thermal etc
Hartmut,
Right, I wasn't aware of the strong dependance of GBP on Ic or Id. A quick data book check confirmed it. I sometimes see changes in an amps ringing behaviour depending on Iq. I attributed that to switching effects when going from B to AB. In hindsight, it may have been the effect you described. I learned something again. Thanks for your patience.
On the Naim, I know at least one model where they strongly advise their own speaker cables for reasons of stability. Apparently, the cable has just the right cap and inductance to keep the amp stable. They apparently don't use the output inductance and zobel. Together with your statement on the thermal (in)stability, they seem tricky amps to live with.
Cheers, Jan Didden
Hartmut,
Right, I wasn't aware of the strong dependance of GBP on Ic or Id. A quick data book check confirmed it. I sometimes see changes in an amps ringing behaviour depending on Iq. I attributed that to switching effects when going from B to AB. In hindsight, it may have been the effect you described. I learned something again. Thanks for your patience.
On the Naim, I know at least one model where they strongly advise their own speaker cables for reasons of stability. Apparently, the cable has just the right cap and inductance to keep the amp stable. They apparently don't use the output inductance and zobel. Together with your statement on the thermal (in)stability, they seem tricky amps to live with.
Cheers, Jan Didden
Re: thermal etc
Jan,
the Naim instability with capacitive loads comes from bad stability margin through obsolete design methods. Miller cap compensation is stone age. See my remarks and modification hints in the Naim schematics thread.
My modified Naim NAP140 is stable with all cables, and under all conditions.
Oh, and Hiraga Le Class A tend to blow after years of use - they are getting really hot. I had to repair 2 of them in the last years.
regards,
Hartmut
janneman said:
Jan,
the Naim instability with capacitive loads comes from bad stability margin through obsolete design methods. Miller cap compensation is stone age. See my remarks and modification hints in the Naim schematics thread.
My modified Naim NAP140 is stable with all cables, and under all conditions.
Oh, and Hiraga Le Class A tend to blow after years of use - they are getting really hot. I had to repair 2 of them in the last years.
regards,
Hartmut
"Pass amps are not really in the forefront regarding low distortion, so thermal distortion may be 3rd or 4th order effects in them, but in very low distortion/high precision amps thermal dist may have a much more pronounced effect."
The character of Pass amps is a function of simplicity and
low feedback. If you make the circuit more complex and
apply lots more feedback, the thermal distortion is still going
to be in the dirt.
Just as I have yet to see a commercially sucessful (or otherwise
legendary) amplifier with ultra-low distortion, or ultra-high
slew rate, (etc. etc.), I am waiting to see a low thermal
distortion amp make its mark.
Actually, the early Stasis amps qualify in this regard, as the
output stage was designed for constant current / constant
voltage, and this kept the thermal character quite constant.
The front end was all cascoded everywhere, so the same
tended to apply.
The character of Pass amps is a function of simplicity and
low feedback. If you make the circuit more complex and
apply lots more feedback, the thermal distortion is still going
to be in the dirt.
Just as I have yet to see a commercially sucessful (or otherwise
legendary) amplifier with ultra-low distortion, or ultra-high
slew rate, (etc. etc.), I am waiting to see a low thermal
distortion amp make its mark.
Actually, the early Stasis amps qualify in this regard, as the
output stage was designed for constant current / constant
voltage, and this kept the thermal character quite constant.
The front end was all cascoded everywhere, so the same
tended to apply.
"Just as I have yet to see a commercially sucessful (or otherwise
legendary) amplifier with ultra-low distortion, or ultra-high
slew rate, (etc. etc.), I am waiting to see a low thermal
distortion amp make its mark."
Halcro seems to be making their mark in the ultra-low distortion arena. When I first saw their ads several years ago geared exclusively toward claiming the lowest distortion specs of any other amplifier, I found it rather ironic that they were playing that to an audience that still mocks the THD wars of the 60s and 70s.
And Lavardin amplifiers are all predicated on low thremal distortion. They don't actually call it thermal distortion however. Instead they call it "Memory Distortion," claim to have discovered it themselves, and herald it as "the greatest discovery in analog audio design over the last twenty years."
