Without using a tool such as this?
Stability Analysis of Power Supplies with the Bode 100 - YouTube
I already have an oscope, and no problem buying a function generator, but is there a guide to test stability? Basically how to infer a bode plot with just an oscope and a generator.
Stability Analysis of Power Supplies with the Bode 100 - YouTube
I already have an oscope, and no problem buying a function generator, but is there a guide to test stability? Basically how to infer a bode plot with just an oscope and a generator.
The old fashioned way:
One frequency at a time.
Look for the gain of 0 and then look at the phase.
OR
Try using a sound card and small amplifier. There are cheap (or free) programs available (haven't needed one yet so I can't recommend any) that will do the sweeps and the amplitude / phase plots.
All you need then is an audio transformer and some circuits to play with.
One frequency at a time.
Look for the gain of 0 and then look at the phase.
OR
Try using a sound card and small amplifier. There are cheap (or free) programs available (haven't needed one yet so I can't recommend any) that will do the sweeps and the amplitude / phase plots.
All you need then is an audio transformer and some circuits to play with.
The usual way to find out if a power supply is stable (we are talking about a regulated PSU here with a feedback loop of some sort I assume) is to apply a switched load and look at the output with a scope. Recovery from a load transients should be clean and asymptotic with no ringing. If you have ringing, you can assume the phase margin is inadequate and the ULGF loop gain slope approaching 40dB/decade. Why don't you post your circuit so we can have a discussion and point you in the right direction . . .
The rough-and-ready way to test stability is with a load transient response. This can be as crude as whacking a load resistor across the supply output terminals with the oscilloscope set to "norm" trigger. If you get an overdamped response, you're golden - if you get a lot of ringing, it's time to sharpen your pencil.
A simple variable load is an amplifier, remove any local bypass capacitors use a resistive load on the amplifier and drive it with whatever nasty signal comes to mind, 50 Hz square wave is a good place to start. Put an AC coupled oscilloscope between ground and either supply rail of the power supply output. The posts above describe what to look for in transient response.
There is no need to test at all frequencies, square wave has plenty of harmonics and it is the usual way to test regulation.
There is no need to test at all frequencies, square wave has plenty of harmonics and it is the usual way to test regulation.
I'm also looking for some circuit for building a power supply performance test, mainly 2: 1) Loading test, 2) Ripple test. Does anyone like solution for that?
Last edited:
This is probably the simplest and doesn't even require a function generator. How effective is this really, meaning is there a way I can miss instability if I see no ringing at light load and full load?
If you test at the extremes of the load range, the Stanley-stoneage expedient of simply knocking a resistor across the output terminals acutually forces the loop to slew over a fairly wide range to accommodate the load transient. This is more likely to smoke out potential problems than a well-mannered small-signal loop response measurement. In the course of preparing an engineering report on an SMPS, I do both techniques, as most potential customers want to see numbers on gain crossover frequency, phase margin, and gain margin. This takes several thousand dollars worth of kit to measure. A resistor is much cheaper, and tells you pretty much what you need to know if you're not obsessed by numbers and the need for CYA.
- Status
- Not open for further replies.