for your work 
Maybe I spoke a little too soon. The guys in Santa Clara might say otherwise about the OP275 but as for the AD797 it was low noise data acq. and instrumentation. Having Walt Jung on the staff during those years made for all the fun audio apps that we did together. BTW the arguments about BB vs. LTC vs. ADI don't bother me at all, here it's all for enjoyment and leisure.
PDF of Wurcer's/ADI Patent
amazing USPatent and Trademark Office -- I ordered these on Sunday morning and they were in the "Inbox" today -- a legal holiday in the U.S. -- here's a link to the PDF of the patent -- due to the size of the PDF I am only going to keep it up there for a few days, elsewise I will runover over my bandwidth limitation:
http://tech-diy.com/wurcer.pdf
amazing USPatent and Trademark Office -- I ordered these on Sunday morning and they were in the "Inbox" today -- a legal holiday in the U.S. -- here's a link to the PDF of the patent -- due to the size of the PDF I am only going to keep it up there for a few days, elsewise I will runover over my bandwidth limitation:
http://tech-diy.com/wurcer.pdf
transducer said:
Ron,
Sorry for the delay, I am following the forum irregulary these days.
Yes, I would leave out the gnd plane if I had to redo it again. Of course, there may be an effect I am aware of, so don't sue me, but give it a try. The important point is the stringent star grounding and remote sensing as mentioned in the 3rd article of the Super regulator series.
Jan Didden
scott wurcer said:
Scott, hello, nice to meet you here,
I have a question on this. As you are probably aware, we (or at least I) had some trouble getting the version with the 797 stable. How is the neutralizing cap influencing this? Could it be that I have been overlooking the obvious all the time (not the first time, ha).
Jan Didden
peranders said:
Correct. I think I actually made the point in the article. I recommended NOT to use very high quality foil caps at the supply output (those .1 or 1 uF ones everybody seems to insist on) as they promote instability.
In fact, with those low output iompedances, the ONLY purpose of the output caps is to stabilise the supply. Since they are in fact in parallel with the Zout of, as Scott says, 0.000001 Ohms, it is difficult to imagine the output cap to have any role in supplying power at all.
Jan Didden
Folks, I have not followed this thread, so please excuse any oversight. However, with feedback regulated power supplies, it is best NOT to use a really high Q, (good) cap at the output. The reason is that the output looks like a synthetic inductor, BECAUSE the output impedance change with frequency will act exactly like one. Therefore, if you put a really good cap at the output, the Q of the resonant circuit formed will make it ring. This usually is seen as an extra noise, but it could show itself in other ways. SO, either you put a resistor in series with a good cap, or you use a lossy cap. Take your choice.
Last but not least
This has been discussed at length by W Jung, J Carr, AWL, and even myself on the forum. I wonder if an error amp with a wide flat open loop gain would allow the use of a cap whose impedance dominated the output impedance at a frequency range where the regulator still flat. You raise an excellent point and this resonant point is probably the Achilles’ heel for most regulators that have an open loop gain that starts to fall at low audio frequencies. BTW I would vote for a low ESR cap and a small value resistor. This might allow the use of a small value cap in parallel whose impedance would be come dominate at a frequency above the point at which low Q cap and regulator output inductance resonance occurs.
AD825 anyone?
This has been discussed at length by W Jung, J Carr, AWL, and even myself on the forum. I wonder if an error amp with a wide flat open loop gain would allow the use of a cap whose impedance dominated the output impedance at a frequency range where the regulator still flat. You raise an excellent point and this resonant point is probably the Achilles’ heel for most regulators that have an open loop gain that starts to fall at low audio frequencies. BTW I would vote for a low ESR cap and a small value resistor. This might allow the use of a small value cap in parallel whose impedance would be come dominate at a frequency above the point at which low Q cap and regulator output inductance resonance occurs.
AD825 anyone?
