Power supply performance testing? How To

[Zero Cool]

Thanks Jan!

Zc

[ghemink]

Originally posted by Zero Cool


I have seen these programs but how do you get the signal into the sound card? I would imagine you use the line in of course, but how do you condition the signal from speaker level and DC down to line level without effecting the results??

I have the SIA SIFT SMAART Pro software that is a FFT that i use for sound system setup. and that is a very powerfull FFT, and i have a FFT in soundforge and Wave lab. I did some testing of the crossover in my BMW's 10 channel amp with the SMAART software. that worked well enough to see the crossover slopes etc. but i had no way to calibrate the input levels etc to some reference. Any of these would work well enough to see noise and harmonics etc. but how do i connect a sound card to a power supply?? I am going to have to block the DC with a capacitor of course.

How about giving us a write up of the setup you use etc.

Zc

How about AC coupling it with a cap from the power supply rail to the inputs of your sound card. Probably safest is also to reduce the voltage with a voltage divider after the cap (or before). Are those soundcard inputs protected agains overvoltage? When you switch on the supply, there will be a pretty high voltage at the input of your soundcard, dependent on how fast the power supply ramps up and what the x-over frequency of the RC filter is (due to the C for AC coupling). You can DC couple, but then you need to reduce the signal with voltage dividers and then you loose resolution as you are interested basically in the AC signal that is rising on top of the DC voltage.

By the way, I monitor my power supply rails using a scope, which is of course safer as scopes are supposed to be more robust :-)
Although much more expensive than a soundcard :-(

Best regards

Gertjan

[jackinnj]

Originally posted by janneman

Where do you inject the test current? Either at the point of the regulator where it's internal feedback is taken > you measure the lowest Zout; or at the point where the supply is connected to the load > you measure reality in use. Conclusion: always connect the load to the same point as the supply internal feedback point. You can easily have a 1mOhms supply connected to the load through 10mOhm of wiring. Pretty wastefull.

And oh yes, sweep the test frequency.

Jan Didden

Seems that most power supply "stability" testing is done by inserting a transformer into the error signal chain. The transformer is driven by a sweep generator or network analyzer to derive a plot of gain and phase, the measurement points being the primary of the transformer (the secondary of the transformer usually has a 50 ohm resistor across it so that the impedance matches that of the analyzer) National Semiconductor, Lecroy, Tektronix and others have good application notes on the procedure. Of course, you need a really good transformer.

You could also use a high speed opamp -- Micrel had a small article in EDN about a year ago which I posted on DIYAUDIO. The opamp takes the place of the transformer with some caveats. I think Galinski wrote the article so search unde his name.

Of course, I like to bang my power supplies by rapidly changing the load and seeing what happens. You can do this with a FET switch .

Noise testing -- you don't need an audio-precision tester -- Texas Instruments and Linear Technology will show you how to do it -- if you look at the TI product folder for the TL431 adjustable reference they have a schematic for a noise testing circuit with the correct filters. Walt Jung also described a circuit in the 1995 series of articles in Audio Amateur -- his was based upon the now discontinued SSM2017 from ADI -- the SSM2019 can be used instead and has very low THD.

When you are measuring in the microvolts you have to be very careful of your "probing" -- in fact, rather than probing it seems optimal to solder 50 ohm cable to the points of interest. Best if the DUT is placed in a shielded container (we don't all have Faraday shielded rooms) -- Jim Williams from Linear Tech uses a Danish Butter Cookie tin !

[jackinnj]

Here's a link to Galinski's design note on using an opamp for Bode analysis:
http://www.edn.com/article/CA450603.html?text=galinski

[ghemink]

Originally posted by jackinnj


Seems that most power supply "stability" testing is done by inserting a transformer into the error signal chain. The transformer is driven by a sweep generator or network analyzer to derive a plot of gain and phase, the measurement points being the primary of the transformer (the secondary of the transformer usually has a 50 ohm resistor across it so that the impedance matches that of the analyzer) National Semiconductor, Lecroy, Tektronix and others have good application notes on the procedure. Of course, you need a really good transformer.

You could also use a high speed opamp -- Micrel had a small article in EDN about a year ago which I posted on DIYAUDIO. The opamp takes the place of the transformer with some caveats. I think Galinski wrote the article so search unde his name.

Of course, I like to bang my power supplies by rapidly changing the load and seeing what happens. You can do this with a FET switch .

Noise testing -- you don't need an audio-precision tester -- Texas Instruments and Linear Technology will show you how to do it -- if you look at the TI product folder for the TL431 adjustable reference they have a schematic for a noise testing circuit with the correct filters. Walt Jung also described a circuit in the 1995 series of articles in Audio Amateur -- his was based upon the now discontinued SSM2017 from ADI -- the SSM2019 can be used instead and has very low THD.

When you are measuring in the microvolts you have to be very careful of your "probing" -- in fact, rather than probing it seems optimal to solder 50 ohm cable to the points of interest. Best if the DUT is placed in a shielded container (we don't all have Faraday shielded rooms) -- Jim Williams from Linear Tech uses a Danish Butter Cookie tin !

I just "bang" my power supplies by suddenly changing the resistance of the dummy load by switching manually a mains switch that connects additional dummy loads resistors to the load. This works very well.

You can see what I did in the class D section in the K6 SMPS thread.

[jackinnj]

Originally posted by ghemink



I just "bang" my power supplies by suddenly changing the resistance of the dummy load by switching manually a mains switch that connects additional dummy loads resistors to the load. This works very well.

You can see what I did in the class D section in the K6 SMPS thread.

that's not going to tell you a whole lot about a power supply which may be prone to oscillation -- something which should be well considered in switching amplifiers with a lot of potential for clock noise. but I acknowledge that I use the "larger hammer" method myself.