7055-C giving low USB voltage even with AUX

[mete]

I am using a QA403 audio analyzer (USB 2) and it reports USB voltage and current (and warns if USB voltage is too low). With my PC, USB voltage is around 4.7V and the current between 0.8-0.9A. I see a very minor 50Hz component when looking at the FFT of its output and I wondered if this can be eliminated with 7055-C. I tried connecting it in various ways and giving AUX power with different things (USB hub, USB charger, USB battery etc.) but QA403 always reports the USB voltage below 4.5V when 7055-C is in between. Is this an expected behavior ?

[Daniel (Intona)]

Do you hear an audible click (internal relay) when supplied with AUX power?

[mete]

I believe I did, it is not connected now so I cant check. Does the operation of relay indicate the AUX power being used ?

[Daniel (Intona)]

Yes it does. 4.5V is reasonable if wires getting long (USB devices should work down to 4.2V).

When I look at [External Link Removed for Guests], there is already a built-in USB isolation. And on the screenshot, I can see a 50 Hz deflection. Maybe this is some capacitive coupling into the measurement wiring?

[mete]

Probably it is OK for normal USB devices but QA403 specifically requires I think >4.6V (and ~900mA). It still works with less than this but it is not recommend, I guess it is not designed or tested for that.

Meanwhile I tested a few cables and there is a large difference. The cable that works the best for me is actually the long one (probably 5m, Lindy brand not something special), I guess being long it is better constructed than others. It gives 4.65V and connected to a powered hub. I will get a shorter 20awg cable and use a USB hub near the device. I have to keep QA403 in a distance, so I need a long cable either for the device or for the hub.

Yes, the audio section is isolated. I was trying to minimize the influence of mains so I wanted to try also the USB isolator but it seems the issue is mostly about connection to DUT or DUT itself. It seems not very obvious to eliminate this but I have quite minimized it by changing the location of things around and the cables and connectors. There is no mains component when looking at the noise floor of the device (attached image with no signal) or when it is operated in a loopback (attached image with 1kHz signal).

[Daniel (Intona)]

I see the device has + and - inputs on separate BNC connectors. This looks quite uncommon to me. Are you using both + and - inputs? If not, I'd try to build an adapter that tightly connects a twisted pair cable (cat5 or so) to +/- and the screen to both BNC GND. If you measure unbalanced, tie the minus wire on DUTs GND but with a resistor in series. The resistor should have the same value as the DUT's output resistor. If it even doesn't have one (which is odd), add 47R right before going into the cable. This way, you get a real balanced line (even when you DUT has unbalanced output) which is quite immune to 50Hz coupling.