post-trigger and external clocks

[shoneill]

I'm desperately looking for some help on an issue I've never had to deal with before.

I need to read in a video out signal from a linear CMOS sensor.  I have a start pulse (ST), a stop pulse (EOS), a trigger indicating the correct time to read (Trig) and the video signal itself (Video) which is an analog signal.

Looking at the scheme shown below I figured some post-trigger acquisition at the frequency of the base clock would suffice, using the ST and EOS signals to set the scan boundaries.  However I have one rather large problem.  The sensor comes with an evaluation board which locks the clock at 300 kHz.  The video out signal is a quarter of this (75 kHz).  My card only does 100 kHz (MIO-16XE-10).  I'm assuming that I'm going to get some "fuzzy" signnals from the sensor as my card is basically too slow, but what I want to know is if any efforts to get this up and running on this card are a waste of time.



Secondly, assuming the signal can be recorded (even if only badly), how can I couple the measurement frequency of the DAQ card to the trigger signal of the evaluation board?

I am dreading upgrading the card, as it's kind of hard to find 16-bit cards which still have drivers for LV 6.1.......  The PCI-6035 is the only one I can find.  Maybe I should grab one before they're obsolete!

Shane.

Message Edited by shoneill on 09-19-2006 03:34 PM

Message Edited by shoneill on 09-19-2006 03:35 PM

[Andre_Saller]

I can not recommend using this board to acquire your video-signal. According to the nyquist theorem, you would need AT LEAST 150 kHZ sampling frequency to get a good signal, better if you have even more. Check out this link for details:

http://zone.ni.com/devzone/conceptd.nsf/webmain/0B04C09D4A44C78186256C3F007F8B02

If you would have a board with at least 300 kHz, you could use the clock as an external clock for the analog input directly, thisa would make the application easy, and you would have an oversampling of 4 which should give you a clean enough acquisition of your signal.

Of course you could use a  counter as a freuqnecy devider (by4). Connect your clock to the input of the counter, and route your output (would be a 75 kHz signal) to the external clock input of yourt anaolog input. But you would still have the problem, that 75 kHz sampling-frequency is much to low for a 75kHz Signal.

So there is no way to acquire this signal in a clean and failure-free way with this board, you definitifly should get a board with a faster aquisition-rate.

Hope this helps!

[shoneill]

Thanks,

This is what I had thought, but it's been a few years since I even connected something to a DAQ card.

I need to do a proof-of-concept measurement, so I think the "fuzzy" measurements (Each single point being basically an iteration of an entire video out pulse, rise time and fall time included) would have sufficed.

However, I think I'll simply drop the evaluation board which locks the base clock at 300kHz and connect everything to the DAQ card (Including the base clock signal).  This way I can drop the base clock to something more manageable, say 100kHz.

On another aspect of this:  Can I use the signal labelled "Trig" on the original graphic to connect to the external clock of the AI connections (a 75kHz signal with a duty cycle of 12.5%)?  As previously stated, fuzzy signals are less worrying to me at this point in time as I just need to get the synchronisation up and running.  Once it's running, it's easier to persuade the boss to get a faster card.....

Thanks

Shane.

Message Edited by shoneill on 09-20-2006 10:17 AM

Message Edited by shoneill on 09-20-2006 10:18 AM

[shoneill]

Slight addendum to my original post:

I don't need the video curve, just a distinct value for each plateau.  If the "plateau" includes the entire rise and fall flank, then that's something I can live with at the moment.  In the diagram shown below, each plateau is the same height.  In reality, this isn't so, and I simply need to know how the height varies versus pixel number.

Shane.

[shoneill]

Update:

My proof-of-concept worked.

I used both start and stop triggers and linked the "Trigger" to the scan clock of the DAQ card.  OK, I only got roughly every fourth pixel due to the card being too slow, but I was at least able to get something which resembled a signal which reacted correctly to partial masking of the sensor.

What I now need is a 16-bit 1MHz card supporting coupling digital out to an external clock which has drivers for LV 6.1....

Anyone know of such a card? I believe the new M-series don't have any 6.1 support, is this correct?

Shane.

[shoneill]

I found out myself....

The M-Series cards are only supported by DAQ-mx, which doesn't run with LV 6.1.

Darn.

Looks like we'll have to upgrade LV
or make do with a less useful card

I'll have to wait until my boss is in a good mood to suggest upgrading LV.....

Shane

[shoneill]

Upgraded,

LV 8.20
PCI 6251
DAQmx

Working fine.



Shane.