Per-Anders, you need to get rid of that 0.1 uF cap on the output of your circuit. I know you won't take my word for it, but maybe you'll take Walt's. To quote his article from Audio Electronics at the top right of page 16, "With this compensation network in place on one of Mark Kovach's PCBs, I was able to apply high-Q capacitive loads to the output of the AD817 +/- regulators without oscillations. But, alas, this trick definitely does not work with the AD797".
'Nuff said.
'Nuff said.
I have this need/no need for the capacitor in mind but I am going to test this if/when I make the pcb's. I have great faith in you guys so I'll believe that it's more or less true.andy_c said:
Not a huge demand for SMD pcb's but I would like to test it at least. Maybe a group buy?
Maybe a couple of beta testers are willing to volounteer.
Okay, here's a plot of the impedance vs. frequency of the exact same parts Per-Anders is using (vendor = Kemet, 0805 package, X7R dielectric, 0.1uF). This is from the Kemet web site. You have to register to get this data, so it's kind of a pain.
It's clear that the series resonant frequency is 10 MHz and the ESR is about 0.07 Ohms. So now you have R, L and C. What does this do to stability? Let's make a very optimistic assumption of infinite ft of the pass transistor and very low impedance seen by its base, so that the open-loop output impedance of the regulator is purely resistive, with value 26mV/ILoad. Let's say we're delivering 52 mA. So we have an open-loop output impedance of about 0.5 Ohms resistive. At the series resonant frequency of 10 MHz, the attenuation is -20 * log(.07/(.07 + 0.5)) = 18.2 dB. To be continued...
It's clear that the series resonant frequency is 10 MHz and the ESR is about 0.07 Ohms. So now you have R, L and C. What does this do to stability? Let's make a very optimistic assumption of infinite ft of the pass transistor and very low impedance seen by its base, so that the open-loop output impedance of the regulator is purely resistive, with value 26mV/ILoad. Let's say we're delivering 52 mA. So we have an open-loop output impedance of about 0.5 Ohms resistive. At the series resonant frequency of 10 MHz, the attenuation is -20 * log(.07/(.07 + 0.5)) = 18.2 dB. To be continued...
Attachments
So what does this overly optimistic analysis tell us? Well here's an amplitude and phase plot of the simple RLC circuit showing the roughly 18 dB of attenuation at 10 MHz. Now if you have the AD797 data sheet handy, look at figure 14, open loop gain and phase vs. frequency. We have unity feedback because of the capacitor bypassing the feedback resistor. If we look somewhat below 10 MHz, we see in the plot below that there's about 50 degrees additional phase lag. This is also right where the phase of the AD797 starts to look squirrely too. So eyeballing it a bit, it looks like the unity loop gain frequency will be about 7 MHz and the phase margin will be about 10-20 degrees. Not good.
Attachments
If I may......
" Would the realistic Ft of the pass xsistor (instead of the infinite one) not actually improve matters, because the open loop gain will be less?"
That is an excellent question. Actual it is very likely that addition phase shift will make things even worse. Instead of a 6db per octave roll off per octave we now have a 12db per octave roll off per octave. Phase margin is a figure of merit over how far into the 12db per octave rate when the open loop response is at unity gain. You have oscillator when the phase shift is 180 degree in the region were the open loop is still greater than one. You don't have to be at the point where the circuit is an oscillator before the trouble begins. The closer you come to the 180 degree phase shift the more over shoot the circuit has. Overshoot is not a desirable quality in amplifiers or power supplies; which are essentially amplifiers with big DC offset.
" Would the realistic Ft of the pass xsistor (instead of the infinite one) not actually improve matters, because the open loop gain will be less?"
That is an excellent question. Actual it is very likely that addition phase shift will make things even worse. Instead of a 6db per octave roll off per octave we now have a 12db per octave roll off per octave. Phase margin is a figure of merit over how far into the 12db per octave rate when the open loop response is at unity gain. You have oscillator when the phase shift is 180 degree in the region were the open loop is still greater than one. You don't have to be at the point where the circuit is an oscillator before the trouble begins. The closer you come to the 180 degree phase shift the more over shoot the circuit has. Overshoot is not a desirable quality in amplifiers or power supplies; which are essentially amplifiers with big DC offset.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.