Patchwork Reloaded: Circuit Optimization and Board Layout.

[AndrewT]

deleted

[MJL21193]

Originally posted by Nico Ras

I am not slating the design, please don't be offended. If you have a real amplifier that performs quite differently from the sims, why even bother with sims obviously all the other data will be equally useless such as low THD, Slew, PSR, etc.

Hi Nico,
If it weren't for the simulator, we wouldn't be having this conversation. I wouldn't have any amp design or redesign, I wouldn't know diddly squat about any of this.
The amps that I have built completely agree with the simulations, right down to the measured voltage and current. I have been diligent about using accurate models in the sim to give the best prediction. Some things may be slightly off, but not by much.

I know there was some confusion and that you were under the impression that I was still using the revised mirror. I have thrown out that idea and will stick with the standard mirror.

Originally posted by Nico Ras

I have looked at the 4.7 Ohm 100 uF combination, what are you filtering 100 Hz ripple? This filter will do nothing.

My simulation and real world results say otherwise. This filter works to improve PSRR at higher frequencies. Increasing the resistance to 22 ohms and capacitor to 220uF will make it effective at a lower frequency. I suggest you look at this in Microcap. It should be very apparent on a PSR plot.

Originally posted by Nico Ras

If the 200K makes a slight difference on the THD, even in the sim it is fractional considering that the amp performance is already at 0.003%, in the real amp it probably is non existent, why 200k, what about 150K or 220K.

The bootstrap resistor value has been optimised for the last version of this amp. I can see if it needs adjustment for this version.
It has a positive effect, not a negative one. It lowers THD at high frequency. Whether this is a simulation only result is unclear. I would sorely love to have a distortion analyzer to verify these things. Maybe in the new year.

Originally posted by Nico Ras

Anyway, I think as it stands now this is a pretty nice amp and may be considered by other DIY enthusiasts as a very nice project.

Thank you Nico.

[MJL21193]

Originally posted by Nico Ras
have you considered a capacitance multiplier, or what about a series shunt regulator. It may give you that 100dB PSR.

Hi Nico,
I'm not looking for PSRR that low.
Others are, but I believe the best way to accomplish this is to improve the power supply. Leave the amp itself as simple as possible.

Originally posted by homemodder
Hi MJL21193

I was wondering how that ccs mirror performed. When I saw you apply it my immediate thought was that the ltp would be unbalanced.

Hi homemodder,
Unbalanced would seem to be the intention of the original design.
I don't know if it will bring the improvements that I saw in a real amp circuit, I don't know if it will have a negative impact on PSRR in a real amp circuit. If I had a distortion analyzer I'd take the time to add it to a working amp and see what the actual results would be. Until then, I'll forget about it.
IF it does as it shows in my sim, it is worth pursuing. The PSR can be dealt with and the DC offset can be handled with a simple pot of even a servo.

[MJL21193]

Originally posted by MJL21193

My simulation and real world results say otherwise. This filter works to improve PSRR at higher frequencies. Increasing the resistance to 22 ohms and capacitor to 220uF will make it effective at a lower frequency. I suggest you look at this in Microcap. It should be very apparent on a PSR plot.

I have ran the sim with a few different values for the RC filter:



The green trace is with no filter. The yellow is with 4R7 on the rails with 100uF.
The blue is 10R on the rails with 220uF and the red is 22R on the rails and 220uF.

[Nico Ras]

I don't have to look at Microcap, you can simply calculate that the -3dB point of 4.7 Ohms and 100uF just does not filter much 100Hz.

F=1/(2xPIxRxC), -3dB is at 338 Hz, why would you want to filter at this frequency.

[Nico Ras]

22 Ohms and 220uF attenuates the 100 Hz by 10 dB, a little better.

[danielwritesbac]

And, what happens if you model 2.2uF from V+ to V- at the amplifier board?

EDIT: 3.3R+perfect cap+3.3R = 250v polyester cap (as used in speakers), wheras 0.8R+perfect cap+0.8R = decent quality 250v ecap.

[MJL21193]

Originally posted by Nico Ras
22 Ohms and 220uF attenuates the 100 Hz by 10 dB, a little better.

Hi Nico,
Are you only looking at one frequency? Did you have a look at the plot I posted?

Originally posted by danielwritesbac
And, what happens if you model 2.2uF from V+ to V- at the amplifier board?

Hey Dan,
Not much, I think.

[danielwritesbac]

Originally posted by MJL21193
Hey Dan,
Not much, I think.

Then it can be a nice surprise. I updated the post for more specific information that just came out of the Dick Smith ESR meter when I measured the caps. Its still an estimate, though.

[MJL21193]

Originally posted by danielwritesbac


Then it can be a nice surprise. I updated the post for more specific information that just came out of the Dick Smith ESR meter when I measured the caps. Its still an estimate, though.

You're slipping again Dan! Didn't you take your meds today?

We are talking about power supply ripple rejection of the amp circuit. The resistors and caps form a filter to drive the effects of this ripple from the power supply down in audibility.