Quote:
Originally Posted by astral highway
But when I’ve been looking at a pair of low and high side drivers together, I’ve blown up a few instantaneously, suggesting a shorted output that was not apparent. They’re pricey little things so I don’t want to repeat this finding.
|
Looking at the low side is generally easy, because typically your power supply's ground reference is sufficiently close to the scope's ground reference that no noticeable current flows when you clip on the scope ground and everything carries on working. The tricky part is the high side, as you've observed.
The first way I'd try to do this is to make sure the power supply I'm testing is not connected to ground. Run it from a battery, or from a power supply whose output isn't grounded. Most lab power supplies are like this. Make sure nothing on your power supply is grounded, anywhere. Then you can clip your scope ground wherever you like and make measurements relative to that reference point. This works well if your reference point is a pretty low-impedance one that doesn't move much, such as the positive supply rail. However, if your reference point is something like the opposite gate drive or a bootstrap rail which has a lot of fast edges on it or a relatively high impedance, you may find that grounding it causes trouble. This method also means you can't make more than one measurement simultaneously with the same scope with different reference points.
Sometimes it's worth trying to use the two scope inputs in 'A-B' mode as a differential amplifier. How well this works depends on the scope. Some have a good common mode rejection ratio like this, others don't. Seeing the scope in the photos, I remembered an experience I had with one of my clients, doing some power supply design. They had a very similar GWInstek scope, and using the 'A-B' mode was a total disaster. It turned out that the common-mode range was limited to the screen height, so it was impossible to get accurate traces on which we could actually see anything. This is a pitfall of modern digital scopes which don't have 'proper' analogue input stages and try to do everything with digital maths. I went home and fetched an analogue scope to do the job with.
The best way of tackling such measurements is traditionally a differential amplifier. I have a couple of Tektronix 7A13 modules which are outstanding: 100+MHz bandwidth, good CMRR, and for most practical purposes a common-mode range of several hundred volts. I don't know what their modern equivalent is, sadly, though the Powerscout-type isolated scopes work very well. They exist as both CRT and LCD instruments.
Incidentally, a good way of getting really accurate views of fast pulse waveforms such as gate drives is to use your scope's 50 ohm input termination, if it has one, then use a piece of thin 50 ohm coax (top tip: there are lots of cheap cables on eBay made with RG174 wire or similar. They're great for hacking apart for this sort of thing) from there to the point you want to measure. Strip the end quite short, so less than a centimetre of braid and centre core are showing. Solder the braid on to your ground point and solder a surface-mount resistor of a few hundred ohms or so on to the point you want to measure, then to the centre core of your coax. Hey presto, an instant and cheap probe which will give you accurate results up to several hundred megahertz, without the vagaries of dangly probe ground wires and compensation capacitors. This doesn't solve the ground problem, though.
Chris