A scope 100MHz or better set to a high sensitivity is usually enough to track a source of ingress. Moving a probe (ungrounded to the dut) around the circuit can lead to the source of ingress. If you can nail the frequency of the ingress then you can identify the source on a board. The amplitude will become greater as the source is approached. One can do the same if the source is external and nearby.
I found my cable modem puts out a near 15MHz burst tone, 30mVpp at distance of 18".
The burst was causing false triggering of the zero crossing detector (comparator) in my oscillator. Additional hysteresis on the comparator fixed the problem. The point being we have a lot more RF source around us these days than ever before and it's not going to get any better. Our equipment must be able to tolerate and operate in this kind of noisy environment.
I found my cable modem puts out a near 15MHz burst tone, 30mVpp at distance of 18".
The burst was causing false triggering of the zero crossing detector (comparator) in my oscillator. Additional hysteresis on the comparator fixed the problem. The point being we have a lot more RF source around us these days than ever before and it's not going to get any better. Our equipment must be able to tolerate and operate in this kind of noisy environment.
I can't find my RF sniffer, but the scope probe did show that it's the USB -- I tried my ADUM4160 Isolator, but couldn't get them to talk.
I also loaded up the USB cable with various ferrites and toroidal cores -- enough to prevent any data from going up or down the pipe!
I think the QA400 uses the USB high speed which is why the ADUM4160 Isolator won't work.
So were back to the USB problem.
There are jumpers to move between high and low speeds -- didn't work on either.
I thought the isolator worked in 1.5 Mbit/s "Low Speed" and 12 Mbit/s "Full Speed.
High speed is 460Mbit/s for USB 2.0. Dick Moore bought one to try with the EMU0204.
The EMU0204 uses high speed. The isolator wouldn't work with the 0204.
I have Dick's isolator here. I'll have a look.
Data sheet says this "Low and full speed data rate: 1.5 Mbps and 12 Mbps".
Is this what you have ADuM3160?
Correction High speed is 480Mbits/s.
ADuM4160 is the same.
Last edited:
I can point you to a lot of good software work on a USB UAC2 interface. Making it work on USB3 would require an SOC with a USB3 interface. Otherwise it's way too much work.
A USB 2 isolator does exist that will handle High Speed; USB Extension | Icron Technologies I have used this USB 2.0 Ranger® 2204 - Icron Technologies Corp. with the EMU 0404 and 0202 and Praxis successfully over 100' of cat 5. The fiber versions should offer a lot of isolation, but they also bring a power supply and local noise generators. And they are $1,000. You can be a really fine laptop for that and get better isolation running from the battery with Wifi for connections. I pushed my Analog Devices rep hard on USB high speed and he had no plans or roadmap for one. Because of the mixed signalling its a difficult problem to solve.
A USB 2 isolator does exist that will handle High Speed; USB Extension | Icron Technologies I have used this USB 2.0 Ranger® 2204 - Icron Technologies Corp. with the EMU 0404 and 0202 and Praxis successfully over 100' of cat 5. The fiber versions should offer a lot of isolation, but they also bring a power supply and local noise generators. And they are $1,000. You can be a really fine laptop for that and get better isolation running from the battery with Wifi for connections. I pushed my Analog Devices rep hard on USB high speed and he had no plans or roadmap for one. Because of the mixed signalling its a difficult problem to solve.
Data lines? Are you talking about USB? USB needs to be close or it radiates enormously. The power supply noise won't affect the data. If its really bad you will get data loss not bad data. The protocol is pretty robust.
The noise is most likely switching supplies coupling through the input ground. Way back, early in this thread I have some pictures showing how low the noise can be if the computer (laptop) is on battery power. Even with that the leakage from my hand on a plastic part of the pc would show in the plots. It only takes a little common mode leakage.
The noise is most likely switching supplies coupling through the input ground. Way back, early in this thread I have some pictures showing how low the noise can be if the computer (laptop) is on battery power. Even with that the leakage from my hand on a plastic part of the pc would show in the plots. It only takes a little common mode leakage.
yes, USB. Its usually bad practice to have switching supply lines bundled with data... analog or digital. There are two data wires/pins and two USB pins for power. The shield may be connected to power... in which case it could have one end of shield connected and a separate wire for power return.
I am primarily referring to that 100KHz noise though.... its a fixed or limited freq range which something might be operating at (QA400 PS?). It is computer independant... mine has the same 100KHz noise (on cpu battery power only).
Otherwise its 100KHz noise is from QA400 processing pushed up there.
Thx-Richard
I am primarily referring to that 100KHz noise though.... its a fixed or limited freq range which something might be operating at (QA400 PS?). It is computer independant... mine has the same 100KHz noise (on cpu battery power only).
Otherwise its 100KHz noise is from QA400 processing pushed up there.
Thx-Richard
Last edited:
The existing linear circuitry in the QA400 seems to run off a very low voltage -- perhaps 2.4V. I didn't probe around a lot as the fluke leads aren't really designed for SMTs.
Since we're dealing with an instrument with limited bandwidth (the AP will perform FFT out to 103kHz, the QA is 20kHz) it might be useful to transformer couple the outputs to protect them from DC -- AP does this. Then you could also select the output impedance with insertion of a resistor. The outputs of the QA400 aren't happy with low-ish "Z" loads and a compound amplifier around the LME49710/LME49600 would help.
The inputs really need protection. If you're going to build a new low noise, low distortion front end, a low noise power supply has to go with it, especially if clamp diodes are going to be used. I liked the Boonton 1120 front end's protection scheme, but no longer have access to that manual.
As I've blown up many front ends measuring power supply noise and Zout, I would suggest a "sacrificial" socketed opamp, clamp diodes, a bit of RFI protection (and in my case, the ability to "pre-charge" the input coupling capacitor.)
(I note that new Jung regulator boards are available from the Diyaudio store.)
Since we're dealing with an instrument with limited bandwidth (the AP will perform FFT out to 103kHz, the QA is 20kHz) it might be useful to transformer couple the outputs to protect them from DC -- AP does this. Then you could also select the output impedance with insertion of a resistor. The outputs of the QA400 aren't happy with low-ish "Z" loads and a compound amplifier around the LME49710/LME49600 would help.
The inputs really need protection. If you're going to build a new low noise, low distortion front end, a low noise power supply has to go with it, especially if clamp diodes are going to be used. I liked the Boonton 1120 front end's protection scheme, but no longer have access to that manual.
As I've blown up many front ends measuring power supply noise and Zout, I would suggest a "sacrificial" socketed opamp, clamp diodes, a bit of RFI protection (and in my case, the ability to "pre-charge" the input coupling capacitor.)
(I note that new Jung regulator boards are available from the Diyaudio store.)
These board accomodate both SOIC and DIL for the opamps, and have also provisions for a remote sense low-pass to kill any oscillation tendency with remote sense.
One lives and learns
Jan
You know, I have a spare board with that circuitry. I'm itching to add a small micro and turn it into an autoranging sound card interface...
jan
The signal around 100 kHz is very likely a only partially supressed image of the audio signal (at Nyquist mirrored information, a natural process in DA conversion). Could be improved with a better anti-image filter, but should not affect most measurements anyway.
Samuel