Bob Cordell's Power amplifier book

Jump to Latest
#10,622
Mark Johnson said:
Maybe the original intention was to have the relay contacts' swing-arm attached to the amplifier's output jack. In the "play" position, the output jack (swing arm) becomes connected to the amplifier output (L1) and the amp drives the loudspeaker load. In the "mute" position, the output jack (swing arm) becomes connected to ground, and the amplifier output (L1) is floating. When muted, the amplifier is not connected to the loudspeaker.

_
Click to expand...
I have exact experience with it. When my amplifier power ON, one fuse at PSU amplifier blown. I'm confuse because I think my amplifier work good in simulation. But my relay in speaker protector is wrong. I blown several fuses. My amplifier is OK, thank to the fuse.
 
#10,624
#10,625
tiefbassuebertr said:
Concerning a wanted method of detecting a tendency to oscillation resp. ring ring - yet before it actually comes about I start this Thread:
https://www.diyaudio.com/community/...-ring-in-audio-power-amplifier-stages.392385/
Maybe there are to find helpful hints here on diyaudio (maybe even in this thread) or anywhere on the web.
Thank you very much.
Click to expand...
I built a so-called parasitic oscillation sniffer a long time ago. This addresses the fact that parasitic oscillations often appear as a burst on a high-amplitude sinewave signal, making them sometimes difficult to see, much less measure their frequency. The sniffer basically consists of a wideband amplifier with a few stages of fast emitter-follower-based second-order high-pass filters, leaving only the burst to see. Bandwidth was in the tens of MHz, so as to be able to catch the high-frequency local parasitic oscillations that Vertical MOSFETs were capable of. An amplitude detector would flash an LED and also provide a trigger source for the scope.

Cheers,
Bob
 
  • Like
Reactions: Damon Hill
#10,631
dadod said:
Hi Bob,

Finally I bought the second edition, and reading it carefully and slowly.
I checked bias spreader on page 412 used for ThermalTrak OPS and found the same error as in first edition.
Here is the link where you confirmed the error.
https://www.diyaudio.com/community/...r-amplifier-book.171159/page-222#post-4083455

BR Damir
Click to expand...
Hi Damir,

Thanks for bringing this to my attention, and I am sorry that it slipped through the cracks in the 2nd edition. I looked back at the post you linked to, and that was helpful. In looking at the 2nd edition and your previous post, I'm now not sure that some of what I said was wrong. Namely, IIRC, R1 and R3 are indeed the resistors that control the proportion of compensation from the temperature of Q1 and the temperature of the TT diodes. If you look at the extremes of the ratio, it can be seen. If R3 is open, the Vbe multiplier just looks like an ordinary Vbe multiplier controlled by the temperature of Q1, and just in series with TT1 and TT2. In this condition, the temperatures of TT1 and TT2 have no influence on the Vbe multiplier itself; they are not part of the Vbe multiplier action.

At the other extreme, if R1 is open, TT1 and TT2 are inside the Vbe multiplier and their voltage changes with temperature get multiplied, so have more effect. This is pretty much the same issue for Figure 17.19 a, b and c. However, some of the math errors you pointed out may actually be errors anyway. It's been a long time since I worked on that 🙂.

In any case, if my explanation above is correct, it is my fault for not explaining it better in the book.

Thanks again for your feedback.

Cheers,
Bob
 
#10,634
Has anybody built the MOSFET amplifier with error correction?
Click to expand...

It crossed my mind at one time. Working with Bob, we have designed pcb's for DH-220C and BC-1. DH-220C is popular, but only 3 folks have signed up to build a BC-1, so I do wonder if designing more amplifiers is the best use of my time. So it comes down to who's intersted? For one person, not one of my top priorities. The mechanical design aspects need to be considered for any design. Its possible to use the 3Ux300mm Modushop chassis we used for the BC-1 design, since the infastructure is in place. Not like I need any more amplifiers however.
 
  • Like
Reactions: SomekPoland
#10,635
SomekPoland said:
First of all, thank you @Bob Cordell for an excellent book!

Has anybody built the MOSFET amplifier with error correction? I'm interested in somethin for my KEF LS50 Meta. It seems like a nice match. Should handle 4Ohm loads with ease.
Click to expand...

Bob:

This posting reminded me that I wanted to ask this earlier.
Have you thought about or played around with adding the error correction circuitry to a lateral MOSFET output stage?
Tanberg did many years ago. I think it was the model 3016.
Would be an interesting option to incorporate into the modified Hafler ...
 
#10,636
rsavas said:
It crossed my mind at one time. Working with Bob, we have designed pcb's for DH-220C and BC-1. DH-220C is popular, but only 3 folks have signed up to build a BC-1, so I do wonder if designing more amplifiers is the best use of my time. So it comes down to who's intersted? For one person, not one of my top priorities. The mechanical design aspects need to be considered for any design. Its possible to use the 3Ux300mm Modushop chassis we used for the BC-1 design, since the infastructure is in place. Not like I need any more amplifiers however.
Click to expand...
If I would decide to build it, I would want to design the PCB myself. I finished a good course and I'm itching to put it into good use. I appreciate all the help, if you wish to get involved. Maybe it would be interesting idea to then drop the project into the diyaudiostore if Mr. Cordell and Mr. Donald agree.

Right now I'm refurbishing a Sony T-F770ES, besically rebuilding it from ground-up. That will take me another month I think, but then...
 
#10,639
petr_2009 said:
1) I agree with you, every note of a piece of music begins with a break of the first or second order.

Hit the piano key - get an infinite spectrum. Hit the cymbal with a stick - get another portion of the infinite spectrum, and so on from the beginning to the end of the piece of music.

And in order to reliably amplify such signals, the amplifier must be as broadband as possible, strictly speaking - with an infinite bandwidth.

2) For example, Hafler said that the amplifier should be as broadband as possible, although I have not seen anywhere in his statements which band is enough. Cyril Hammer argued that an amplifier with feedback should have a negligible signal propagation delay - only a few ns, and this requires a bandwidth of tens of MHz.
Click to expand...
1) The initial transient of a musical instrument may produce a infinite spectrum span. In other words, a infinite high frequency content that needs a infinite bandwith of the amplifier to reproduce the original signal correctly.

However... The amplifier never gets to see that original signal because every single stage in the recording chain and in the reproduction chain forms a low pass filter that limits the spectral content. Starting with the microphone... Game over.

2) I think we can look at feedback as a iterative process. Every time the signal has travelled through the amp stages and arrives at the base of the second transistor of the differential pair, the amplitude deviation from the original waveform at a point in time is corrected to some degree. Approximately, what matters, is how many times does the propagation period fit into the signal period, before phase errors noticeably start to falsify the process? Or, how many times does the time period of the amplifier bandwith fit into the time period of the test signal ?

Looking at a steady 1kHz sine for example, the delayed signal arriving at the feedback node is attenuated due to the phase shift, in effect the feedback depth is reduced more an more as signal frequeny rises.
Secondly, the correcting feedback signal arrives at the wrong time. The differential stage does see a wrong value, especially bad for classB spikes.
So imho Hafler did mean what he said: as broadband as possible, the introduced error starts immediately. Less bandwith = more error.

Transistors do not allow to built amplifiers with infinite bandwith, feedback can not work perfectly. It is what it is.
 
You must log in or register to reply here.
Top Bottom