So then does the distortion continue its downward trend with reduced resistance or is there a sweetspot?
I only have the fixed load. I would expect there is a range of values that would reduce 2nd harmonic. Tool low and then distortion would increase again - probably the 3rds start to creep up.
you have been pointed to better arguments before - this is unworthy polemic on your part
loading the VAS with a shunt R always causes more distortion - especially IMD difference products that fold down to frequencies where the VAS would have given much higher loop (Voltage) gain without the shunt
I have pointed out that the shunt does reduce VAS output Z and as such may not give too bad a result if the major distortion mechanism is the nonlinear input current from the driver/output stage
at best you may nearly break even from the shunt load reducing gain and providing proportionately reduced low Z for the next stage
however the diff pair and the VAS are operating over a larger range just to supply the unused current in the shunt and therefore shunt load does increase distortions - again not just THD but IMD products too
a good tool for exploring the loop gain, driving point impedance relations, tradeoffs is Cherry's “Estimates of Nonlinear Distortion in Feedback Amplifiers” JAES V48#4 2000
which tests the impedance matrix/sensitivity modeling combined with perturbation theory style nonlinear modeling against a physical amp
Last edited:
Jcx,
Don't be ridiculous.
I measured the damn amplifier on the bench.
Clearly what is happening is that the non- matching gain between the halves dominate the distortion measurement at the levels I did the test at. Loading the VAS as I did improves distortion probably for the reason I quoted. However, no doubt at higher lading the mechanisms you point out dominate.
I repeat: loading the VAS on the e-Amp with 2 x 33 k to ground lowers the distortion (1 KHz at 150 W into 8 Ohms). I did not originally put this option in the design to lower the distortion - I originally did it to explore what happens when the loop gain is made lower but the -3 dB bandwidth pushed well above the audio band which is what this loading does.
Ovation e-Amp: A 180 Watt Class AB VFA Featuring Ultra Low Distortion
Don't be ridiculous.
I measured the damn amplifier on the bench.
Clearly what is happening is that the non- matching gain between the halves dominate the distortion measurement at the levels I did the test at. Loading the VAS as I did improves distortion probably for the reason I quoted. However, no doubt at higher lading the mechanisms you point out dominate.
I repeat: loading the VAS on the e-Amp with 2 x 33 k to ground lowers the distortion (1 KHz at 150 W into 8 Ohms). I did not originally put this option in the design to lower the distortion - I originally did it to explore what happens when the loop gain is made lower but the -3 dB bandwidth pushed well above the audio band which is what this loading does.
Ovation e-Amp: A 180 Watt Class AB VFA Featuring Ultra Low Distortion
Last edited:
'unworthy polemic' means what in this context? Do please clarify what your complaint is?
Your technical arguments seem sound here but focus only on distortion. As if getting the lowest THD (with a reference to IMD in passing) is the be-all and end-all in amplifier design. Is there any evidence (listening based or any other) that this is the case?
To support what Owdeo says, I have a stunning 24 bit Verve reissue of a Dizzy Gillespie recording entitled "Have trumpet will excite" This was recorded in NYC in 1958 and I am pretty certain it went to analog 8 track through a tube mixer. I dont wish to appear arrogant. What I do object to is those critics with tin ears who rhapsodise about the "stunning imaging" of a recording AFTER it has been mashed and remixed and passed through hundreds of opamps. Self designed mixing desks for a long time so had to use opamps as a component. For most pro audio they are altogether too common.
Thank you for your comment. But rich people here (in my country) like an amplifier which have harmonic profile like that and with much higher THD.
I want to design an amplifier to fit their taste, although I like an amplifier with much lower THD (accurate).
Despite their not-so-great musical attributes, the early direct-to-disc recordings were often quite stunning, compared to what we'd gotten accustomed from studios and their many-gain-stage consoles.
in other words within margin of error for any general principle
there can be interesting cancellations with 2nd order diff component of a 3rd order distortion mechanism summing with the 2nd order nonlinearity 2nd antiphase
this sort of harmonic+IMD cancellation pops up in articles every few decades, there doesn't seem to be a textbook treatment
I wouldn't recommend design an amp dependent on these mechanisms
I would much rather see the 1 kHz CCIF difference measurement if stuck with just one measurement - IMD sweeps can be more illuminating
I specifically mention IMD, difference products folding down where they can be more audible because masking curves are steeper from below at reasonable listening levels
and critical bands are denser at low/mid audio frequencies, increasing the chance that the difference product is in a relatively empty critical band making audibility of IMD difference products much more likely than harmonics and sum IMD products
Last edited:
My sound card can resolve down to about 14 ppm. I expect the real distortion of the e-Amp is somewhat lower than the 28 and 20 ppm I reported, but the absolute figures at this level are not really a concern.
