesn't mean amps that use them are without their own mistakes and bad ideas. I do think that the lack of any output protection is a booby trap waiting to blow up with the first whoops speaker short.
That's an insult to Peter Baxandall. He was quite capable of measuring things, as well as calculating them.
Baxandall diode "misguided"?
Self confirmed that such a diode did reduce distortion in the quasi-amp. Not perhaps as much as a fully complementary design, but an improvement over the plain quasi.
Also take a look at the emitter currents in Linsley-Hood's 75W. Despite having no degen resistors in the input of the diff amp, and the imbalanced I mentioned, it does not cut off for a fast square wave input. It does show ringing, indicating a marginal instability but that does not reach the output.
The reason for this apparent freedom from transient distortion is due to the phase lead compensation capacitor as opposed to the usual Miller.
So I think you really aren't giving John Linsley Hood much credit either.
That does not mean to say that there has not been a lot learned since the 1970's -obviously, things are much better understood now. While JLH got some things wrong, as he said himself, his aims were at least in the right direction. Baxandall well understood the design aspects well. I beleive it was he who suggested the transitional Miller approach to Self.
Self confirmed that such a diode did reduce distortion in the quasi-amp. Not perhaps as much as a fully complementary design, but an improvement over the plain quasi.
Also take a look at the emitter currents in Linsley-Hood's 75W. Despite having no degen resistors in the input of the diff amp, and the imbalanced I mentioned, it does not cut off for a fast square wave input. It does show ringing, indicating a marginal instability but that does not reach the output.
The reason for this apparent freedom from transient distortion is due to the phase lead compensation capacitor as opposed to the usual Miller.
So I think you really aren't giving John Linsley Hood much credit either.
That does not mean to say that there has not been a lot learned since the 1970's -obviously, things are much better understood now. While JLH got some things wrong, as he said himself, his aims were at least in the right direction. Baxandall well understood the design aspects well. I beleive it was he who suggested the transitional Miller approach to Self.
No, it's not about any special attachment but it is because i have a bunch of good NPN output transistors including TIP35C & TTC5200. As output TR they have similar characteristics so i believe taking advantage of QC i can make some decent amplifiers for my friends & well wishers(or for myself). Currently i'm listening to Natsemi LM3886, which as per my taste is quite good. The problem is today complimentary amps are more popular & i didn't find much about QC amps in web except old capacitor coupled type which i don't like normally.
Regards
Ironically a LM3886 is a quasi amplifier.
As far as the process of putting power resistors on a wafer or integrated chip
It is incredible more simple to put 2x NPN power transistors on a chip.
Than trying to match a PNP and NPN power transistor.
Far as high power " chip amps " used for error compensation
for targeting radar.
There is numerous 40 to 150 watt chip amps my father tested
for National Semi for military radar.
By mid 90's target accuracy was 98%
assuming explosion radius of typical sidewinder missle.
the 2% error can largely be ignored.
And the 98% was done with Quasi Amplifiers
Anyone that has seen a Quasi on a bench with and without
the diode. Can easily tell you the non linearity common with
a quasi are easily fixed with the diode. Mainly AC current
not DC but AC current is equalized between the NPN's
As far as the process of putting power resistors on a wafer or integrated chip
It is incredible more simple to put 2x NPN power transistors on a chip.
Than trying to match a PNP and NPN power transistor.
Far as high power " chip amps " used for error compensation
for targeting radar.
There is numerous 40 to 150 watt chip amps my father tested
for National Semi for military radar.
By mid 90's target accuracy was 98%
assuming explosion radius of typical sidewinder missle.
the 2% error can largely be ignored.
And the 98% was done with Quasi Amplifiers
Anyone that has seen a Quasi on a bench with and without
the diode. Can easily tell you the non linearity common with
a quasi are easily fixed with the diode. Mainly AC current
not DC but AC current is equalized between the NPN's
Ironically again Germanium power transistor more commonly PNP
the original RCA patent was PNP Quasi
But yes dominant Bipolar power transistor NPN
Just for fun I was playing with PNP Quasi
for the historical endeavor.
Being fascinated by T03 metal package power transistors.
Current production T03 not much over 4 MHz Ft
Finding it comical one of the faster T03 available at 18 MHz
is a PNP. I played around with it since .01 to .008% distortion
is feasible at 20K with a Ft of 18 MHz
the original RCA patent was PNP Quasi
But yes dominant Bipolar power transistor NPN
Just for fun I was playing with PNP Quasi
for the historical endeavor.
Being fascinated by T03 metal package power transistors.
Current production T03 not much over 4 MHz Ft
Finding it comical one of the faster T03 available at 18 MHz
is a PNP. I played around with it since .01 to .008% distortion
is feasible at 20K with a Ft of 18 MHz
Yes, a well done discrete quasi can provide excellent performance just like Natsemi chipamps. Anyway till date the best quasi complimentary (using bjt's) amp i found in web is this.. a double differential QC from Elektor magazine.
Last edited:
Feasible , double differential can offer lower THD
Then again with the usual upgrades or improvements to the topology discussed
in this thread. It can too get similar results. Aka often called blameless
and then additional cascode Vas to stretch it even further.
Far as earlier opamps making a jump with lower distortion.
And its affiliation with the Hitachi mosfet amplifiers.
The topology I showed you in post #17 is the low part
count way of getting THD close to what a Double Diff can do.
I have never done thermal evaluation of the output topology
shown in the Elector amp. much like a triple but involving CFP
stages. I just stayed away from it. Mainly my ignorance and not
wanting to deal with it. But I can make up excuses.
Interested in double diff amps that is a fun rabbit hole.
But likely use complementary. Since CFP/ Quasi will put
a limit on high frequency THD and still be stable.
Then again with the usual upgrades or improvements to the topology discussed
in this thread. It can too get similar results. Aka often called blameless
and then additional cascode Vas to stretch it even further.
Far as earlier opamps making a jump with lower distortion.
And its affiliation with the Hitachi mosfet amplifiers.
The topology I showed you in post #17 is the low part
count way of getting THD close to what a Double Diff can do.
I have never done thermal evaluation of the output topology
shown in the Elector amp. much like a triple but involving CFP
stages. I just stayed away from it. Mainly my ignorance and not
wanting to deal with it. But I can make up excuses.
Interested in double diff amps that is a fun rabbit hole.
But likely use complementary. Since CFP/ Quasi will put
a limit on high frequency THD and still be stable.