Spurious signals using differential quad encoder

[saratoga]

I'm getting occasional bad counts on my pcie 6323 that cause the instrument to drift out of position with where the NI board thinks it is.  Although the signals out of the encoder do not look that noisy, experimenting with digital PFI line filtering has not made much difference. 

I suspect the problem is that I've tied the negative differential line of both phases to the common digital ground on the NI board, negating the noise immunity provided by differential signaling.  Can anyone recommend a 5v 4MHz differential line to TTL adapter suitable for use with a NI PFI input?  I see some discussion in old threads, but the product links are discontinued or dead.

[Kevin_Price]

I expect some of those old threads were ones I was in, and I see that my recommended device has become a dead link.  Looks like the little company got bought out and searching the site didn't bring up anything that looked familiar.

You could look into their inline converters, maybe one of those would work?  The board I used was bigger, had screw terminals, and could convert in either direction I think (though I only needed 1-way).

Sorry, but good luck finding an alternative, and please post back if you have success with one.  Someone in the future will appreciate the link to a non-obsolete source.

-Kevin P

[saratoga]

I found these RS422 converters:  http://www.bb-elec.com/Products/Datasheets/422ttlx_0515ds-pdf.pdf

But the maximum baud rate is only 115kbit, whereas my encoder can do a few MHz, which is a pretty large difference.  I could breadboard my own, but I'm a little wary of trying to do relatively high speed timing circuits on a DIY board.  I was hoping to find something that was known to work just to help with troubleshooting.  

[saratoga]

I ended up buying these:

https://cnc4pc.com/c46-differential-to-single-ended-converter.html

For $8 I figured it was worth a chance.  They're actually pretty good, analog bandwidth is ~30 MHz, so a 4 MHz TTL signal is has clean edges.  They use the TI 26ls32ac line driver IC.

Initially I thought they didn't work well, but looking at the datasheet, I realized that countrary to what the website says, they are NOT wired correctly for normal ethernet patch cable.  If you plug it into a normal cable, the B differential pairs will share a twisted pair, but the Index and A differential pairs will be mixed (one half of each twisted pair will be each signal), and so crosstalk is very high on those pins.  However, it was simple enough to make a new patch cable with with the correct wiring for the pinout.  I suppose whoever designed that part was not aware that TIA/EIA-568 does not order the patch cable wires sequentially on the connector.  

One other note, they're 10k ohm terminated, so they don't drive low impedance sources at all, and its pretty important that the output of the differential driver be close to the NI board to avoid excessive ringing.  I found with a ~2 foot cable ring down was small enough to be handled by the debounce filter on the NI board, but I made it shorter to be safe.