Hello:

I have a 3.3V 100 pulse-per-second signal coming from an FPGA. The signal is isolated and converted from 3.3V to 5V using a digital isolator.

The signal needs to be driven over a 50 ohm coaxial cable. The load impedance ranges from 50 ohms to 1 Mohm

Currently, I use a Philips (NXP) 74F3037 NAND Driver.

http://www.nxp.com/#/pip/pip=[pip=74F3037_2]|pp=[t=pip,i=74F3037_2]

The chip has 4 outputs, 2 of which I parallel together for increased drive strength. The other 2 are NC. The output resistors on the 2 outputs are 100 ohms (50 ohms parallel). See the attached PDF for details.

What I plan on trying is:

1/ Reducing the output impedance. I currently find the response is over-damped.
2/ Connecting more outputs together. Perhaps there isn’t enough drive strength?

Previously the rise-time was not critical, but now, for measurement purposes ONLY (1m coax into a scope) - a fast rise-time is required.

In the previous design a gate-driver was used and I followed the same concept in this design. I assume the previous designer (no longer around) used a gate driver because the original rise time requirement was better than 1 us and, as I mentioned, the signal is only 100PPS.

I’ve done some reading and I typically see current-mode opamps are used for high performance; however, I don’t have much experience in that area which is why I followed suit with the previous design and stuck with a gate-driver.

Does anyone have any suggestions for this design?

Appreciated,
Gavin

 

You won't get a clean signal through any significant length of coax without also having proper 50 Ohm termination.

Anything else will cause reflections and give a fuzzy edge or ringing.

50 Ohm in series at the drive end plus a 50 Ohm load should give a good response.

Some of the fastest line drivers deliberately use relatively slow edges, as the receive level settles faster that way due to the lack of overshoot and ringing.

I've used a few 74AC04 inverters in parallel to drive coax for a TDR setup before now, any fast logic gates should do.

If monitoring with a scope, ONLY use a x10 probe - that should have capacitive as well as resistive input matching, and is the only thing that will allow you to see the true waveform. (Unless the scope has a 50 Ohm termination option, but the standard is 1M for use with a probe).

 

I've used a few 74AC04 inverters in parallel to drive coax for a TDR setup before now, any fast logic gates should do.

A 50 Ohm BNC termination on a BNC 'T' connector will be more than acceptable for measurement on the scope. BTW, TDR is one of the most fun things you can do with a scope and a pulse gen.

 

Hi Robert:

Thanks for your reponse. You're right, the 50 ohm load termination makes a significant difference (my scope can do 1M and 50R).

The problem was not the driver but the capacitance of the TVS causing the slow rise time (large RC time constant). I'm going to search for a lower capacitance TVS that will hopefully still provide adequate protection.

-Gavin