Using +5V power supply output of NI 7833R PCI DAQ card question

[gt7319c]

Can I use the +5V power supply output of a NI 7833R PCI DAQ card to both: (1) supply power to an external CMOS-input logic IC and (2) pull-up the DAQ card's TTL-output voltage levels to the external device's CMOS levels via a pull-up resistor?  Essentially, I'm trying to develop a buffer circuit to attach at the far end of the 1m long RDIO cable I'm using with the card to get a couple more feet of good digital waveform signal integrity before reaching the logic device I'm really interested in communicating with. 

The documentation for the 7833R cautions against shorting the DIO lines directly to power or ground, but the pull-up resistor should take care of that problem if it has a sufficiently large resistance.  However, the documentation also cautions to "never connect the +5V power terminals to analog or digital ground or to any other voltage source on the NI 78xxR device or any other device."  Does "voltage source" in the above quote refer one of the other +5V power terminals on the card, or to any of the analog or digital outputs of the card?  If the later is true, would the pull-up resistor still provide sufficient protection?

[tikir]

Hello,

the question is related to this: 

http://forums.ni.com/ni/board/message?board.id=250&thread.id=49284  

which is still not answered unfortunately. Doesn't exist any sufficient spec out there?

We used the 7388 +5V power to supply three LVPECL translators each driving approximately 2m cables into 50 Ohms, consuming ~150mA in sum. The card should deliver up to 1A. The AWG28 thickness of the power lines in the cable will reduce the available voltage at the cable's end, if a significant current is drawn.

We did not use any pullups, because the card delivers around 3V of logic output level, which is sufficient to drive TTL inputs safely, so one can use HCT or ACT logic for instance. It's not enough for HC of course, so use alternatives. The maximum output current is 4mA, as stated in the spec, but see the measured output impedance in the link above.

A dedicated level translator will be the safer solution than the pullup, especially at high frequencies or steep switching edges.

Regards, Timo 

[gt7319c]

Hello Timo,

Thank you for the quick response.  After some more research I found an AHCT logic quad buffer IC that should do the trick.  The IC I'm trying to drive at the far end of my setup (a laser diode current switch) has a CMOS input, so the AHCT device should work well to translate the TTL signals from the DAQ card to the CMOS levels without using a pull-up resistor.

I am still concerned about proper impedance matching.  Is there any further circuitry I need to implement at the input of the buffer IC given that it will be at the end of the RDIO cable?  How about at the input of the CMOS device I'm trying to drive?  The cable distance between the buffer IC PCB and the laser diode switch will be < 1 m, but I'm trying to preserve as much digital signal integrity as I can up to the maximum switching frequencies of the DAQ card.  I plan on using shielded twisted-pair cabling for the final leg of my circuit, but I haven't identified the wire gauge, yet.

Chris

[tikir]

Hello Chris,

heavy distortions of the signals occur, if the cables aren't terminated properly. The driver should provide 50 Ohms impedance (a RF series resistor with its output) and the receiver too (a RF resistor to GND), if using RG58 or similar cable. No T-connects with significant impedance are allowed, use splitters if neccessary.

I'm not sure, if the AHCT logic is able to drive 50 Ohms, maybe using several gates of one IC in parallel. The series output resistors decouple the gates from each other.  

If you try to use differential signaling over twisted pair, you could try LVPECL translators like MC100EPT20 as TX and MC100ELT21 as RX. Of course this cable has to be terminated in respect with its own impedance characteristics, it is different from the asymmetric RG58.

Take care of symmetric wire lengths in the layout, a proper RF design is recommended. 

Regards, Timo