I have been doing a little bit more testing and trying to figure out the noise and ripple results better..
Main problem is they always seem to have about 20mV noise on the oscilloscope probes even before I start actually taking measurements. So these results are likely a lot worse than they really are..
These are the 5 V noise figures..
- cccccccc.png (4 KiB) Viewed 8403 times
So actually have about 28mV P-P.
This is the ripple figures..
- dddddddd.png (3.83 KiB) Viewed 8403 times
Which are to 20mV P-P. Though the average figures are more around 10mV.
The 12 V rail is to be a little worse for some reason..
- bbbbbbbb.png (4.11 KiB) Viewed 8403 times
Here we have 80mV P-P noise. Ripple is actually about 10mV. But it is difficult to work out if it is just interference on the scope probe is all actually noise on the power rails. But I am still looking into this currently.
Considering this power supply does not have the output ripple filter, it's still doing really well. Plus it has much less output capacitance than my previous power supply design. Thing to bear in mind here is that they will be capacitors on the motherboard as well which will lessen the noise and ripple a lot. So of course again it is important to use quality capacitors on the motherboard.
EDIT:
Oddly the spikes I'm seeing are at 175Hz
can only assume the scope is picking this is interference somewhere.
I put the digital filter setting to filter out anything above 5 MHz on my scope.. And this helped a great deal.. The regulators do not run anywhere near that speed.. In any case, the results seen are a lot better now..
- eeeeeeee.png (3.83 KiB) Viewed 8395 times
Realistically the worst spikes are 40 mV P-P. Majority the ripple is at around 20 mV P-P. This is of course under a 1.2 amp load.. The 5 V line is also around 20 mV P – P and this is under a 2.3Amp load. Where 20mV is 0.02 Volts, just to put it in perspective
Aside from the fact that my measuring techniques suck, the actual results of the power supply should realistically be better than what I am actually measuring.
EDIT2:
I found this in the IC datasheet which I had forgot about..
The device minimizes excessive output overvoltage transients by taking advantage of the overvoltage powergood
comparator. When the regulated output voltage is greater than 106% of the nominal voltage, the
overvoltage comparator is activated, and the high-side MOSFET is turned off and masked from turning on until
the output voltage is lower than 104%.
So based on 12V @ 6% is 0.72V. So the voltage spikes are never going to be higher than 12.72V. Which is part of its protection system
I do not really like the idea of routing mains on the PCB when it is all low-voltage stuff there. Then the other way would be to simply solder the transformer wires direct to the voltage selector switch, but then not everyone would want the switch, and they would still be the problem of where to mount it and how. So I am thinking the voltage selector switch just not going to happen.
