Power Supply Resevoir Size

[jacco vermeulen]

Originally Posted by mightydub

1) rails 20V higher

Transformers must be cheap overthere.

[AndrewT]

Originally Posted by mightydub

1) rails 20V higher (my rule of thumb, yours may vary) than the maximum output swing so that the rails can sag under load but still be comfortably above the maximum output voltage;

That's just about equivalent to specifying a 180W into 8ohms, (53.7Vpk & +-60Vdc supply rails) amplifier and never using it @ >=100W into 8ohms, (40Vpk).
What a waste of resources and design effort.


Most "Rule of Thumb" guidelines will do far better than applying "-20V" to the supply rails!

If one were working with a 60W amplifier (+-36Vdc supply rails) then the "-20V" rule becomes silly.

[fas42]

But, that 20V does appear at a simple glance to be what's needed. For the sake of consistency, in comparing like with like, I used and adapted Tom's last posted circuit, for the moment "ditched" his fancy capacitor model because there were some spurious glitches appearing in the sim:

VoltageSag01.gif

The PS rails with no load, the output stage disconnected, sit at 82V. Then, reconnected and idle current flowing, the rails drop to 71V. Finally, add a rather vicious, constant load of summed 20Hz, 635Hz, and 20,000Hz (635 midway point ...) sine waves which peak just below clipping, and see what happens to the rail:

VoltageSag02.gif

Sags right down to 59V in less than half a second -- that 20V margin doesn't look that bad after all now ...

And what happens to the output with droopy rails? A nice bit of clipping ...

VoltageSag03.gif

Frank

[AndrewT]

Fas,
what is the supply rail voltage when the amplifier is delivering full power?
I ask because the 9Volts lost from no load to quiescent load seems enormous, compared to "normal" power amp PSUs. I'd expect to see the supply rails of a 100W into 8r0 amplifier sagging to ~+-45Vdc @ full powwer, from a quiescent of ~+-50Vdc and zero load of +-50.5Vdc. This is 0.5V sag, cf. your 9V sag.

[fas42]

Originally Posted by AndrewT

Fas,
what is the supply rail voltage when the amplifier is delivering full power?
I ask because the 9Volts lost from no load to quiescent load seems enormous, compared to "normal" power amp PSUs. I'd expect to see the supply rails of a 100W into 8r0 amplifier sagging to ~+-45Vdc @ full powwer, from a quiescent of ~+-50Vdc and zero load of +-50.5Vdc. This is 0.5V sag, cf. your 9V sag.

Remember, at the moment, I'm just using Tom's transformer parameters as is, I haven't looked closely at how "reasonable" they are ...

At idle, the output stage dissipates about 13W, 5W in Q3 and Q4, 1.6W in Q1 and Q2. Then, driving with a pure 1kHz tone to just clipping, the voltage sag is stable at 57V. Voltage p-p is just over 100V, and power dissipated in the 4R load is 320W.

Frank

[AndrewT]

320W into 4r0 is 50.6Vpk (101.2Vpp).
That is a reasonable unclipped maximum from a loaded PSU supplying +-57Vdc, i.e. 6.4Volts of losses through the amplifier. Particularly when the 57Vdc is actually modulated with the PSU ripple (what is the loaded PSU ripple when delivering maximum power?). This leaves the Vsupply minimum ~55V, or really a 4.4Volts loss through the amplifier.

Can you change the transformer data/model to more realistically mimic what a "normal" transformer shows when unloaded/loaded? Maybe Gootee could supply data.

[fas42]

Originally Posted by AndrewT

Can you change the transformer data/model to more realistically mimic what a "normal" transformer shows when unloaded/loaded? Maybe Gootee could supply data.

This is where the action is. The transformer, as "supplied", has secondaries looking like this at full power, 320W into 4R:

TransSec01.gif

Nasty looking, eh? But how realistic ...?

If I back off to idle, things improve somewhat ...

TransSec02.gif

Then, disconnect all load to the rails, and we get the textbook result:

TransSec03.gif

Interesting times ahead ...

Frank

[fas42]

Sorry, ignore that figure of 82V I posted earlier for no load conditions. As Andrew pointed out, there was something awry there, and it turned out to be a combination of me being confused with a result taken under different conditions, earlier on, and not allowing sufficient time for the voltage to settle. A much more accurate value is a touch under 73.4V, so the transformer model is most likely to be of good accuracy ...

However, that doesn't change the main story that significantly, on the need for close to 20V "safety margin": 73.4V no load, 57V for full power, no clipping ...

Frank

[liching1952]

you can limit high ripple due to small Capacitance by using analog stabilisation or a Switch mode power supply

[gootee]

Here is the transformer model.

The only transformer models I have made are for transformers with around 22-30V outputs, since those are the only ones I measured (see measuremnt steps in model image).

So what is being used in the simulations is this model, but with a much higher input voltage amplitude than normal, in order to make the output voltage higher.

I just noticed, too, that the inputs and outputs are labeled. When identical labels are used in spice, those points in the circuit are joined, as if with a wire. So in the simulations so far, the secondaries of the two transformers have been in parallel. I will change that and see if it makes a difference, and will post a new overall model with the correction.

P.S. Frank, I lowered the input voltage to 321 V, to get an unloaded output voltage of roughly 71.136 Volts.

Cheers,

Tom