Don't know whether you'd consider the Halcro or Lavardin amps as "commercially successful" but if nothing else, their marketing hyperbole makes for some cheap entertainment.
se
legendary) amplifier with ultra-low distortion, or ultra-high
slew rate, (etc. etc.), I am waiting to see a low thermal
distortion amp make its mark."
Halcro seems to be making their mark in the ultra-low distortion arena. When I first saw their ads several years ago geared exclusively toward claiming the lowest distortion specs of any other amplifier, I found it rather ironic that they were playing that to an audience that still mocks the THD wars of the 60s and 70s.
And Lavardin amplifiers are all predicated on low thremal distortion. They don't actually call it thermal distortion however. Instead they call it "Memory Distortion," claim to have discovered it themselves, and herald it as "the greatest discovery in analog audio design over the last twenty years."
Don't know whether you'd consider the Halcro or Lavardin amps as "commercially successful" but if nothing else, their marketing hyperbole makes for some cheap entertainment.
se
Lavardin actually discovered thermal distortion
Steve,
they do not only claim it, they really discovered it. Mr Perrot, the brain behind Lavardin, showed in a series of technical articles that thermal distortion occurs and how to get rid of it in circuit technologies. Though he kept his final best circuit secret. He did it in the mid-80ies under the nickname "Hephaistos" in the DIY mag L'Audiophile. In the 90ies he then raised several patents on the final circuit technology. He did several papers on that subject for the AES, too.
What he showed in the 80ies, was: He made a test setup, where active devices are measured under conditions of square wave power dissipation variations. He showed the extent at which the operating points ripple shortly after rise or fall. Interesting was that only tubes and the 2SK30 JFET showed no ripple at all.
He did also a whole range on amplifier stages, inputs, outputs, showing different topogies and measured them a lot.
In a very late L'Audiophile issue they compared his power amp to a 300B and claimed that it was better.
To call it memory distortion/effect now is just marketing, in my opinion.
regards,
Hartmut
Steve Eddy said:
Steve,
they do not only claim it, they really discovered it. Mr Perrot, the brain behind Lavardin, showed in a series of technical articles that thermal distortion occurs and how to get rid of it in circuit technologies. Though he kept his final best circuit secret. He did it in the mid-80ies under the nickname "Hephaistos" in the DIY mag L'Audiophile. In the 90ies he then raised several patents on the final circuit technology. He did several papers on that subject for the AES, too.
What he showed in the 80ies, was: He made a test setup, where active devices are measured under conditions of square wave power dissipation variations. He showed the extent at which the operating points ripple shortly after rise or fall. Interesting was that only tubes and the 2SK30 JFET showed no ripple at all.
He did also a whole range on amplifier stages, inputs, outputs, showing different topogies and measured them a lot.
In a very late L'Audiophile issue they compared his power amp to a 300B and claimed that it was better.
To call it memory distortion/effect now is just marketing, in my opinion.
regards,
Hartmut
Hartmut,
One can consider that Mr Perrot has discovered the so named thermal distortion, but transient instability and linearity errors caused by thermal effects in transistors amplifiers (in medical equipment, for instance) were well known many years ago, in 1960 if I remember. The phenomenon was particularly noticeable after a very large overload, such caused by patient grounding fault in ECG : 10 to 20 seconds were necessary for the amplifier to recover its full precision... (0.5% or so). At present time, this effects are completely cancelled by sophisticated designs.
Regards, Pierre Lacombe.
One can consider that Mr Perrot has discovered the so named thermal distortion, but transient instability and linearity errors caused by thermal effects in transistors amplifiers (in medical equipment, for instance) were well known many years ago, in 1960 if I remember. The phenomenon was particularly noticeable after a very large overload, such caused by patient grounding fault in ECG : 10 to 20 seconds were necessary for the amplifier to recover its full precision... (0.5% or so). At present time, this effects are completely cancelled by sophisticated designs.
Regards, Pierre Lacombe.
"they do not only claim it, they really discovered it. Mr Perrot, the brain behind Lavardin, showed in a series of technical articles that thermal distortion occurs and how to get rid of it in circuit technologies. Though he kept his final best circuit secret. He did it in the mid-80ies under the nickname "Hephaistos" in the DIY mag L'Audiophile. In the 90ies he then raised several patents on the final circuit technology. He did several papers on that subject for the AES, too."
