Super Regulator, collecting the facts

[janneman]

Originally Posted by peranders

Can't you block the DC with a cap?

Yes that's what I do. AP output > cap > reg output.
The cap must be pretty big if you want to measure accurately to low freq, but a 1000uF 25V electrolytic is cheap & probably in your junk box anyway. We're not interested here in super-duper caps anyway.

Alternatively, you could modify the AP zout measurements script to measure Vgen before the cap to calculate Iout, and with the other channel measure signal at the reg output and calculate Zout.

jd

[ikoflexer]

Guys, may I have your attention for a bit? I don't have access to an AP unit but have an old HP 3585A spectrum analyzer. It doesn't have two channels, so I can forget about phase measurements. It would be nice though to at least get a measurement of the output impedance magnitude.

This is what I have in mind. The 3585A has this feature that one trace can be saved in memory and then subtract it from a new trace. Say, measure trace A, the AC voltage across a one ohm resistor, so that this is actually the AC current induced in the load by the tracking generator, which is coupled to the load via a power amp (truepath TA2024 class-d little amp) and capacitor. Then save this to memory as trace B. Then measure a new trace A as the AC voltage component across the load. Since the values of A and B are logarithmic, their difference is actually voltage/current division, the "impedance." I'm thinking that a 3Hz resolution BW, while not ideal, is adequate.

I've tried it and it works indeed, meaning that I was able to get a reading for (A-B) for a lm317 and a 7812. But I'm just not sure that my interpretation of the final (A-B) result is correct. The nice feature is that after the cables are hooked up the actual measurement is very easy and fast to do, and in the case of this spectrum analyzer, it can be run to well beyond audio frequency.

The attached drawing shows the setup.

Any comments are very welcome.

[janneman]

Originally Posted by ikoflexer

Guys, may I have your attention for a bit? [snipped to save bandwidth]
The attached drawing shows the setup.

Any comments are very welcome.

Nice idea!
I think what you want to measure as current is the current through a resistor feeding the amp output into the reg output, rather than the current through the resistor as shown. You want to know the current that is 'absorbed' by the reg (the virtual Zout).
You can do that by measuring the generator (amp) Vout and divide by the serie resistor, as the Vac at the reg output will be much lower than the amp Vout so the error is very low. Say you set Vout at 10V, Rseries at 100 ohms. Neglecting the low (mV's) of ac volts at the reg output, Iout is 100mA ac. Store this trace in dB.
Then measure the ac volts at the reg output, store as trace in dB, and subtract the two traces.
There's one problem I see, and that is that dB are a ratio, not an absolute value. The difference curve will show you how Zout varies with freq, but not the actual value. So you need to do an absolute measurement at a spot freq to calibrate your setup.
Say your ac signal at the reg output, at 1kHz, is 1mV. You know that the Zout there is 10 milliohms. So you now know that the point where your difference curve goes through 1kHz is actually 10 milliohms. The rest is easy

jd

[ikoflexer]

Jan, thanks, I'll try your suggestion. Regarding the dB though, the HP can give dBV, which is relative to 1V (it can also show dBm). That should be calibrated already, no?

Unfortunately I don't have a HP-IB card, as they seem to go for some big $$ even used on eBay. Would have loved to be able to get all this done from a PC, and then do any post processing easily.

[janneman]

Originally Posted by ikoflexer

Jan, thanks, I'll try your suggestion. Regarding the dB though, the HP can give dBV, which is relative to 1V (it can also show dBm). That should be calibrated already, no?

Unfortunately I don't have a HP-IB card, as they seem to go for some big $$ even used on eBay. Would have loved to be able to get all this done from a PC, and then do any post processing easily.

Yes that should help, not sure about the actual scale settings available. Can you chnage scale legends?

Check out the USB HPIB adpater from Prologix, LLC || Home great value for 150 $ or so!

jd

[ikoflexer]

Originally Posted by janneman

Yes that should help, not sure about the actual scale settings available. Can you change scale legends?

Check out the USB HPIB adpater from Prologix, LLC || Home great value for 150 $ or so!
jd

Yes, the display can be changed to V, dBV, dB, and dBm. The reference can also be changed. But the trace on the screen stays in the log domain. The marker only shows different unit values. Of course if I could extract the values into a PC the rest is easy as 1,2,3.

The GPIB-USB seems very nice, so I could do everything from my laptop, that'd be great.

The other thing about this setup is that I could insert in the loop a low noise amp with 20/40/80 dB.

[ikoflexer]

I'm very tempted to put one of these together in a hurry. I got all the parts, except for the GPIB cable and socket, which are probably going to cost more than the rest.

http://www.webalice.it/hotwater/PicPlot.htm

Edit: AND they have a USB version too. I'm drooling...

http://www.webalice.it/hotwater/USBpicplot.htm

[janneman]

Originally Posted by ikoflexer

I'm very tempted to put one of these together in a hurry. I got all the parts, except for the GPIB cable and socket, which are probably going to cost more than the rest.

http://www.webalice.it/hotwater/PicPlot.htm

Edit: AND they have a USB version too. I'm drooling...

http://www.webalice.it/hotwater/USBpicplot.htm

In that case you'd also want Ulrich Bangert's EZ-GPIB utility.

Downloads

jd

[ikoflexer]

Good one! Another one that seems useful might be the open-source GPIB toolkit. As soon as I get the circuit working I'll report what happened.

[jackinnj]

You had sent me the link to the AP Technical Note quite a while ago -- wasn't till I pulled it down that I realized I had it already -- I'll have to rewrite the macro a tiny bit.

There's a lot of HPIB code floating around -- put "KE5FX GPIB" into google!