[time-nuts] Impedance question

[Magnus Danielson]

Hi,

First off all, for 5 and 10 MHz and indeed PPS, we have rise-times quick
enough that we need to view the cable as a transmission line, which is
exactly what Bob has done.

Second, one should be careful to assume the impedance of a source, as
some sources is far from 50 Ohm at any frequency or they may be 50 Ohm
at the carrier, but not the side-band. Whenever you have a mismatch of
impedance at the source end, things reflect from the source. In my
normal business, the key number is the Return Loss, which indicate just
how damping a signal has on reflection.

In the receiver end, one is tempted to leave it open, because it deliver
almost twice the amplitude of the signal locally than if impedance
matched terminated. The trouble is that un-terminated power reflects and
traverses backwards in the cable onto the source, which impedence match
may or may not reflect power. If the source reflect power, it will now
return back to the receiver, and may shift both amplitude and phase.
Cable loss reduces the effect.

Thus, the return loss of both source and receiver needs to be considered
along with the cable-loss, which is experienced twice.

For higher stability, thermal effect on cables shifts phase around, so
that's when you for sure want to go terminated to reduce that effect.

If, and only if, the source return-loss and cable-loss is high enough,
one may safely use open/high-impedance input, else should proper
termination be used.

Further, termination can also be considered to improve from some
noise-injection.

Personally I prefer to run terminated whenever I can. The main reason I
do not have 50 Ohm on an input is when I put a BNC-T there and sniff the
signal passing by and have termination at another point. For higher
frequencies that is naturally out of the question.

Cheers,
Magnus

On 2020-09-08 02:40, Bob kb8tq wrote:
> Hi
>
> You actually have at least three components to your setup:
>
> 1) You have the *source* impedance of your signal
>
> 2) You have the impedance / length / loss / stability of the cable connecting the 
> source to the load
>
> 3) You have the load impedance. 
>
> In an ideal world, you would have cable that is truly 50 ohms at all frequencies. 
> With that cable a 50 ohm termination at *either* end of the cable would absorb
> all reflections. Either it is eliminated on hitting the load ( = load is matched) or
> it is eliminated when it returns to the source ( = source is matched). Either way,
> you have no ringing reflections. 
>
> Indeed you often terminate *both* ends of the cable. That helps with cables that
> may not *really* be 50 ohms. You get reflections reduced at both ends, so the
> net is a bit better. 
>
> ====
>
> In the case of a pure sine wave, things can be a bit “interesting”. Whatever the
> impedances involved and whatever the cable impedance, it all works to a phase
> shift and a loss. Loss is never a good thing. Phase shift (as long as it is perfectly
> stable) is really not much of an issue. 
>
> ===
>
> So what *is* your setup? 
>
> PPS outputs may or may not be source matched. PPS outputs that are matched 
> and running into a matched load result in logic levels that are 1/2 what you expected
> them to be ( = you have a 2;1 attenuator). Normally PPS is matched at one end *or*
> the other, but not both.
>
> Cable length can matter. A 1 KM cable is a very different thing than a 1 M cable. 
> Generally if the cable is < 1/10 the wavelength of the signal ( 10 MHz has a 30 M 
> wavelength) the cable is better treaded as lumped elements than with transmission 
> line equations. The typical cables normally used tend to fall into this ( < 3 M) range. 
>
> Yes, this can go on and on …..
>
> Lots of fun !!
>
> Bob 
>
>> On Sep 7, 2020, at 8:03 PM, SimBeej <t3acherbj at gmail.com> wrote:
>>
>> Dear fellow Time-Nuts,
>>
>> I have a question about impedance matching (and I apologise in advance for
>> my blatant ignorance on this matter, but I don't have a background in
>> electronics and hardware, so am having to learn along the way).
>>
>> When using frequency counters (in my case a 53230A and SR620) for making
>> frequency or time interval measurements, should I be choosing 50 ohm for 1
>> mega ohm as the input impedance (where the input to the counter might be
>> either a 10 MHz signal or two 1 pps signals from a variety of oscillators,
>> depending on whether I am doing frequency or time interval measurements)?
>> Initially I thought it would be best to match the impedances (in which case
>> I should be using 50 ohm), but now I am not so sure.
>>
>> When I tried to read up on it, I found there is a lot of conflicting
>> information out there. I trust the Agilent Application Note 200
>> (Fundamentals of Electronic Counters)  and it says "for frequencies up to
>> 10 MHz an input of 1 mega ohm is usually preferred". However, the same
>> document also states that "the higher the impedance the more susceptible to
>> noise and false counts the counter becomes". And could there possibly be a
>> problem with reflections if the impedances aren't matched?
>>
>> If anyone out there has a good handle on this sort of stuff and can provide
>> me with some advice, it would be greatly appreciated.
>>
>> Thanks.
>>
>> Belinda
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>> and follow the instructions there.
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.