Amp design attempt number 2 (simpler)

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Vclip is the minimum voltage across the amplifier, but between the voltage rail and the output.

If you think about how big the output signal can be before clipping starts, you imagine the space available, from zero volts to the rail voltage. Most people realize that the ripple voltage has to fit at the top of that range and the signal has to fit at the bottom. But if you look at an amp schematic, you can realize that the amplifier itself must be in there somewhere, too. Its voltage has to fit between the ripple and the signal.

If you simulate the onset of clipping, and plot the signal and ripple on the same plot, you will see the ripple start to gouge chunks out of the signal while there are still a few volts between them. That is the amplifier's Vclip voltage, under those conditions. It varies at least with rail voltage and load impedance, and type of output transistor.
 
Thank you gootee. Have done as you say and Vclip (the difference between the rails and the amp output) is 3V.

Plugging it into the spreadsheet has yielded approx 21000uF per rail for the power I'm after.

Thinking of using 4 * 5600uF per rail per channel. I thinking of not using any film caps at the main reservoir as any benefit is negated by the wiring and their presence would only increase the chances of getting oscillations that would then require snubbers.

The ripple current capabilities will be at 85 degrees 6.34A (*4) = >24A. Enough for my purposes since the SOA protection would be kicking in at that level.

Paul
 
Use GEAR solving method.
Thank you Paul,
please forgive my rudeness, but I must ask, as I already asked elsewhere on this forum with no replay - why Gear? As I understood Helmut's reasoning on keeping Gear in LTspice, it is there simply to maintain look and feel of PSpice (that I'm still fond of), but otherwise a Trap is supposed to be more accurate... or I got it completely wrong ?!?

Thanks
 
Thank you Paul,
please forgive my rudeness, but I must ask, as I already asked elsewhere on this forum with no replay - why Gear? As I understood Helmut's reasoning on keeping Gear in LTspice, it is there simply to maintain look and feel of PSpice (that I'm still fond of), but otherwise a Trap is supposed to be more accurate... or I got it completely wrong ?!?

Thanks

From what I understand TRAP is indeed more accurate but some circuits suffer from convergence problems. GEAR is supposed to be good for power electronics and can help with convergence problems.

I suppose the logic is that an amp is really a power electronics type circuit so the use of GEAR is ok.

I had a long battle with DC operating point problems with this circuit in the past.

If you have problems with solving in this circuit change R21 to either 20K or 22K.
 
See the rather-enlightening image of the LT-Spice simulation plots of cap currents and output voltage, in the post at:

http://www.diyaudio.com/forums/power-supplies/240955-resevoir-capacitors-chip-amps.html#post3599692

This post might also be helpful:

http://www.diyaudio.com/forums/powe...evoir-capacitors-chip-amps-2.html#post3607549
Thanks for these gootee.

I haven't looked at your spreadsheet in detail but you don't seem to have anything for the Power Transformer winding resistance & leakage inductance.

IMLE, this is by far the biggest (and alas most expensive) factor in a simple Power Amp PSU.
 
Makes sense.
I often play with oscillators so skipping the operating point calculation is often a way to go with uncooperative schematics.

I guess I'll have to edit some symbols to fit model names to the missing ones, as I'm short of ECW20N20b, ECW20P20b, and PMBD914

Was using the oic command for a while with parasitics simulation.

For PMB914 just substitute surface mount version of 1N4148 (MMSD4148 - I think). It works almost exactly the same.

The MOSFETS to use from the exicon file are the ECF20x20. You may be able to substitute the 2SK1056 + complement. There are Bob Cordell models for these. Haven't looked at this in detail. You may have to adjust the resistive bias for 150mA bias per device.

All these things I forget...
 
Thanks for these gootee.

I haven't looked at your spreadsheet in detail but you don't seem to have anything for the Power Transformer winding resistance & leakage inductance.

IMLE, this is by far the biggest (and alas most expensive) factor in a simple Power Amp PSU.

That is correct. In that spreadsheet, I did not include any effects of the power transformer windings' leakage inductances or resistances. Most people don't have those numbers and, with that speadsheet, I wanted to try to provide a simplified method to estimate the minimum required capacitance to prevent clipping, for a given rated max ouput power.

I have also posted another spreadsheet, which includes a scalable transformer model, including the leakage inductances and the resistances, which solves the differential equations for the circuit including transformer, rectifier, reservoir capacitance, and a constant-current through an amplifier with a resistive load. There is a link to that in Post 82, at

http://www.diyaudio.com/forums/soli...n-attempt-number-2-simpler-9.html#post3532280

Cheers,

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