I'm sorry, no. They did not really discover it. Thermal distortion had been well known since the earliest days of solid state electronics.
Now, he may have been the first among the ranks of high end audio designers to have stumbled across it. But that's like saying Columbus discovered America, which would come as some surprise to those folks who were standing on the shore to greet him.
Unfortunately, the high end has something of a knack for reinventing things and claiming them as new discoveries.
In fact, there had already been a number of circuits developed to specifically compensate for such things as thermal distortion, which is all Perrot's circuit does.
Here's an excerpt from a patent issued to Tektronics which was filed for back in 1977, which I dare say is at least a few years prior to the mid-80s:
"Two major sources of amplifier error are non-linearity and thermal distortion. The mechanisms which produce these undesirable traits are inherent in the fundamental physical properties of semiconductor pn junctions."
I'm sorry, but Lavardin's claim of discovery is an empty one.
se
I'm sorry, no. They did not really discover it. Thermal distortion had been well known since the earliest days of solid state electronics.
Now, he may have been the first among the ranks of high end audio designers to have stumbled across it. But that's like saying Columbus discovered America, which would come as some surprise to those folks who were standing on the shore to greet him.
Unfortunately, the high end has something of a knack for reinventing things and claiming them as new discoveries.
In fact, there had already been a number of circuits developed to specifically compensate for such things as thermal distortion, which is all Perrot's circuit does.
Here's an excerpt from a patent issued to Tektronics which was filed for back in 1977, which I dare say is at least a few years prior to the mid-80s:
"Two major sources of amplifier error are non-linearity and thermal distortion. The mechanisms which produce these undesirable traits are inherent in the fundamental physical properties of semiconductor pn junctions."
I'm sorry, but Lavardin's claim of discovery is an empty one.
se
but the published THD+N is .05%. Plus a S/N of better than 120dB! Steve Eddy said:
Perrot's circuit does not compensate, it avoids or minimize it. Circuits which look like they are compensating, are different, have a look at e.g. the Van de Plassche (spelling ?).
Steve Eddy said:
as you seem to know a lot of it, could you quote the patent number or the link into a patent server, please.
regards,
Hartmut
<i><b>Perrot's circuit does not compensate, it avoids or minimize it. Circuits which look like they are compensating, are different, have a look at e.g. the Van de Plassche (spelling ?).</i></b>
Of course they're compensating. What do you understand the word to mean? According to my dictionary (Merriam-Webster's Ninth Collegiate):
<b>3 a :</b> to provide with means of counteracting variation <b>b :</b> to neutralize the effect of variation
What do you think all that additional circuitry is for? It's there to cause a constant current/voltage in the primary differential transistors and thereby counteract the variance that would otherwise take place and result in thermal distortion.
<i><b>as you seem to know a lot of it, could you quote the patent number or the link into a patent server, please.</i></b>
That quote was from patent number 4,146,844. It's just one of perhaps a dozen or so patents issued to Tektronix which relate to the issue of thermal distortion. I used that particular quote because it pointed to the fact that thermal distortion is implicit in the parameter changes with temperature in semiconductor devices and that these parameter changes with temperature were known quite a long time before Mr. Perrot rediscovered them.
Audio designers sometimes limit the scope of their studies to the realm audio and ignore research being done in other disciplines which involve analog circuit design. Instrumentation being one of the more obvious.
se
Of course they're compensating. What do you understand the word to mean? According to my dictionary (Merriam-Webster's Ninth Collegiate):
<b>3 a :</b> to provide with means of counteracting variation <b>b :</b> to neutralize the effect of variation
What do you think all that additional circuitry is for? It's there to cause a constant current/voltage in the primary differential transistors and thereby counteract the variance that would otherwise take place and result in thermal distortion.
<i><b>as you seem to know a lot of it, could you quote the patent number or the link into a patent server, please.</i></b>
That quote was from patent number 4,146,844. It's just one of perhaps a dozen or so patents issued to Tektronix which relate to the issue of thermal distortion. I used that particular quote because it pointed to the fact that thermal distortion is implicit in the parameter changes with temperature in semiconductor devices and that these parameter changes with temperature were known quite a long time before Mr. Perrot rediscovered them.
