20-08-2017, 10:21 AM
(20-08-2017, 06:51 AM)ppppenguin Wrote: I'm wondering how DMMs achieve a very high input impedance on the mV range while being a standard 10M on all other voltage ranges. I suppose that the basic meter has a very high input impedance from a FET input opamp. Higher voltage ranges need multiplier resistors and so must have a defined impedance. I also wonder how well protected the mV range is against overvoltage. A thousand volts, or the maximum allowed for the meter, applied to any normal voltage or resistance range shouldn't bother the mater at all. Is the Hi Z mV input likely to suffer?
The input travels via a 10M resistance straight to to the IC. Inside the IC, that signal goes straight to an amplifier, but at the input to the amplifier are a bunch of FET switches that are used to connect external resistors to form a potential divider with the 10M, according to the required range. In mV, none of these resistors are engaged, hence the high input impedance. The fact there are two millivolt ranges for both the AN8002 and AN8008 is because of the ability to switch the input amplifier between x1 and x10.
In terms of input protection, you have that 10M in series with the IC, and the spec for the IC states ±2mA max via the body diodes. So that's pretty good. Remember Joe Smith's tests, where the AN8002 survived transients of 2.5kV - these transients were applied to the meter at all settings of the dial apart from the current ranges. The breakdown, when it happened at 3kV, was between the PCB tracks that form the range switch. Dave Jones has subsequently done some similar tests on the AN8008: https://www.youtube.com/watch?v=Ru2rb7bFhag
The input impedance spec is given as "larger than 60MΩ" for the AN8002 (the other Aneng meters don't mention it in the manuals I have). So that fact we're getting results in the GΩ region is pretty nice, but well beyond the spec.
As to humidity, this is low-voltage, so providing there's no condensation, I really doubt it would be a big problem here - certainly not enough to make it fail the 60MΩ spec. It's 60% in my cellar where I've done my tests - but the dew point is 12 degrees, and nothing down here is that cold apart from the occasional glass of beer
Better meters use guard traces and similar techniques to mitigate against leakage on the surface of the PCB (contamination from handling being an obvious issue), but these cheap meters are an interesting study of what can be achieved for minimal cost and effort. The really interesting test will be how they perform long-term. If anything, they might expose other meters as being over-engineered, which is just as bad as under-engineering in a capitalist world.
