If you watch the video you will notice that there is a DC and AC component to the waveform. I have stated many times that the output from this generator is a 220VACrms 60HZ that was full rectified and unfiltered. This normally provides me with a quick check of the AC, DC and AC+DC functions. In this case, the meter can not seem to handle the rectified signal reliably. The 8002 had a similar problem.
Ah - I hadn't seen reference to a DC offset before. That changes things...
I used my arb to get us away from the AC mains so others could perhaps repeat the test in a safe manor. Again I full wave rectified the output of the Arb. Yes, a basic function generator is all you would need. At 400Hz, sine, 10Vrms, full rectified the 8002 will read 10.01VAC. My HP will read 10.002. Switching the 8002 to DC, it reads 20.19 and the HP reads 20.198. Switching back to AC and the 8002 read 10.01. So no problems.
Now lets increase the amplitude to 15VACrms with no other changes. The 8002 reads 15.01AC and the HP reads 15.004. Switch to DC and the 8002 reads 30.62 where the HP reads 30.63. No problem. Now switch the 8002 back to AC and it reads 3.093AC! Of course, the HP still reads 15.004. With the 8002 still in the AC mode, disconnect the leads and reconnect and the meter still reads 3.040. I disconnect again and reconnect, 15.01. Hit and miss.
I'd done similar tests, but none of my function generators create enough voltage to replicate this problem. So I (carefully!) set up a test with a variac+isolation transformer and a bridge rectifier, and can confirm the behaviour - but read until the end
I found that it was absolutely fine with an AC input voltage (to the bridge) of 11V, which gives an AC component after the bridge of 3.8V and a DC component of 9.7V. No problem at all...
But once you go beyond that, the AC reading from the AN8002 and AN860B+ start to fall. Adjusting to give 15V prior to the bridge gave an AC component of 5.1V as measured by my 87V, and 4.3V according to the AN8002. The AN860B+ does a very similar thing. At 30V AC in, the post-bridge AC component is 10.7V, but the AN8002 reports it as just 0.7V!
I might have imagined this, but the AN860B+ did give correct readings initially, but once I switched it to DC and then back to AC, the bug came and hasn't gone away since. It's almost like something was changed by that initial test, as unlikely as that seems.
On the upside, I can report that the AN8008 doesn't have this problem
I wondered if other meters with the same IC could be affected. The BSIDE ADM08A seems fine, as does my UNI-T UT210E - although it has to be said that the DC offset did cause the auto-ranging to take quite a while to sort itself out.
And quite by chance, I discovered that the BM235 also does strange things with this test. With the same 30V AC into the bridge, it gives the expected 10V AC after the bridge. But cycle through DC -> VFD -> AC and it displays 0.6V AC. If you unplug the signal and put it back in again, it does sort itself out, so in practice this is less likely to catch you out if you're probing a circuit. Interestingly, if you turn the meter on and off again with the signal still plugged in, it continues to give incorrect readings.
BUT...
Having said all the above, I wasn't happy about some of the numbers I was getting, and especially as I started to get some inconsistencies after those tests. I wasted a bit of time checking my connectors, etc, and then decided that my large bridge rectifier might be causing problems with the junction capacitance or similar. So I added a 10k resistor to the output of the diode bridge. This improved things a lot.
Now, with 30V AC going in to the bridge, I was getting 26V DC and 13V AC after the bridge (verified with a Fluke 187 and an 87V - I brought out the 187 for the AC+DC mode). And, the AN8002 and AN860B+ behave perfectly. But the BM235 didn't change...
Time to reach for the 'scope. As expected, with no 10k resistor, the waveform is nuts. The screen-grab shows it with 3 multimeters connected to the output, but if you disconnect those, it get even more funky! With the 10k resistor, all is well.
I've no idea what causes the AN8002 and 860B+ to give such strange results without the 10k resistor - it's possible that some unfortunate combination of the waveshape combined with the particular properties of the AC conversion is to blame - but it looks like in a realistic power supply circuit, this particular problem shouldn't rear its head. But certainly, it's worth bearing in mind. And the BM235 behaviour is worthy of further investigation.
Does your setup have any sort of load across the output of the bridge, beyond the meter itself? And does adding a resistor help at your end?
All the best,
Mark