The fact is, measurements show that distortion goes down with VAS loading in my specific implementation and there is a plausible explanation for it - no 'unknown mechanism' or '3rd audiophile force' at work.
Theory says distortion must go up if you load the VAS. Measurements show that is not necessarily so.
The fact is, measurements show that distortion goes down with VAS loading in my specific implementation and there is a plausible explanation for it - no 'unknown mechanism' or '3rd audiophile force' at work.
Theory says distortion must go up if you load the VAS. Measurements show that is not necessarily so.
ground loops, feedback take off point errors, probe loop area coupling to rail nonlinear currents...
I wouldn't take my mother's or my own measurement at a single frequency, level, loading harmonic distortion number at that tiny distortion level, tiny difference as "proof"
many more data points, level, frequency sweeps need to be matched against predictions, qualification of the circuit implementation, measurement technique
I wouldn't take my mother's or my own measurement at a single frequency, level, loading harmonic distortion number at that tiny distortion level, tiny difference as "proof"
many more data points, level, frequency sweeps need to be matched against predictions, qualification of the circuit implementation, measurement technique
Last edited:
I've put a link in the earlier post to the amplifier write up.
I would not claim the result applies to all implementations etc. In sims, you do not see this effect because the component matches are generally much better in the models. I could build another e-Amp and find, fortuitously, that the two halves matched nearly perfectly, and as a result, the 2 nds would already be as low as they could go. In this case, loading the TIS may cause 3rds to increase and distortion to go up.
I would not claim the result applies to all implementations etc. In sims, you do not see this effect because the component matches are generally much better in the models. I could build another e-Amp and find, fortuitously, that the two halves matched nearly perfectly, and as a result, the 2 nds would already be as low as they could go. In this case, loading the TIS may cause 3rds to increase and distortion to go up.
The cable debate is is dangerous territory, as long as it is quality copper I like it 🙂. There is a lot of electronics theory to both support and negate cable claims, but the only real rule of thumb when running the long cable from preamp to monoblock amps is that preamp output impedence is sufficiently matched to cable capacitance, and that the cable is shielded properly. It's amazing what a little rf injection can do to colour the sound with bad shielding or none at all 🙂.
Colin
A little to add to the mention of opamp sonic signatures, by golly they do exist say my ears. I started out working with these little buggers, and each has their own strength and weaknesses. In fact so much so I have been able to compile a reference list of opamps to tailor the "sound" of a circuit based on each sonic signature. I can even say I would never double up an Analog devices or Burr Brown in one circuit in many cases. Scott Wurcer would even like that I don't mind an ad797 as a flat gain mc gain stage but hate it in an riaa filter stage, it just sounds too solid state there, even with nulling.Then again it was designed as a mic preamp preamp right?.
Colin
Colin
I agree that many "Original" very good sounding recordings were made with valve/tube gear in the late 50's & 60's etc 🙂 Not just jazz, but pop etc !
I used to worry about having my Ed Teller autographed book (of which there are many) and a Robert Oppenheimer (they are quite rare) side by side on a bookshelf.
I don't remember what Scott said the particular anticipated market for the 797 was, but I don't think it was audio, even though the datasheet leads with such applications. And if one is trying to do a traditional single-stage preamp with RIAA compensation, it has a lot of current noise for MM cartridges. A wonderful part for low voltage noise and moderate/high gains, to be used with low impedance sources (like dynamic microphones and most MC cartridges).
A lot of people don't understand parallel (current) noise.
The NE5534 remains a formidable device for MM because of its low noise current. I looked carefully into this, and you really have to go for discrete JFET's to significantly better it from the noise perspective.
Perhaps, Owdeo, you could enlighten us as to what your playback chain consists of, what program material you consider listenable, whether you have any credentials as a pro audio or even consumer audio designer, etc etc. So far we just see you as a knocker of Bonsai who at least puts his designs online to invite comment. Do the same, Owdeo, and we will have some basis for intelligent comment rather than your simplistic and quite annoying mudslinging towards MF !!!!
Not the Royal "we" abraxalito. I am but a humble audio engineer and jazz trumpet player who prides himself as having a good jazz CD collection and a reasonable playback system. I am a committee member of the local Melbourne AES chapter (M16662). I have always had analog design as my principal passion, and count Bob Pease and Jim Williams among my heroes. I try to speak honestly in plain language and avoid meaningless subjectivist jargon. I hope I will always make a positive contribution to the diyaudio community. My construction trajectory includes all the Marshall Leach Amps, the Bob Cordell distortion analyser, and I currently own an amp by Hugh Dean who I count as a very close friend. I enjoy talking with you guys and I hope we all continue to learn from each other. My lab is well-equipped with HP and Tek gear incl a SG5010 and AA5001. My scope is a 2465A and I have a 7L5 spectrum analyser too.