Audio designers sometimes limit the scope of their studies to the realm audio and ignore research being done in other disciplines which involve analog circuit design. Instrumentation being one of the more obvious.
se
Nelson Pass said:
Since we talk about thermal distortion: What is it really, in real life? Temperature doesn't change very quickly so does it produce effects that you actually can hear? Does this type of distortion change operation points of the circuit and therefore maybe set the amp in worse operating conditions? I asked Mr Walter Jung abouts this but I didn't get a good answer. I gather that this phenomena is with single chip solutions (opamps)?
P.Lacombe said:
Pierre,
the articles and researches from Mr Perrot dealt only with behaviour of active devices within their power envelope, that is normal operation. He did not looked at overload or saturation effects. IIRC, he loaded the devices with loads changing from something like 40 to something like 60% of power dissipation, while keeping well inside the current and voltage limits. This seems to be trivial, because going outside would easily damage the active device under test. This seems to be a big variation, but I must admit that I am testing preamp powersupplies with square waves of half max current, just to have something meaningful at all on the scope.
Also, he showed the ringing in the milli-seconds time frame, that is exactly the listening frequency range. Ringing or overshoots or settling behaviour in that time frame cannot be compensated by servos like those used for nulling offsets of DC coupled power amps.
regards,
Hartmut
peranders said:
I'm not sure what temperature you're referring to that doesn't change very quickly.
As mentioned previously, the phenomenon has to do with the physical nature of the pn junction itself which is rather exceedingly small and while ambient temperature doesn't change very quickly, the instantaneous temperature of the pn junction can change quite rapidly owing to its small physical geometry.
As for audibility, I've no idea.
Well, rather the opposite. It's basically the change in the operating points of the transistor which results in the distortion, not the other way around.
It can be worse, yes. Largely I think because of thermal "crosstalk" between the output stage and input stage owing to their being constructed on the same tiny silicon substrate.
Some 10 years ago or so I started experimenting with monolithic power opamps. When I compared the monolithics to hybrid designs (which are more of a discrete circuit built up on a much larger substrate) I found I preferred the hybrids to the monolithics. That preference may have had something to do with thermal issues.
se
Hartmut,
I was talking about trivial effects of overload because this was the first effect which has been noticed, and surprisingly without any capacitor in the circuit : clearly, it was thermal memory effects. But similar effects have been soon discovered in normal operation, causing transient gain error.
Gérard Perrot seems to be the first to have related this phenomena in audio amplifiers, but in my opinion has not discovered them.
Regards, Pierre Lacombe.
I was talking about trivial effects of overload because this was the first effect which has been noticed, and surprisingly without any capacitor in the circuit : clearly, it was thermal memory effects. But similar effects have been soon discovered in normal operation, causing transient gain error.
Gérard Perrot seems to be the first to have related this phenomena in audio amplifiers, but in my opinion has not discovered them.
Regards, Pierre Lacombe.
Audibility of Thermal Distortion
Pierre, Steve,
you teached that Mr Perrot did not discover that subject. Thank you for your informative input. Though I got the impression, that you don't want to give him any credits at all for having covered the subject and measured and listened input and output stages and writing articles nearly 100 pages altogether.
All,
yes, we can hear thermal distortion. My first DIY preamp had it. It had a phono stage with mirrored input LTPs, which are feeding their respective VAS LTPs then. Output was taken from the collectors of positive phase ends of said VAS LTPs.
When I fired it up and listened to it, it was like my turntable had wow. I then measured the operating points and saw, that the points of the VAS LTPs where unbalanced and even moving from second to second. What happened: the changing bias of the input LPTs was amplified and moved the bias of the VAS stages. There was no feedback to correct that.
I then removed one half of the mirrored schematic and loaded the remaining VAS LTP with a current mirror. Bingo. Good sound and stable operating points then.
So, at least when the net signal is very small (moving coil input), then thermal distortion can be heard, and it is beneficial to the sound, when that distortion is minimized.
regards,
Hartmut
Pierre, Steve,
you teached that Mr Perrot did not discover that subject. Thank you for your informative input. Though I got the impression, that you don't want to give him any credits at all for having covered the subject and measured and listened input and output stages and writing articles nearly 100 pages altogether.
