Noise from DAQ signals sent through a 2532 switch

[gimNPC]

Hello,

I may have a hopeless case here, but it's worth a shot....

I am trying to send DAQ analog voltage time traces through a 2532b switch, and then to my DUT. When I measure the DUT without the switch, my noise level is very low. When I route signals through the switch to the DUT, the noise is increased (as expected) but to a degree that makes this system totally useless for my application. Is the 2532 not designed to route analog signals? Can I hope to send analog signals through my switch with say a doubling in noise instead of a 15x increase in noise?

I have posted a schematic of how my system is configured. Note that wherever possible, I am using BNCs. I even threaded the 2532 ribbon cables through a 1/2 in. tinned copper braid, so the system is about as well shielded as is possible with this hardware. I've also posted signal sent to the DUT (in this case a dummy 1Mohm resistor) using a 1Mhz sampling rate.

Thanks, everyone!

[Frank-L]

gimNPC,

Can you better explain how the DUT is being measured? Is the DAQ card outputting a voltage or measuring a voltage (or both)? The diagram is hard to interpret where the dut is actually being measured. Can you elaborate more on this?

Is the switch breakout box shielded?

Are there any noise sources nearby?

[gimNPC]

Hi Frank,

Sorry I should have included more information. The DAQ is outputting a voltage signal, for example a 1Hz 1V sin wave. This signal serves as the external voltage source for a potentiostat, such that the potentiostat sources the signal it receives from the DAQ. The potentistat has hi and low leads (working electrode and counter electrode) which are routed through the switch before going to the DUT. The potentiostat then measures the actual applied voltage and the measured current and outputs them as voltage signals which are fed directly into the DAQ.

The breakout boxes are all shielded and so are the cables. The DUT is in a faraday cage.

As far as noise sources, the building ground is not terrific but it is pretty good. 60hz noise is present, but the fft spectrum of the noisy signal also contains prominent signals at 80kHz and higher. The potentiostat is designed for high bandwidth, low current applicaitons and so has very favorable noise. The PXIe is the only computer on in the room, and all electronics are plugged into a isolated power supply (it think it is generically called a UPS). I've turned off all other equipment in the room including the lights.

Anyways, the noise only shows up when I route the signal from the potentiostat through the switch. When I send the signal from the potentiostat directly to the DUT, the noise is great (see attachments in previous posts). I would blame the extra noise on a ground loop except that as I have set it up, routing the signal through the switch does not introduce any extra paths to ground.

Thanks for taking a look at this!

-Gim

[Frank-L]

Gim,

Thanks for the additional information!

I believe that I now understand how the system is set up.

Are you routing any other signals through the switch during the measurement? If so what voltage and frequency?

Do you have any other signals attached to the switch? If so what voltage and frequency?

What does the top chart in both images mean (the bottom chart is the FFT, is the top chart time domain measurement?)?

Sources of noise we should be aware of:

Thermal EMF: DC voltage offset of ~20-50uV

Channel to channel or Open channel leakage between channels: -85dB to -35dB depending on the frequency and terminal block used.