time to digital conversion in NI 6110

[JHS2020]

I will be configuring the NI 6110 for externally triggered analog to digital conversion. My analog signal is a 50 nanosecond transient (A fluorescence lifetime from a photomultipler). What will be the time to digital conversion upon receiving an external trigger ? What will be the jitter in this time ?  This will help me estimate the average value that the digitizer will produce for each external trigger received. This information is not available in the manual. 

[Kevin_Price]

This isn't an answer to your question, just an additional caution.

Your 6110 has a max sample rate of 5 MHz.  I don't know whether the front end analog triggering circuitry has a higher bandwidth than that, but in the absence of further info I wouldn't assume it.

Thus, I'm not at all sure that the 6110 can be counted on to *detect* a 50 nanosec trigger transient.

I've never worked with PMT's but have been in many discussions here with others who use them.  They typically have some kind of PMT-specific conditioning instrument that can turn the analog transient into a digital pulse.

If you have the means to do that reliably, you'll likely be able to work out the precise timing details you're looking for.   For example, many NI DAQ devices have a minimum (and also default) delay from trigger to sample clock equal to 2 cycles of the timebase used to derive the sample clock.  For now, I'll assume your 6110 is one of them.

    That was a mouthful, so the summary is that your device likely uses its internal 20 MHz timebase, and 2 full cycles would be 100 nanosec.  But because most stuff down at that level is edge-driven, you might need to account for the random phasing of when the trigger edge arrives relative to the timebase edges.   That's pretty much a +/- 1 cycle uncertainty, so I'd expect the delay to be able to vary from 50-150 nanosec to account for the randomness of the phasing.

-Kevin P

[JHS2020]

Kevin, 

Thanks for the reply. 

Will look into options for PMT signal conversion options.

In my case the trigger signal will be 1 MHz, well within the 5 MHz of 6110. 

Also I'm planning on using the incoming trigger as the sampling clock, bypassing the internal timebase. 

Also the analog signal is at 1MHz, due to the 50 ns transient, it has a very low duty cycle. The external trigger/sampling clock is to sample such low duty cycle signals.

-JHS

[Kevin_Price]

I'm not sure I'm interpreting your last message correctly.  Let me describe my interpretation, you can correct me where wrong.

1. There's a PMT signal that produces a 50 nanosec duration pulse.

2. You plan to look into ways to convert this to a TTL pulse.

3. You would then use this pulse as an external sample clock for an AI task.

4. You expect these pulses to arrive at a 1 MHz rate?  (And is that max?  Typical?)

5. Then I'm confused with your comments about 1 MHz analog signal with low duty cycle due to the 50 nanosec transient.

    If you sample the analog PMT signal at 1 MHz, you will not see the transient.  You will get 1 single sample to represent everything that happens over the course of a microsecond.  The transient lasts 5% of that, so you'll usually miss it entirely.  Even the times you happen to land within the transient, you'll only get 1 data point to represent the whole thing.

   Or maybe I misunderstood something.

-Kevin P

[JHS2020]

Kevin, 

PMT signal is an analog signal with a 50 ns transient. 

The transient drops off exponentially and dies out within 50 ns. 

Goal is to get one sampling per transient at the peak value before it drops off.

My questions is if I supply external clock pulses that are precisely timed with the arrival of the PMT transients to the NI card, what would be the sampled voltage. Will the sampled voltage be at the peak of the transient or any where else. Say if there is a jitter/uncertaininty window from the arrival time of clock pulses to when the actual sampling happens then the sampling will happen at random points along the transient.

I just wanted to know if this time jitter/uncertainity window is known for this particular card. Here Im assuming arbitrary precision for the arrival times of the clock signal and PMT signal. What are the limitations of the ADC in 6110?

Thanks

[Kevin_Price]

Ok, now I understand.  But I also don't know the answer.  The best clue I found in the specs was the spec for "interchannel skew" which was stated to be "1 nanosec typical."

I understand this to mean that for the different AI channels that are *supposed* to be simultaneous, the timing skew between their actual sample times can typically be 1 nanosec.  "Typical" makes me think it's more like an average, not a max.

It seems to me that this spec is based on variation in digital signal propagation times through the circuitry, much like the spec you *really* want.   It seems plausible and fairly logical that for the spec *you* want, you're looking at repeatability along 1 signal path.  I'd tend to think that shouldn't be any *larger* than the interchannel skew, and could quite plausibly be smaller.

Strictly speaking though, this is all speculation on my part.  You'd need someone from NI to answer this, and that's gonna be tough to find one of the (likely) very few people who can.

-Kevin P