|Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting|
Comments on Greiner
The article by R.A. Greiner in the November, 1977, issue of Audio, "Amp Design and Overload," is quite interesting. While the article makes a number of quite valid and very important points, there are a number of issues on which comment is appropriate.
Inasmuch as Prof. Greiner has referred to my work, his interpretations should be viewed in the proper context. They are not, for several reasons.
Reference 7 cities the initial two parts of a four-part series on slew -induced distortion (SID), which appeared in February, 1977. Therefore, many of the comments in the text relevant to THD tests (i.e., Figure 5) are off base, simply because they have already been answered (1.3).
The third part of this series "Transient IM tests for SID" (1) was not referenced and discussed, and it makes points which are certainly germane to Greiner's discussion. The work of coauthors Mark Stephens of Sound Technology and Craig Todd of Dolby Labs sheds highly important light on the efficacy of the sine -square test for TIM (2) and correlates results of this form of test with the THD and two-tone IM methods.
Without going too deeply into the many issues involved, it can be stated that the results of this work and further studies (3) generate very serious questions as to the validity of the open loop bandwidth criteria (Greiner, rule 3b Pg. 60) and reference (2) Our tests show that TIM or SID performance is relatively independent of the amplifier open-loop bandwidth as long as a minimum slew rate criteria is satisfied.
In one specific instance an IC op-amp shows superlative results for TIM and THD tests with a 20 Hz open-loop bandwidth, but a 10V/µS slew rate.
It is not apparent to me how an open -loop bandwidth of less than 20 kHz can in itself cause distortion, if amplitude clipping and slew rate limiting are avoided. In fact, many measurements have been made to test for such distortion, with results thus far always indicating slew rate as the dominant frequency dependent non linearity. I would like to see some test results which support the open -loop bandwidth criteria, as independent of slew rate. Could it be that rule 3 should really say "the open loop large signal bandwidth," which then specifically qualifies it to mean slew-limited bandwidth? Our work has shown that a safety factor of 4 or more is justified here, not just as an equal bandwidth.
This is not to say that such a large open -loop, small -signal bandwidth would not be a valuable asset; to the contrary. In practice, however, such wide open -loop bandwidths are highly uncommon, and if you apply the criterion generally to the whole audio signal chain, the result is an impossibly large requirement for open-loop bandwidth by the time the signal reaches the power amp (1MHz!). This is because the required open-loop bandwidth increases for each stage, and there are usually at least three of them in the path.
It is unfortunate that Greiner has not, in his attempt at a "unifying" article, made use of all the information available. Perhaps if this had been done, some of his skepticism as to the validity of THD tests for SID detection would not be so apparent. Also, Part IV of reference (1) includes an interesting correlation of electrical/audible results, while reference (3) includes a theoretical prediction of SID performance. Both of these may have influenced some of the comments in the article.
---Walter G. Jung; Forest Hill, Md.
1. W.G. Jung, M.L. Stephens, C.C. Todd, "Transient IM tests for SID," JAES preprint [Feb., 1977]. "Slewing Induced Distortion in Audio Amplifiers," Audio Amateur [March, 1977].
2. J. Curl, E. Leinonen, M. Otala, "Method for Measuring Transient IM Distortion." JAES 25, 4 [April, 1977].
3. W.G. Jung, M.L. Stephens, C.C. Todd, "Slewing Induced Distortion and Its Effect on Audio Amplifier Performance." AES preprint 1252 [May, 1977].
And from Madison ...
I am sorry that Mr. Jung is upset that I did not reference Parts III and IV of his articles on distortion in operational amplifiers. Only parts I and II were available to me in Spring of 1977 while I was revising my paper for Audio. I made reference to the first two parts because I thought they were an important contribution to finding good ways to measure slew induced distortion in amplifiers. I have now had a chance to review the final parts of this work and, as I said in the Audio article, this work is quite convincing. Mr. Jung and his colleagues have carried out an immense number of measurements and shown good correlation of these measurements with listening tests. It certainly looks like their technique can be used to select operational amplifiers for various uses.
My skepticism of the technique relates only to the question of how this measurement technique, or any other for that matter, can be related to a detailed understanding of the actual internal mechanisms that cause the distortion. It is possible that the overload mechanisms may be slightly different, or occur at a different point in the circuit, in power amplifiers as compared to operational amplifiers.
Thus, I would like to see these same techniques applied to a large variety of power amplifiers which are used under load conditions that approach real loudspeaker loads.
It seems quite possible that the same measurement techniques will give good results for power amplifiers and that this test should be applied more widely. In any case, Mr. Jung's articles should be required reading for amplifier designers, manufacturers, and audiophiles alike.
With reference to rule 3b in my Audio article, I can only assume a misunderstanding of it by Mr. Jung.
The conditions for application of rule 3b are stated in the parenthetical sentence following it. I did, of course, mean the bandwidth under large signal or full power conditions. This rule is rather difficult to follow since it has implications not only for individual parts of a system but for the system as a whole. The rule, as stated, does guarantee that there will not only be no slew rate limiting but that there will be no internal overdrive within the feedback loop. It is possible to not follow this rule exactly if the internal active stages in the amplifier have good amplitude headroom and if the high frequency content of the program material is not too great.
All in all, I would consider my attempt to show that essentially linear measurements, and concepts, can be made to correlate well with non-linear measurements and phenomena to be supportive of the work of Mr. Jung. For too long writers have implied that transient measurements somehow had no relation to steady state measurements when in fact they do. While this relationship gets more diffuse as the nonlinearities get very large, we would not usually want to drive an amplifier much beyond its linear output capacity in any case.
In summary, I find no major disagreement with Mr. Jung on either the validity of his method of measurement or his desire to sort out the status of the many amplifier types available.
--R. A. Greiner; Professor of Electrical Engineering, University of Wisconsin Madison, Wisconsin
A problem of growing concern to the serious audiophile of today is the radio frequency interference being picked up by high -quality audio equipment.
Although the problem is often caused by an improperly or illegally operated transmitter, or by "interference prone" audio equipment, it is usually difficult to enforce either legal or technically.
Proper operation of most private transmitters or enlist the help of the transmitter operator in dealing with the situation (If the operator can be located at all). This is especially true if the transmitter is located in a moving vehicle.
It does not seem fair to require the audiophile to modify his expensive equipment to free it of interference from such transmitters, especially when the modification may very well cause a noticeable deterioration of the frequency response, phase response, etc. of the audio system. However, in the case of an impossible to locate offending transmitter or if the operator is reluctant to co-operate, modification of the audio equipment may be the only reasonable solution for the audiophile.
In such a situation a switchable filtering device would be highly desirable because any adverse affects of the filter on the sound of the system need only be endured when the offending r.f.i. is present, and said adverse affects on the sound will almost certainly be less annoying than the r.f.i. itself.
--Walter M. Scott; Ill Knoxville, Tn.
(Source: Audio magazine, Mar. 1978, )
= = = =
Prev. | Next