Driving 50 Ω coax cable for doing measurements with oscilloscope

Question

This is a follow up question to Using 50 Ω coax cable instead of passive probe.

In the linked question I asked whether it would be possible to use a 50 Ω coax cable instead of the passive probe for making measurements.

While it was confirmed to be possible, it isn't ideal because - from my understanding now - the high capacitance of the coax (which will attenuate high frequency content) and the fact that the transmission line isn't properly terminated with a 50 Ω load which will lead to reflections.

When writing the linked question, I kind of was under the impression passive probe coax would be the same as regular 50 Ω or 75 Ω coax which it isn't (excerpt from an answer to linked question, emphasis mine):

A passive probe is more than just a 9 MΩ resistor and a piece of coax. [...]

Second, in order to deal with the fact that the scope input (when switched to 1 MΩ) is not matched to the coax, the coax itself is actually constructed with a deliberately lossy (resistive) center conductor.

I also found a nice graph from an article showing the difference:

Where we can see that a passive probe's coax is indeed lossy compared to a regular coax cable.

Now I'm re-considering my approach because I kind of want to do this right, even though I'm technically not working with RF.

My question basically boils down to if there are dedicated devices/probes/chips that have a high impedance which are able to drive the 50 Ω coax. The idea is to use a short coax cable (to minimize capacitance) from the PCB under test to another device/PCB/probe that has a driving circuit that will drive the longer coax cable which will go to the scope - in an effort to not have to place anything on the PCB under test except the UMCC connector.

Because as far as I can see, if I want to use a coax connection to my scope that isn't lossy (like the passive probe ones, which are hard to get) I have to use 50 Ω termination and have something that is able to drive that 50 Ω load.

Maybe I'm all overthinking this or the approach to solving my problem which is ultimately getting a robust and reliable probe connection to my PCBs is uncalled for.

---

This section is not directly related to the question asked, but it shows the bigger picture/problem I'm trying to solve here:

The main reason I'm not happy with passive probes is the relatively high inductance of the ground clip and that the ground spring isn't really able to be easily fixated on the DUT. I have scopes from different brands (Rigol and Keysight) and I'm looking for a solution that is brand agnostic if possible. Also, I want to keep "test points" small on the PCB - that's why I think UMCC connectors are an appealing solution.

Answers from the linked question also mention using something else than coax:

A clip-on test point for a 10:1 scope probe would be around an order of magnitude less loading.

Better still, you can also get sockets that mate directly with scope probes and provide both the signal and the ground connection.

As mentioned in a comment on one of the answers, they make connectors that mate directly with scope probe tips. I’ve seen them on EVAL boards from chip manufacturers.

I have searched for those kind of sockets but didn't really find something that wasn't too big or being able to be SMT soldered like the UMCC connectors. It's also unclear to me how compatible such sockets are (i.e. if a probe socket designed for Tektronix Scopes will work with a Keysight scope and vice versa).

Answer 1

In you previous question, in bold you wrote:

I'm not talking about RF

That makes a great deal of difference. If what you are measuring is low frequency or DC, then the transmission line characteristics of the cable are not that important.

a passive probe's coax is indeed lossy compared to a regular coax cable.

Yes, indeed that is generally the case. However, if you connect a cable with, say 250 \$\Omega\$ (or even more) resistance to an oscilloscope input with an input impedance of 1 M\$\Omega\$, the 250 \$\Omega\$ introduces very, very, little attenuation. Not the many decibels of attenuation shown in your chart.

The idea is to use a short coax cable (to minimize capacitance

If the frequencies of the signals you are measuring are low, and the signal edges are not terribly sharp, then the capacitance of plain 50 \$\Omega\$ is not going to upset your measurements very much.

I have to use 50 Ω termination and have something that is able to drive that 50 Ω load.

Again, you don't need to terminate your transmission line if you are using only low frequencies and not RF. However, if you are using high frequencies or sharp transistions, and you do not want to load you circuit with 50 \$\Omega\$, you could add a 450 \$\Omega\$ resistor in series with the center pin of the on-board coax-connector, and use a 50 \$\Omega\$ pass-through-terminator between the coax and the oscilloscope's high impedance input connector. Together, they will cause a "10X" attenuation, but virtually all oscilloscopes have a 10X option. This will reduce the loading on the circuit under test from 50 \$\Omega\$ to 500 \$\Omega\$.