I was talking about the book as such. 1984 is 30 years ago; no measurements since then?
Since we are talking about quiescent dissipation, then there will be negligible losses in the rectifiers. I don't think the quiescent losses in the transformer (due to magnetising current) will be anything like 5W. That would mean that a transformer connected to the mains but with secondary open-circuit would get perceptibly warm.
That is not my experience.
I did not understand your answer - you could have just said "More specific about what? - your numbers are correct" ;-)
Anyway I guess a good design will have to work from low supply rails in order to ensure that turn on/off thumb is minimized.
The class XD idles somewhat higher ? is this better than a good blameless - is it in the new book
Best regards
Jens
You have done a great job of exposing many things via measurements which one might not know by sim alone. The practical side is just as important or more so. From measuring - it opens up new explorations into the mechanism that causes this or that is tuely -IMO - important both as a process and in the results/findings.
THx-RNMarsh
Last edited:
What we now need are some/more asymmetrical waveform tests to see which topologies come out better or what new surprises, if any, there will be found and then the why.
THx-RNMarsh
THx-RNMarsh
You seem to be implying that powerful amplifiers are not feasible, which is obviously not the case. A DC-coupled amplifier must have an output relay to protect the loudspeaker from DC faults. It is simple to use this to mute the output for a second or two at switch-on, so thumps become irrelevant.
Yes and yes. Crossover Displacement gives you a low-power region of true Class-A operation, with no downside. (unlike Class-AB) There is a complete chapter on my Crossover Displacement technology in APAD6.
[/QUOTE]
Of cause good powerfull amps can be designed without thump and relays.
I was referring to the ability for a circuit to be well defined within a broad supply range will result in a circuit less likely to produce turn on/off thumps. This simply because the energy in the system during transition from undefined state to defined state and back is minimal when the transition happens at low rails.
Great will read up when I get my copy - thanks
\\\Jens
Last edited:
But you do summarize (quite effectively) back in #125 . . . not worth the effort.
Thanks for the link. (a good deal more helpful than some people) The Stereophile results show some very mediocre THD plots- flat at 0.007% across the band at 80W/8R. That's ten times more than a Blameless at most frequencies.
Hifi Engine don't seem to have the RB-1080 service manual.
ElektroTanya | Service manuals and repair tips for electronics experts does have the service manual.
It's a double-input-stage circuit looking very much like Figure 8.14 in APAD6. The THD plot in Stereophile also looks very much like my Figure 8.16. The latter shows a mixture of 2nd and 3rd harmonic, produced partly at least because there is no EF in the VAS, and so Early effect and non-linear Cbc in the VAS transistors work their evil.
Thanks for the link- that allowed a most interesting comparison of performance, and gives a good cross-check on my Chapter 8.
Check out model RA or RB970. Its the exact same design except minor difference in board layout and components. This is a design that has been in use for over 20 years. It was designed by Stan Curtis and has over the years seen only minor modification and been highly regarded over the years winning awards in What Hifi and several other publications usually as best amplifier in the 300 to 350 pound range. A newer model won an award as late as 2012, see here one such award Rotel RA-10 review | Stereo amplifiers | Reviews | What Hi-Fi?
Mr. Self, I am sure you realize that the Stereophile measurements are "end to end" for a finished amplifier, while yours appear to be (correct me if I'm wrong) on the bench. In my opinion, for anything under 0.001% it is the implementation (wiring (in particular for the power supply), shielding, signal path, ground loops, etc...) that controls the overall results. Obviously a bench version has much less constraints than a finished version, even for the same amp.
What I'm trying to say is that comparing the Stereophile measurements with your own is not really apple to apple, you are in fact not comparing two topologies.
I believe that your less than optimal results with symmetrical VAS topologies are due to the known "fighting VAS" issue. Long story short, at HF the two VAS devices in a symmetrical topology are acting (due to the compensation cap miller effect) as two (almost ideal) voltage sources in parallel, which is always a bad idea to do. As a result, here a rather large common mode current develops, which compromises the performance.
Probably the most elegant solution was already mentioned here: a common mode control loop (CMCL). Such a loop does not necessary require highly matched or thermally coupled devices. See again the PGP amp for an example of CMCL (Edmond has more on his web site). Without such a CMCL, reaching 0.0001% THD20 @100W/8ohm would be impossible.
I'm sorry, but I can't agree with you here. The RB-1080 has an unbalanced input that goes directly into the power amplifier. I assume that Stereophile used this, though I don't think that is confirmed anywhere in the article. If so, there is no extra circuitry in the path and the tests are equivalent to my tests of PCBs on the bench.
The implementation can degrade the THD at much higher levels than that. The mechanisms are spelt out in APAD6, but you are absolutely right that the power supply wiring layout is potentially the most troublesome. If the power tracks are on the PCB then you are committed and it is very difficult to try varying layouts without a major re-lay of the PCB. Managers (I use the word loosely) frown on this sort of thing. Power supplies should come in on wires you can wiggle about and re-route as required.
It is however perfectly possible with a little care and knowledge to get results in a production box that are as good as on the bench. I have done it many times.
I think I am, and the RB-1080 is a most convincing demonstration of the shortcomings of one kind of push-pull VAS.
I don't think that is a good interpretation. At the moment I don't really have much more to say than is in Chapter 8 of APAD6.
I am not sure that a configuration which needs a servo to make it work properly can really be described as elegant. However, I do agree with you that this approach should be further investigated, but I am not convinced it will solve all the problems.
Many thanks for thoughtful input!
Last edited:
About 13 years ago, I measure my amp with push pull VAS. I get 0,006% THD at 1kHz, 100W/10ohm. I use standard components, and no matching at all.
At that time, I do not know it is importance to measure THD at 20kHz and I did not read your book and Bob Cordell book yet.
At that time, I do not know it is importance to measure THD at 20kHz and I did not read your book and Bob Cordell book yet.
I guess that Rotel RB1090 is almost identical. See att. This kind of topology is very popular in high power sound reinforcement and studio amps. (Mackie, Alesis, etc)
Attachments
Last edited:
I would, really like to have your comment on this here http://www.diyaudio.com/forums/soli...-6th-edition-douglas-self-21.html#post4011827. If you think it's not worth to be implemented, please, explain why.
BR Damir
- Status
- Not open for further replies.