dimitri said:
Thanks for the advice, but I’ve already been there (google almost does a better job of finding DIYaudio threads than the DIYaudio search engine does). The link given in post 55 of that thread is to a corrupted .pdf file, according to three attempted downloads on my computer.
Attachments
Re: Re: CFB clamp
Hi Jan,
Finally, here it is (for better or worse.
):
www.data-odyssey.nl
or jump directly to the details at:
http://home.tiscali.nl/data.odyssey/page5.html
(sorry for the missing schematic last Friday)
Cheers, Edmond.
janneman said:Hi Edmond,
That is one of the idiosyncracies of the forum world. You ask: anybody interested in such-and-such", and only luke-warm, if any, replies. But if you post your design, we all jump on it. For better or worse.
Jan Didden
Hi Jan,
Finally, here it is (for better or worse.
):www.data-odyssey.nl
or jump directly to the details at:
http://home.tiscali.nl/data.odyssey/page5.html
(sorry for the missing schematic last Friday)
Cheers, Edmond.
Re: Re: Re: CFB clamp
The schematic was easy to find. Look at the link, erase the last bit, and I got your whole list of diagrams. You may want to look into that.
Jan Didden
Edmond Stuart said:
Hi Jan,
Finally, here it is (for better or worse.
www.data-odyssey.nl
or jump directly to the details at:
http://home.tiscali.nl/data.odyssey/page5.html
(sorry for the missing schematic last Friday)
Cheers, Edmond.
The schematic was easy to find. Look at the link, erase the last bit, and I got your whole list of diagrams. You may want to look into that.
Jan Didden
Edmond Stuart said:
You may want to revisit your power supply design:
- R63 defines the CCS current and it's probably around 1mA (you didn't specify the Vcc1 and Vee1 values). This 1mA through Q35/Q38 has to account for the Q39/Q40 base current and the LM329 current. I doubt only 1mA would be on the safe side, LM329 needs a minimum current of 0.6mA. You may choose to add resistors from the LM329 devices to Vcc1/Vee1 and/or increase the CCS current to 3...5mA.
- Even if it's low noise and low impedance, the LM329 would benefit from a 1uF film in parallel. Over 2KHz, the LM329 dynamic impedance increases exponentially.
- It is most likely that the whole PS won't be stable, unless you compensate the loop by throttling the BW of the error amp.
PMP amp
First, this is a current project, not yet finished. So it will certainly leave room for improvements.
R63 defines NOT the CCS current, rather Vbe/R65 ~= 1.4mA of which most of it flows through the LM329.
As for your suggestion: "add resistors from the LM329 devices to Vcc1/Vee1" No, that's not a good idea, as it will spoil the PSRR and the division ratio of R62/R64. BTW, Vcc1/Vee1 = +/- 50V, see the schematic of the OPS.
As for noise, thanks for the hint. I've put 1uF across the LM329s.
As for stability, this PSU is stable, although the step response shows a small overshoot (10%). I've cured this by increasing the BW, see C19 and C20.
Thx for your remarks.
syn08 said:
First, this is a current project, not yet finished. So it will certainly leave room for improvements.
R63 defines NOT the CCS current, rather Vbe/R65 ~= 1.4mA of which most of it flows through the LM329.
As for your suggestion: "add resistors from the LM329 devices to Vcc1/Vee1" No, that's not a good idea, as it will spoil the PSRR and the division ratio of R62/R64. BTW, Vcc1/Vee1 = +/- 50V, see the schematic of the OPS.
As for noise, thanks for the hint. I've put 1uF across the LM329s.
As for stability, this PSU is stable, although the step response shows a small overshoot (10%). I've cured this by increasing the BW, see C19 and C20.
Thx for your remarks.
Re: PMP amp
Correct.
Even after silently changing R65/R66 from 560 ohm to 470 ohm, you are certainly still very careful not to waste a couple of mA's, I'll let Al Gore know about. If the Q39/Q40 base current is so small, why have you chosed those expensive 20W TO220 devices for the series regulators? This is supposed to be a poor man's amp, so a little care with the costs won't harm.
Right. Could you give us some numbers regarding how the input voltage change/ripple will modify the LM329 I/V, and how this would impact the whole power supply PSRR? And how this power supply PSRR would impact the amp performance?
