121gw capacitance measurements

[pontuss]

Hi all, I'm very new to this forum and did some searches here regarding my question, but nothing came up.

I have a 1 meter cable between two racks, a DSUB15 "straight" cable, pin 1 to pin 1 etc, male to female.

The cable transports high speed (300 Mbsp) LVDS differential signals and it uses twisted pairs (TP) within the cable.


For some reason the cable itself has all pairs combined with a white companion, so we have blue-white, green-white,
yellow-white and orange-white. This makes every one confused, including the cable producer.

The cables passes test, which is a DC test.

But the cable is sometimes bad in reality, the white cable part of the TP, may be swapped to an adjacent signal,
and signal integrity is lost, meaning an transmission error every other day.

So my colleague came up with an idea, twisted pairs should have more cap than other parts in the cable...?
Yes! This actually is a way to determine if we use the correct TP pairs, we get (on his fluke DVM) cap readings of
say 110 pF for "other TP partners" and 116 pF for correct TP partners. The measurements were +-2 pF so it seems OK.

I tried this with my 121GW DVM, and measured 0.000 nF on both the good and the bad cable!

Is this only observable at higher frequencies?
The 121 GW spec says +-10 pf +- 2.5% on a (was it 10 nF range?)


I like the 121GW alot, so I would like to be able to continue using it with full confidence.

Any Ideas?

[pontuss]

I admit that I have a "bat low" iniication on th DVM.

[pontuss]

Fw version 2.02 IIRC.

[joeqsmith]

See the following:

https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg4442128/#msg4442128

[pontuss]

Thanks for giving me pointers to other related things!
Actually another colleague just got his VNA-nano to test his short wave equipment, and I will probably loan it for a few days
to test other things.

My intention was to help production to test cables they manufacture, so my idea was:
- make the cable
- DC test the cable
- AC test the cable (i.e. measuring cap between pins) If we are in the correct TP cap should be higher.

What I don't understand is that the 121gw fails completely doing this.

But since the Fluke meter can do this, I guess that's what we end up using.

Measuring cap seems to be kind of non-trivial?

[J-R]

Doing some tests with lengths of CAT5e, I think the issue is that the "real" capacitance of that short section of cable is much lower than what the Fluke is stating, especially when factoring in typical test leads.

I tested with my Fluke 287 it reported a 67pF cap as over ~150pF, while the 121GW reported around 55pF.  The test leads are most likely around 30-50pF, but I've calibrated that out on my 121GW.

With a 3m CAT5e cable, I observed ~50pF for non-paired and ~100pF for the paired with the 121GW, but ~120pF for the non-paired and ~180pF for the paired with the 287.

Moving to a much shorter cable, less than 1m, my 121GW also registered zero but the 287 reported ~120pF for non-pairs and ~130pF for the paired.

The output waveforms on DMMs vary drastically between manufacturers and even ranges, so I think with the cable being so short you're just below what the 121GW can manage via its chosen method and perhaps the factory calibration.

I have a few dedicated capacitance meters that excel at jobs like this, the UEi DCM3 and the Beckman CM20A, and of course there is the venerable DE-5000.

[AVGresponding]

You could try the offset method

[J-R]

The "offset" method does work, but it's a bit tricky at the single digit pF differences on the 121GW so I didn't suggest it.  But if you want to give it a try, attach a capacitor in parallel with the measurement to bring the value up (offset) into a range the DMM can read more reliably.  The method I used was bend the 67pF cap leads like this: < > and attach the alligator clips there, then use the remaining | | portions of the capacitor leads to probe the cable (or whatever is of interest).  Using this method I could just barely observe in practice a difference on my 0.3m patch cord.  Basically I got right around 56pF for the non-paired wires but it would slowly creep up to around 61pF over a few seconds on any paired wires.

Using my Beckman CM20A and UEi DCM3 with or without anything in parallel was a total breeze: solid/reliable/instant numbers down to the 0.1pF with no fuss using just standard DMM test leads.

[J-R]

I'm also recalling that on my early-model 121GW if you turned it off with the capacitance mode selected, then some time later flicked it back on directly to that position, there would be some capacitance visible, basically the value of the test leads from what I recall (20-35pF or so).  Then if you used the Mode button to cycle through the other three modes and back to capacitance, it would suddenly read zero.  I don't recall my late-model 121GW doing this.  It would always read zero.  I believe this behavior stems from a change to the input section of the DMM.

As they sit now, my 121GW units are using the hacked firmware where they always revert back to a specific configuration at power-on, because I hate DMMs that I'm always having to guess what mode it will be in when I turn it on.  But this also solves the issue of having the capacitance mode read differently depending on the condition mentioned above.  So they default to resistance at power-on, and I've calibrated the capacitance range with that in mind, which basically is zeroing out the test leads.  I suppose it's personal taste.  Do you want the DMM to read the value of the capacitor, or the capacitor plus the test leads??

But regardless, the dedicated capacitance meters destroy all DMMs easily, so they should be a better choice for this project.

[pontuss]

Wow, i'm glad I got a gw121, not just because it's good meter, it seems to have a very knowledgeable/responsive/supportive support forum.

The offset method was an excellent idea, I added a 100 pF (NP0) cap to the test adapter, and gw121 gave very similar
(differential) results to the fluke!
Both are offset from zero - with different offsets - but both confirming a difference of 6-8 pF depending on correct/bad twist-pairs,
with what it seems (from sloppy measurements, w.r.t. how the std DVM leads where on the bench) a measurement uncertainty of
atmost +-2 pF.

So I'm home!

 Thanks, and I will be back!