Working on my SMPS circuit tonight. This past weekend I fully instrumented
the circuit with 4 digital voltmeters. I have two that are exactly the
same, cheap Chinese DVM's that appear to give consistent data. I am using
those two to measure the voltage on the input shunt and the output shunt.
This way I know how much current is coming in and out.
Problem began tonight when I couldn't get the output voltage down to 18
volts as it should be. Finally it clicked that the input current was only
60% of what it should be. Used the fourth DVM to check the output voltage.
Big difference.
So I changed batteries in both of the two voltmeters that couldn't agree.
That brought the voltage difference down to just about .4 volts difference.
Next, I grabbed a 9 volt battery and put them both on at the same time.
Better, only a .04 volt difference. So they agree on pure DC. They must
process noisy voltage differently.
So I put them both back on the discharge voltage position of the circuit and
put the oscilloscope probes on too. The meters don't agree by the same .4
volts. The scope shows that there is no real ripple on the output (this is
a two-phase SMPS so it does have much lower ripple, plus my discharge cap is
oversized by a factor of three). The scope does show some significant
"ringing" where the mosfets turn on & off. At the 5uS setting there is no
real area under the ringing trace.
Amazingly, I did not carefully read what the oscilloscope thought was the
correct voltage...Not sure why. In this case it would probably be the most
accurate indication of voltage...
So my guess is one of the DVM's can tolerate the ringing and one can't?
Extreme accuracy is not a huge concern for this part of the project. I
would like to be within .1volt on all my meters for this purpose. I am
currently trying to get the feedback loop tuned. Actual output calibration
will be something I worry about later on.
Thoughts?
Dan Nicoson
