Acquiring a single analog data point using external clock signal for radiation detection

[MIIGEvan]

Hello NI Forums,

We're a group working with PET imaging and are trying to use our PCI-6133 to digitize our signal. The setup is this:

• The pulse from our photodiode is roughly 500ns long with a rise time of about 50ns, we want just the value at the peak of this signal. (It's roughly 100-1kHz, we can control the frequency)
• We have a digital pulse that coincides with this analog one, who's falling edge occurs as the analog signal peaks.
• We were hoping to use the falling edge to acquire 1 sample with the DAQ.
• This is usually done with a peak detect/sample-and-hold style system, we're just trying this out.

Our problem is that when we start acquiring samples, all we get is the noise between events. I haven't used Labview since my undergrad but I'm fairly sure I have the acquisition set up properly, and I've looked at the "Continuous Analog Acquisition with External Clock" examples and read up on the DAQ Manual. We get the same results with signal express and the example. My understanding was that the sample occurred exactly at the clock edge, and that the digital processing used the internal clock (which is why you can't just connect a 10GHz external clock). I've verified on the scope that the edge aligns with the peak, so either some timing issues exist between the PCB and the DAQ or I'm just plain wrong with how the sampling works. 

Any help you can offer is greatly appreciated! I've attached our quick and dirty VI as well.

Thanks very much!

[johnsold]

I cannot see how you have the DAQ Assistant set up, but generally it is not a good way to go if you are trying to maximize performance.

If you set up a continuous acqusition task at the maximum sample rate, do you get good data on the pulses when they occur?  You indicate that the pulses have a duration of about 500 ns, but how is the peak shaped?  The fastest you can sample with the 6133 is about every 400 ns, so if your peak does not last that long, you may not get good data.  I did not dig into the data sheets to see if they specify the aperature time for the 6133.

Lynn

[MIIGEvan]

Whoops, i forgot to mention that we shape the pulses!

So the initial pulse is about a 50ns rise time which instantly begins to decay with a time constant of ~42ns. The useful time to sample the peak there might be 10-20ns.

We do shape the pulse with a bandpass filter, however. This shaped signal has a 100ns rise time and about a 300ns fall time with a more gaussian shape. I've attached a scope reading with the digital pulse present. These signals are coming in at about 1kHz max in our testing environment.

When I get back to the lab I'll try to set up a VI without the DAQ assistant.

Thanks!

Evan

[MIIGEvan]

Ok I just tried again using the Continuous Sampling using External Clock example. Some things that I noticed:

When I switched between rising and falling edge on the clock (we would like to use falling edge) I did see a difference. Rising edge gave samples in the 10mV range, while using the falling edge gave samples in the 50mV range. This is actually pretty close to the ratio between these 2 points in the signals, however, these signals should be around 15x larger. So what I'm thinking is that maybe the problem is impedence mismatch or something on our connections.

We're using the BNC-2110 and all of our connections are made using 50ohm BNC cables. The manual doesn't mention what type of BNC, so we assumed 50Ohm. Looking at our clock signal on the scope while also connected to the PFI 0 pin shows ringing (but not when just using the splitter). This seems to be the source of some issues as well, as we were getting far more sampling errors (attempted acquisition before previous was complete) than we should be based on our count rate. I guess I'll go read manuals and things again!

Thanks for the help/interest so far.

[johnsold]

After looking at your pulses and reading your posts, I think your best option might be to build an external sample and hold circuit. Use the digital pulse to control the s/h with hold starting with the falling edge.  Digitize the output of the s/h at about 10 kS/s and you should be able to get reproducible results and you should be able to compensate for any systematic errors due to the shaping circuits.

The BNC-2110 probably uses 50 ohm BNC connectors but that is not the issue.  The input impedance of the 6133 is 10^7 ohms in parallel with 10 pF.  That impedance mismatch is where the ringing probably occurs.  If your generator can drive it put a 50 ohm terminator on a T-connector at the BNC-2110 so that the line is terminated properly.  This may reduce the pulse amplitude by 50 %.  The mismatch in the cable to the 6133 you will not be able to eliminate.

Lynn

[MIIGEvan]

That is a good suggestion as these sorts of measurements are usually done with a sample and hold circuit or a much faster digitizer; we were just exploring this as a possibility. I'd still like to see if I can make it work, and I'll definitely find some terminators to see if that helps. Will that be enough to stop the ringing on the clock pulse? That seems to be causing at least some of the problems. I guess I can't really comment on the amplitudes until I actually know where I'm getting the sample in time so I'll look into that later.

I'll try a few more things and see how it goes. Thanks for your help Lynn!

I'll leave it open for a bit in case anyone else has some input.

Cheers,

Evan