Audio Power Amplifier Design book- Douglas Self wants your opinions

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Your email ended with the words "Oh well". If you have the right tool, you could do a search on these words.
Frankly I'm not that bothered. Pretty busy designing at present.

What? Six inches long? That is more appropriate for certain human attributes. For amplifiers however, in particular with MOSFETs in the O/P stage or in case of output inclusive compensation, this is way too long. If your amps can stand this length, they must be very stable.

They are. I design for quantity production. Dependable stability with any load is an absolute requirement.

Shunt is shunt, no matter how large the caps are and no matter how well the supply rails are decoupled to ground. Cherry stated that these caps are crucial for stable performance. So if it is not (a mild) shunt compensation, what else it could be?

No, really not. For those capacitors to have a significant effect on the compensation they would have to have values of tens of nF. As explained before, their purpose is the prevention of parasitic oscillation in the output stage.
 
I wonder if both these worthies (excuse my use of the 3rd party, Mr. Self) have tried Cherry's own solution to this; an emitter resistor on the VAS (R14 in #2321).
Yes. It does not work.

My own experience is that Cherry's collector base caps on the driver (evil cos they connect to the VAS output) may not be necessary either.
DouglasSelf said:
They are there with intent to control parasitic oscillation in the output stage. From my experience, I would think they are probably very necessary.

No. I haven't tried this circuit for real.
DouglasSelf said:
Whereas I have.
Doug, I was referring to my examples in the Discrete OPA thread.

As it so happens, I HAVE tried pure Cherry on a Blameless topology and did get it to work for real (and without the caps on the drivers too), circa 1990. also TPC and other stuff but pure Cherry was best.

Can you post a circuit which exhibits this instability with pure Cherry? If we can get it to be unstable in SPICE world, I'd like to see if I can repeat my Jurassic real life experiments.

This beach bum apologises for being somewhat lacking these days for real life stuff. :(

At least the exercise might help my poor efforts at getting LTspice to replicate nasty real life problems. :)
 
six inches

[...]
Can you post a circuit which exhibits this instability with pure Cherry? If we can get it to be unstable in SPICE world, I'd like to see if I can repeat my Jurassic real life experiments.
[...]

You also might simulate the effect of (ridiculous) 6 inch long wires to the output devices. A wire, 1mm thick and 6 inches long, has a self-inductance (no pun intended) of 173nH.

Cheers,
E.
 
You also might simulate the effect of (ridiculous) 6 inch long wires to the output devices. A wire, 1mm thick and 6 inches long, has a self-inductance (no pun intended) of 173nH.
We need to simplify as much as possible .. not add complexity which will make things further from real life .. especially if good practice can get rid of certain evils.

6" leads to the O/P devices are the sort of thing which will require Cherry's 33p on collector/base of the drivers.

A "Blameless" with 2 output devices on the PCB and using old fashioned "through hole" should easily fit a 2" x 3" space. And take into account all good layout practice including Mr. Self's recommendations.

I do simulate 'long' leads from PSU to the amp as this is unavoidable.
 
Member
Joined 2011
Paid Member
6 inch - you never have seen a papa pass amplifier?

6 inch and more from input/vas stage to output devices is no problem for a good designed blameless.

If you have ever had something to do with low to medium radio frequencies, you should know the wave length and delay of e.g. 1MHz freqency. All you need to know is how to avoid and/or suppress hf oscillation...

E.g: Papa Pass also uses in his X amplifier series very long ribbon cable from input/vas stage to his power output stages.

Regards,
Toni

Hint: power+/gnd/vas+/gnd/output nfb/gnd/vas-/gnd/power-/gnd
 
Member
Joined 2011
Paid Member
Never simulated a cable, but perhaps this way:
L1 and L2 half of self inductance and C1 with some pF

 

Attachments

  • linesimulation.jpeg
    linesimulation.jpeg
    9.1 KB · Views: 355
Last edited:
6 inch and more from input/vas stage to output devices is no problem for a good designed blameless.

If you have ever had something to do with low to medium radio frequencies, you should know the wave length and delay of e.g. 1MHz freqency. All you need to know is how to avoid and/or suppress hf oscillation...

E.g: Papa Pass also uses in his X amplifier series very long ribbon cable from input/vas stage to his power output stages.

Regards,
Toni

Hint: power+/gnd/vas+/gnd/output nfb/gnd/vas-/gnd/power-/gnd

Hi Toni,

We are talking about Cherry's amps with output inclusive compensation. That's a different kettle of fish, just because of HF issues.

Also, we are talking about lead lengths between the drivers and output trannies, thus not between the TIS output and driver input.

Cheers,
E.
 
Last edited:
History & compensation

.........I think we should try to record the history of audio as it happens.

Frankly I'm not that bothered. Pretty busy designing at present.

Hmm... Not interested in history anymore?

For those capacitors [33pF] to have a significant effect on the compensation they would have to have values of tens of nF. As explained before, their purpose is the prevention of parasitic oscillation in the output stage.

Sure, to prevent parasitic oscillations. Isn't any compensation (in power amps) meant to prevent oscillations? So if 2x33p could stop it, I would call that a significant effect.

Cheers,
E.
 
Member
Joined 2011
Paid Member
Dear Edmond,

sorry - my mistake - next time I will read carefully ...
I meant 6 inches between input/vas to output stage as seen in attached pictures.

Thx for clarification,
Toni
 

Attachments

  • myblameless_input_vas_output_test.jpg
    myblameless_input_vas_output_test.jpg
    94.7 KB · Views: 323
  • myblameless_output_stage.jpg
    myblameless_output_stage.jpg
    102 KB · Views: 301
Member
Joined 2011
Paid Member
compensation technics; OLG; phase margin; practical testing

Dear Douglas,

as you can see in post #298 you have inspired me deeply with your 5th edition. Thank you very much for all your great books (APAD; The Design of Active Crossover; Small Signal Audio Desing ...).

My wish for the 6th Edition (which I will definitly buy):

  • More details about different compensation variants (wether or not they have 3 to 5 characters abbreviation)
  • More details about improving slewrate
  • Phase margin, OLG simulation and/or testing in real world designs
  • More practical testing (max. output power; peak tests ...)
Thank you and best regards,
Toni
 
Cherry's amp and compensating the compensation loop.

[..]
No, really not. For those capacitors to have a significant effect on the compensation they would have to have values of tens of nF. As explained before, their purpose is the prevention of parasitic oscillation in the output stage.

I'm afraid you doesn't understand the purpose of these 33pF capacitors. They are not intended to replace -in part or in whole- the Miller compensation. Opposed to Miller compensation around the TIS, which is meant to define the gain and phase response of the global feedback loop, these 33pF shunt caps are meant to define the gain and phase response of the Miller loop itself. Since the unity loop gain frequency of the latter is at least ten times higher (10..20MHz), this explains why these shunt caps need to be much smaller and certainly not ' tens of nF', which is crazy.

If you still insist they are meant to prevent parasitic oscillations, then, for the same reason, we could argue that the purpose of Miller compensation is also to prevent parasitic oscillations.

BTW, similar issues exists with input Miller compensation. Also in this case we often need additional compensation. See for example Bob's book, 'Compensating the Compensation Loop', page 181 (you do have his book, don't you?).

Cheers,
E.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.