All,
yes, we can hear thermal distortion. My first DIY preamp had it. It had a phono stage with mirrored input LTPs, which are feeding their respective VAS LTPs then. Output was taken from the collectors of positive phase ends of said VAS LTPs.
When I fired it up and listened to it, it was like my turntable had wow. I then measured the operating points and saw, that the points of the VAS LTPs where unbalanced and even moving from second to second. What happened: the changing bias of the input LPTs was amplified and moved the bias of the VAS stages. There was no feedback to correct that.
I then removed one half of the mirrored schematic and loaded the remaining VAS LTP with a current mirror. Bingo. Good sound and stable operating points then.
So, at least when the net signal is very small (moving coil input), then thermal distortion can be heard, and it is beneficial to the sound, when that distortion is minimized.
regards,
Hartmut
Thank you very much for nice link <http://peufeu.free.fr/audio/> about thermal distortion - quite authoritative info.
While searching the net I have found that Mr. Perrot has presented another paper and that he is published in some French Hi-Fi magazine under the pseudonym Hephaistos
Gerard Perrot
Hephaistos Laboratories, Tours, France
Quality Audio Circuits _ Limitations of the Usual Design Methods
presented at the Audio Engineering Society
108th Convention 2000 Feb.19-22 Paris
but this paper isn't listed on AES site. Have anybody read it? Have anybody heard about recent development of the subject?
Thank you
While searching the net I have found that Mr. Perrot has presented another paper and that he is published in some French Hi-Fi magazine under the pseudonym Hephaistos
Gerard Perrot
Hephaistos Laboratories, Tours, France
Quality Audio Circuits _ Limitations of the Usual Design Methods
presented at the Audio Engineering Society
108th Convention 2000 Feb.19-22 Paris
but this paper isn't listed on AES site. Have anybody read it? Have anybody heard about recent development of the subject?
Thank you
Sort of off the point, but thermal distortion or more appropriate settling time being effected by thermal effects, thermal tails and the such, are also a known and real problem in many data acquisition systems (i.e. video sampling) and are a very real effect in common off the shelf op-amps. Sadly enough, most vendors do not properly characterize it, but Comlinear op-amps were in the past characterized for this.
To HiFiZen, I am very interested in your results. Another poster I think accurately characterized what you are looking at as dynamic distortion characteristics versus static distortion characteristics which again unfortunately is about all that is typically characterized for an amplifier. I would love to hear more discussion about transient distortion measurement, but lets leave that for another thread.
To everyone else, thank you for all the information, this has been a good useful thread.
Alvaius
To HiFiZen, I am very interested in your results. Another poster I think accurately characterized what you are looking at as dynamic distortion characteristics versus static distortion characteristics which again unfortunately is about all that is typically characterized for an amplifier. I would love to hear more discussion about transient distortion measurement, but lets leave that for another thread.
To everyone else, thank you for all the information, this has been a good useful thread.
Alvaius
Charles,
what amplifier topologies are there where the first emitter junction is inside the feedback loop? The only one that comes to my mind is a non-differential input stage, where the output signal is fed back to the emitter. This will only work in inverting configuration, so the source impedance better be constant. The DC drift will be quite tricky, and this may be an indication that it'll also be sensitive to thermal distortion. After all, a DC servo does not work in the audio band.
Nelson & Steve,
the Audionet amps seem to have extremely low distortion, and they are getting extremely good reviews (whatever that tells us).
Hartmut,
what does LTP stand for?
Regards,
Eric
what amplifier topologies are there where the first emitter junction is inside the feedback loop? The only one that comes to my mind is a non-differential input stage, where the output signal is fed back to the emitter. This will only work in inverting configuration, so the source impedance better be constant. The DC drift will be quite tricky, and this may be an indication that it'll also be sensitive to thermal distortion. After all, a DC servo does not work in the audio band.
Nelson & Steve,
the Audionet amps seem to have extremely low distortion, and they are getting extremely good reviews (whatever that tells us).
Hartmut,
what does LTP stand for?
Regards,
Eric