Power Supply Resevoir Size

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Frank & Tom:

If you remove the entire amplifier & modulated load from the sim, and replace with a resistor, you will see the droop caused by the transformer "regulation" - which will be mostly due to the leakage inductance and the narrow conduction angle. I'd also look carefully at the rectifier model as well - its probably fine, but there are, for example, 10BQ100 spice models with 100R slope resistance (I found them the hard way)
Don't worry, Terry, I've been around the perfect resistive load scenario a vast number of times; the trick here is to deal with the real world of amplifier linear PS's, and class AB loading, and to see what the best way to handle that is ...

Frank
 
Will your sims allow for the speaker(s) reacting?
Can't help but think that a proper model from AC mains to speaker is a level of work that'd be a 3rd of a phd. Add a good model of the human ear and psychoacoustics and you'd be looking at tenure.

Does the speaker prefer current to be stored in a cap for bursts or flowing via transformer, I dunno, but I do know that its easier to add caps to a psu than it is add transformers to a case thats already been laid out.
 
Just a cautionary thought, Tom ... my belief is that real circuitry should be used to load the PS, otherwise one's thinking can possibly be led down an unnecessarily torturous detour. Ultimately, what one comes up with in terms of results has to relate to real, on the bench, bits of circuitry, because that's what the whole exercise is about ...

Cheers,
Frank

Frank,

By "real circuitry" I assume that you mean "simulated real circuitry" instead of behavioral current sources driven by equations. <smile>

Normally I would tend to agree but in this case there is an important advantage to using the B-sources, and, I can't see any difference in their behavior, compared to "real" circuitry.

I figure I can at least use the the B-sources to examine the output error and then use "real" circuitry for verification.

Cheers,

Tom
 
Frank,

By "real circuitry" I assume that you mean "simulated real circuitry" instead of behavioral current sources driven by equations. <smile>

Normally I would tend to agree but in this case there is an important advantage to using the B-sources, and, I can't see any difference in their behavior, compared to "real" circuitry.

I figure I can at least use the the B-sources to examine the output error and then use "real" circuitry for verification.

Cheers,

Tom
Yep :), but what I'm questioning is the fact that using the u(x) step function is reverting the sim to a "perfect" class B amp, rather than class AB. The fact that you had to bypass with a cap says that it's not matching reality, you're switching the supplies on and off too hard ...

But, there are always many ways to skin a ... -- if it works for you, go for it! :D

Cheers,
Frank
 
Will your sims allow for the speaker(s) reacting?
There's no reason why the kitchen sink can't be thrown at doing a mockup in terms of the detail added; the good thing is that it gives one a highly flexible, and speedy handle on understanding what's really important, and what's not. So a really nasty speaker load, and badly distorted AC mains can be introduced down the track, to check their impact.

Frank
 
Frank, yeah I realise that, and its why this is such a neat thread.

the suggestion was merely to illustrate where the droop is coming from.

Andrews "filter" approach is a nice way of looking at it.

but I think what this thread has shown is that we can clearly see three regions: LF (< 100Hz), MF and HF (> 10kHz) (arbitrary numbers, YMMV)

in the MF region its a capacitor bank with ESR. more C, less ESR = good

the HF region handles the transient response, and is dominated by parasitics - ESR, ESL - of both the components and the construction. low-inductance layout/interconnect and paralleling many smaller caps (if done properly) = good.

this has been most of the focus of this thread. But "optimal" (whatever that is) design of the cap bank taking all this into account wont help with....

the LF region, which the droop aspect highlights. It is dominated by the transformer and its interaction with the DC bus caps (more C = narrower conduction angle = higher peak current = greater voltage drop in xfmr).

I think it fair to say that what frank, tom et al have demonstrated is that the transformer needs to be designed too. probably every bit as carefully as an audio output transformer. low leakage inductance is a must, and even then it needs to be designed in conjunction with the cap bank, rather than just blindly throwing C at it, followed by stupidly oversized transformers.

A toroid is, of course, a good start, but it would be really interesting to look at a range of transformers and see how well they perform. my experience with LF transformers is that they tend to be designed automatically, and invariably emphasise electrical safety and thermal performance ahead of electrical performance.

I've seen designs from a variety of places (including some that really ought to know better) that completely ignore proximity effect - "its an HF thing". no it aint, its just that as long as your conductor diameter < 5mm it wont matter at 50/60Hz, so for little stuff its almost never a problem - but above 10kW or so it can and does matter. really big stuff is of course done properly (thou shalt not screw up power systems).

a physically split bobbin to keep primary and secondary separate is a brilliant way to produce a very high leakage inductance, and makes electrical safety (e.g. hipot testing) a breeze....
 
could it be the complete and utter lack of context? ditto re. supporting material. your comment #426 doesnt actually say anything at all. "capacitance multiplier." what. good? bad? indifferent? Luckily your comment #430 clarifies this, and furthermore puts it on a sound theoretical footing by judicious use of a full stop and the absence of any form of rational argument whatsoever.
 
what if you design the capacitance multiplier such that the nominal base voltage is lower than the expected trough in collector voltage? that'd do the trick. there's always the issue of time constant, but that can be solved at DC by turning the capacitance multiplier into an open-loop regulator with the aid of a zener or two. In either case the droop would reduce loss in the transistor as Vce drops with increasing load.

instead of trying to reduce LF droop by improving transformer coupling (which can never be perfect, and in practice is hard to get above 99%) an alternative would be to make it independant of load. take a low-leakage transformer and add a variable inductor in series with it. make the external inductance inversely related to load, so as to maintain a constant level of droop, so the DC bus caps stay at a constant voltage.

its a hop-skip-and-a-jump from there to a constant-voltage transformer....which may well be good for driving a rectifier-capacitor filter as the waveform is trapezoidal (so very large conduction angle). 1953 called, and offered its technology.....
 
o yeah, lowere the basis voltage. However that means that the output voltage at the emitter als drop and increasing the heat loss over the transistor. No profesional will do this. Keep it simple and just use a big capacitor. There is no such as a transformer stable device. The drop in voltage is just due to discharge of the capacitor. Nothink can change that, only bigger cap can cure to a certain level.
 
...............Does the speaker prefer current to be stored in a cap for bursts or flowing via transformer, ..............
there is no choice. The transformer is very effectively NOT coupled to the PSU for the whole time that the diodes are in reverse voltage, i.e. not conducting. This inactive time is very approximately 90% of the time.

The capacitors in the PSU are the POWER SUPPLY for the whole duration of current demand by the amplifier.

The transformer only recharges the partially depleted capacitors.
I did have "simply and" between transformer and only, but realised that "simply" does not apply, due the complex nature of the charging pulses and the reverse current flows and the resulting ripples as the transformer gets all hot and bothered by the pulsing on it's output.
 
my level of electronics? I an MSc in Electronics teaching on the university. Enough? And what is your level?

Is this a contest of qualifications or are we addressing and trying to solve a critical component in audio amplification?

There is a saying in the sciences that you get to know more and more of less and less until you reach the satge that you know everything about nothing.
 
Is this a contest of qualifications or are we addressing and trying to solve a critical component in audio amplification?

There is a saying in the sciences that you get to know more and more of less and less until you reach the satge that you know everything about nothing.
There is no critical component, just use a big cap instead using nonsence showing incompetence of the maker.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.