You aren't going to be able to parallel terminate that cable (though you may be able to get close enough that it's worth trying). If it were a ribbon cable with alternate grounds, then the characteristic impedance (and hence the correct terminating resistance) would be about 150R. Your bundle of wires is going to be more variable, but not so much different. But if you attempt to parallel terminate it (ie. resistor to GND at the receiving end) then the driver has to supply a huge DC current - 20mA per pin! This can be reduced by making it more symmetric: generate a regulated (and well-decoupled) 1.5V and terminate to that instead of GND (half the current). Your random bunch of wires is also going to suffer from the fact that it will behave differently depending on how many pins are switching at the same time; that can be fixed by switching to a ribbon cable with alternate cores ground.
An alternative approach is to series-terminate at the source end (and no termination at the receiving end). Again the value should match the characteristic impedance of the cable, minus the output impedance of the driver (but that's quite small for the 74HCT you are using). The idea here is that you don't get rid of the primary reflection, but you catch it on the way back and allow precisely one reflection and no more. So the voltage on the cable initially goes to half the desired voltage (150R in series, and 150R characteristic of the cable acts like a potential divider, so you see a half-size signal), then the signal travels to the far end, discovers there's no termination, reflects back at twice the height (so now actually the desired size) and gets absorbed at the source terminator.
So:
- the size of termination resistors you should be playing with is around 150R.
- Switching to ribbon cable with alternate grounds would be ideal if that's feasible for you.
- source termination (in series) seems the most likely to give an easy win.
- parallel termination to a 1.5V rail would be a good solution but probably too much bother.
- parallel termination to GND could work, but you've got very high currents and need to source that energy (lots of decoupling capacitors, extra cores in the cable etc).
An alternative approach is to series-terminate at the source end (and no termination at the receiving end). Again the value should match the characteristic impedance of the cable, minus the output impedance of the driver (but that's quite small for the 74HCT you are using). The idea here is that you don't get rid of the primary reflection, but you catch it on the way back and allow precisely one reflection and no more. So the voltage on the cable initially goes to half the desired voltage (150R in series, and 150R characteristic of the cable acts like a potential divider, so you see a half-size signal), then the signal travels to the far end, discovers there's no termination, reflects back at twice the height (so now actually the desired size) and gets absorbed at the source terminator.
So:
- the size of termination resistors you should be playing with is around 150R.
- Switching to ribbon cable with alternate grounds would be ideal if that's feasible for you.
- source termination (in series) seems the most likely to give an easy win.
- parallel termination to a 1.5V rail would be a good solution but probably too much bother.
- parallel termination to GND could work, but you've got very high currents and need to source that energy (lots of decoupling capacitors, extra cores in the cable etc).
Statistics: Posted by arg001 — Fri Feb 02, 2024 10:02 am