Even the PGP power supply was originally stable in your simulations, isn't it?
But you are right, alternatively, increasing the BW may work as well. Do you think 1GHz bandwidth for this power supply will be enough? As a poor man looking for an amp, I am ready to build it on microstrip technology.
Edmond Stuart said:
Correct.
Even after silently changing R65/R66 from 560 ohm to 470 ohm, you are certainly still very careful not to waste a couple of mA's, I'll let Al Gore know about. If the Q39/Q40 base current is so small, why have you chosed those expensive 20W TO220 devices for the series regulators? This is supposed to be a poor man's amp, so a little care with the costs won't harm.
Right. Could you give us some numbers regarding how the input voltage change/ripple will modify the LM329 I/V, and how this would impact the whole power supply PSRR? And how this power supply PSRR would impact the amp performance?
Even the PGP power supply was originally stable in your simulations, isn't it?
But you are right, alternatively, increasing the BW may work as well. Do you think 1GHz bandwidth for this power supply will be enough? As a poor man looking for an amp, I am ready to build it on microstrip technology.
Re: Re: PMP amp
That's why I hate class-heat.
You're right, TO220 devices are a little overdone (regrettably, my simple simulator lacks a price list). I've replaced them by BD130/140. Are you happy now? BTW, remember, this website is still under construction. The only reason why I made it public in this early stage is that I was forced to do so, because someone else here posted -before his turn- a less optimal variant of CFB-clamp.
The PSRR wrt Vcc2 is -88dB and wrt the output voltage -124dB, simulated at 1kHz.
Wrong! Not investigating a loop stability doesn't mean that I was saying it was stable.
Anyhow, this PSU comprises a simple CFP or Sziklai stage, which are stable, not only according to my sims, but even in real life.
I'd really love if you would build this amplifier too.
Cheers.
PS: I didn't know you could be so funny.
syn08 said:
That's why I hate class-heat.
You're right, TO220 devices are a little overdone (regrettably, my simple simulator lacks a price list). I've replaced them by BD130/140. Are you happy now? BTW, remember, this website is still under construction. The only reason why I made it public in this early stage is that I was forced to do so, because someone else here posted -before his turn- a less optimal variant of CFB-clamp.
The PSRR wrt Vcc2 is -88dB and wrt the output voltage -124dB, simulated at 1kHz.
Wrong! Not investigating a loop stability doesn't mean that I was saying it was stable.
Anyhow, this PSU comprises a simple CFP or Sziklai stage, which are stable, not only according to my sims, but even in real life.
I'd really love if you would build this amplifier too.
Cheers.
PS: I didn't know you could be so funny.
Re: Re: Re: PMP amp
Well, who knows? One problem I am experiencing is that according to all govenment charts I do not qualify as a poor man.
I'm done with the XLR-RCA converter design and PCB (2.8"x1.75" fitting both channels), to the purist's horror, a full IC (LM4562/OPA211) solution with 0.05% precision SMD resistors) and in the process of evaluating the results. As much as I love tweaking discrete amps, sometimes you have to give up and face reality: these bits of silicon offer a level of performance that makes any other approach look like an overkilling expensive adventure. So far, I am unable to measure any XLR-RCA conversion contribution to the amp noise or distortions, as long as the signal is kept under 2V (which is about double of what's required). I will update the web sites ASAP.
Edmond Stuart said:
Well, who knows? One problem I am experiencing is that according to all govenment charts I do not qualify as a poor man.
I'm done with the XLR-RCA converter design and PCB (2.8"x1.75" fitting both channels), to the purist's horror, a full IC (LM4562/OPA211) solution with 0.05% precision SMD resistors) and in the process of evaluating the results. As much as I love tweaking discrete amps, sometimes you have to give up and face reality: these bits of silicon offer a level of performance that makes any other approach look like an overkilling expensive adventure. So far, I am unable to measure any XLR-RCA conversion contribution to the amp noise or distortions, as long as the signal is kept under 2V (which is about double of what's required). I will update the web sites ASAP.
dimitri said:
Probably not, but I don't see any reason why the amp CMMR shouldn't be much better than the source. YMMV, but I'd happily pay $5 for 8 precision resistors rather than spending an aftenoon sorting 1% parts to that tolerance.
OPA211 is available today, OPA2211 (the dual version) will be available soon.