2
\$\begingroup\$

I transmit a signal using LVDS over a CAT6 ethernet cable. However, instead of coming through clean, it always 'bounces' back to the high state: (Pink is the input and yellow is the output) Scope image (Sorry for the glare and angle, there is a window behind me so it's impossible to take good photos)

When the input is turned high, pretty much the same thing happens to the output.

The setup is as follows: Schema handling the input The SN65 chip takes the sync signal (left, 6) and puts a differential output on any channel that is enabled (left, 9..16). Then the differential signal is routed to the RJ45 port (right).

Then there is 5 meter of CAT6 cable.

On the other side there's another RJ45 port that is hooked up quite the same.

Schema for the output

The differential signal leads to this DS90 chip which outputs the sync signal again.

Note that there isn't termination. The designer of this circuit assured me that it's not (or at least shouldn't be) necessary. However, when I manually press a 100 Ohms resistor to the contacts, then the signal is better, but not fixed.

Scope image

Can you spot a mistake that would cause this signal going back to high to happen? Do you have tips on how to improve this situation?

The sync signal is not a clock or anything. It will just be occasionally toggled (once every minute maybe or only at bootup. Depends on what the system will need in the end).

\$\endgroup\$
1
3
\$\begingroup\$

The sync signal is not a clock or anything. It will just be occasionally toggled (once every minute maybe or only at bootup. Depends on what the system will need in the end).

Your differential signal is passing through isolation transformers in your RJ45 magnetics. You cannot pass low frequency signals (or DC) through isolation transformer because transformers like these don't work well at low frequency.

Note that there isn't termination. The designer of this circuit assured me that it's not (or at least shouldn't be) necessary.

It may not be necessary for slow speed signals but it's a moot point because the transformers cannot handle slow signal rates.

\$\endgroup\$
1
  • \$\begingroup\$ Now that you say it, it seems very obvious. Thanks! \$\endgroup\$
    – Geoxion
    Jan 7 '20 at 15:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.