Reason for using 220 Ohms resistors on signal outputs

[giosif]

Hi,

I have recently acquired a "home made" GPSDO which, beside the 10 MHz sine wave output, also provides 10 MHz, 5 MHz and 1 MHz square wave outputs.
The thing I find rather curious is that, for the three square wave outputs, the in-series output resistors are 220 Ohms, rather than the typical 50 Ohms (please see attached picture).
Could someone please explain why/when someone would use a 220 Ohms output impedance?

Thanks!

[iMo]

People usually put several outputs in parallel with some resistors on each output such they get 50ohm source impedance. Perhaps this concept has been misunderstood by the creator of your gpsdo..

[giosif]

A mistake could be an explanation, but I doubt it: although certainly lacking on the aesthetics front, the overall build looks fairly good to me (electronically speaking).

[iMo]

The output of a CMOS (I do assume you are using there) has some output impedance. Adding 220ohm in series to it will even increase the total output impedance. That xxx ohm with a 50ohm cable (my assumption) will create a mismatch (the edges of the signal will bounce there around in a manner which depends on the mismatch). The other question is whether your single CMOS output has got enough power to drive a longer 50ohm cable.

[rstofer]

The resistors limit short-circuit current to save the chip.

If the chip has very fast rise times, it is sometimes necessary to use a resistor to slow down the edges to prevent reflections.  I use a bunch of 330 Ohm resistors on my FPGA projects as does Digilent where I got the idea from their schematics.

They are often used on inputs to prevent the programmer from defining an output at a level that shorts the pin.  Most uC pins default to input (or high impedance) and that's fine for switch inputs.  But things go sideways if the programmer sets the pin to output against the switch.

[giosif]

The output of a CMOS (I do assume you are using there) has some output impedance. Adding 220ohm in series to it will even increase the total output impedance. That xxx ohm with a 50ohm cable (my assumption) will create a mismatch (the edges of the signal will bounce there around in a manner which depends on the mismatch).

Agree and this is what I'm seeing when using a relatively short 50 Ohms cable going to my scope: severely distorted wave.
OTOH, if I use a scope probe connecting directly at the output, the signal looks fine (or better).

The other question is whether your single CMOS output has got enough power to drive a longer 50ohm cable.

I didn't think of that, but I would hope 1m of cable shouldn't be too long.

The resistors limit short-circuit current to save the chip.

If the chip has very fast rise times, it is sometimes necessary to use a resistor to slow down the edges to prevent reflections.  I use a bunch of 330 Ohm resistors on my FPGA projects as does Digilent where I got the idea from their schematics.

They are often used on inputs to prevent the programmer from defining an output at a level that shorts the pin.  Most uC pins default to input (or high impedance) and that's fine for switch inputs.  But things go sideways if the programmer sets the pin to output against the switch.

Hmmm... That is interesting.
It makes sense for the application you mention but, for cases where there is signal output, how is one to marry the short circuit protection with impedance matching?

[magic]

You need to terminate the cable with 50Ω at the scope. If your scope doesn't have internal termination option, use a BNC tee and a BNC terminator or build this gadget:
https://eevblog.com/forum/projects/home-made-feed-through-terminator-for-bnc/

The reason for 220Ω is most likely because this is some 74HC logic chip running at 5V which couldn't drive 50mA into 100Ω load on every pin, or even 100mA if the output is shorted.

If the receiving end is terminated properly, the mismatched source termination makes little difference, except for reduced amplitude.

[giosif]

You need to terminate the cable with 50Ω at the scope. If your scope doesn't have internal termination option, use a BNC tee and a BNC terminator or build this gadget:
https://eevblog.com/forum/projects/home-made-feed-through-terminator-for-bnc/

The reason for 220Ω is most likely because this is some 74HC logic chip running at 5V which couldn't drive 50mA into 100Ω load on every pin, or even 100mA if the output is shorted.

If the receiving end is terminated properly, the mismatched source termination makes little difference, except for reduced amplitude.

I already tried with a 50 Ohms terminator at the scope end of the cable, but the wave distortion remained.

[magic]

What sort of distortion?

If something like in the first post here, it's expected.
https://www.eevblog.com/forum/projects/compact-74ac14-pulse-generator-pcb/

[rstofer]

It might be that impedance matching was never considered.  If the downstream device is CMOS, the load impedance will be quite high.

[rstofer]

A scope picture would be interesting as would knowing the bandwidth of the scope versus the frequency being measured.  To get a decent image, the scope bandwidth probably needs to be 9 times higher than the frequency and, even then, the image will have some rounding.  Scope make and model would also be helpful.

[giosif]

What sort of distortion?

If something like in the first post here, it's expected.
https://www.eevblog.com/forum/projects/compact-74ac14-pulse-generator-pcb/

I take that back: with a (proper) 50 Ohms termination, the square wave is not distorted anymore.
Teaches me to not trust those cheapo 50 Ohms BNC terminators found on eBay... 

[giosif]

Thank you all for the answers!
Turns out I was using a defective 50 Ohms BNC terminator.