What's wrong with this RG58 cable?

[gf]

When I feed a square wave signal from an AWG into my scope, using different RG58 BNC cables of 1m length, I notice a rise time of ~7ns with some of the cables, while others do increase the rise time to 9ns. Since rise times add up geometrically, this would mean that the cable alone accounts for 5.5ns.

So I did a TDR of these cables (shorted at the end), applying a ~2.7MHz square wave signal, and got the attached results. The TRD of the "good" cable looks basically as I did expect (for a shorted end), but the signal of the "bad" one does not return to zero between the pulses.

I'm curious which deficiency of the cable does exactly lead to this behaviour?
Is this still within the specs of an RG58 cable?

Thanks,
gf

[edavid]

The simplest explanation is that it's mismarked 75 ohm cable.

[tggzzz]

What is the risetime of the signal?
What is the risetime/bandwidth of the scope?
How are the scope, signal and cable connected together?
What's the scope input impedance?

I would find it easier to interpret if the pictures showed the signal, then the signal applied to the cable with the cable unterminated/open and terminated in 50ohms.

[jmelson]

Is this still within the specs of an RG58 cable?

There are DOZENS of variants of "RG58", such as RG-58A/U, RG-58C/U and so on, that have SLIGHTLY different characteristic impedances.  They vary between 49 and 53 Ohms or so.  But, what you showed might indeed be mismarked 75 Ohm cable.
One way to tell is to put various resistors on the far end instead of a short.  A resistor that exactly matches the cable's impedance will give no reflection.

Jon

[gf]

Thanks for the responses so far. I'm still wondering whether it is just an impedance issue. Impedance mismatch certainly leads to reflections, but is it also suppused to "blur" the pulse edges? I'll follow-up with more information later.

gf

[macboy]

You didn't mention whether you have 50 ohm termination at the scope. If not, then your AWG is driving a big capacitance (between the cable's conductor and shield) rather than a transmission line. This will absolutely cause degradation of the rise time. You will find that adding termination improves both cables, but likely the 50 ohm one more so.

I agree with the others, the bad cable is definitely not 50 ohms. Long ago I built a little impedance tester which is nothing more than a BNC jack with a 100 ohm pot (wired as a rheostat) for termination.  With the same simple TDR setup as you have already used, the pot is twiddled until the termination matches the cable. Then the resistance is measured with a DMM to determine the impedance of the cable.  I was surprised to find some 63 (maybe 62.5) ohm BNC-terminated coax cables in my stash.

[jmelson]

Thanks for the responses so far. I'm still wondering whether it is just an impedance issue. Impedance mismatch certainly leads to reflections, but is it also suppused to "blur" the pulse edges? I'll follow-up with more information later.

Those long tails on between the pulses do look like an impedance mismatch, I think I can even see the stepped reflections.

My method of TDR is to leave the scope on 1 meg-Ohm, and put a BNC Tee on the input.  feed the pulse generator into one end of the Tee, and the cable to be tested on the other end.  Connect a terminator resistor to the far end of the cable to be tested.  If you see a step and then just a horizontal line, the cable and the terminator match.  If there are stepped reflections equal to the round-trip delay of the cable, you can figure out whether the cable is lower or higher than the terminator.  And, you can see the various connectors or other imperfections in the system.

Jon

[gf]

Thank you guys, you were obviously right. There is even a severe impedance mismatch. Eventually I had to terminate the "bad" cable with ~120 Ohm in order to get rid of the reflection (for a cable which is labeled "RG58 50 Ohm").

Even my "good" cable shows a small amount of reflexion, but I think this is still in the green range.

What is the risetime of the signal?
What is the risetime/bandwidth of the scope?

Specified scope BW is 70MHz -> scope risetime about 5ns. Don't know the AWG risetime. The scope measures 7ns, so the AWG risetime islikely about 5ns as well. Sorry, don't have faster equipment.

My method of TDR is to leave the scope on 1 meg-Ohm, and put a BNC Tee on the input.  feed the pulse generator into one end of the Tee, and the cable to be tested on the other end.  Connect a terminator resistor to the far end of the cable to be tested

Well, that's exactly how I did it.

I would find it easier to interpret if the pictures showed the signal, then the signal applied to the cable with the cable unterminated/open and terminated in 50ohms.

I'm attaching the results for open, shorted and 50 Ohm terminated end, for "good" and "bad" cable, and for comparison I've also added a 75 Ohm SAT cable of the same length. And finally, the "bad" cable with 120 Ohm termination.

Thanks,
gf

[tggzzz]

Thank you guys, you were obviously right. There is even a severe impedance mismatch. Eventually I had to terminate the "bad" cable with ~120 Ohm in order to get rid of the reflection (for a cable which is labeled "RG58 50 Ohm").

That is so gross that I would suspect there is fault in a connector or the core is broken somewhere along its length.

Have you tried measuring the DC continuity between the two centre pins of the connectors, and the two connector shields? Both should be a fraction of a ohm.

What is the risetime of the signal?
What is the risetime/bandwidth of the scope?

Specified scope BW is 70MHz -> scope risetime about 5ns. Don't know the AWG risetime. The scope measures 7ns, so the AWG risetime islikely about 5ns as well. Sorry, don't have faster equipment.

As you can see from your pictures, the short cables plus long risetimes means you can bearly see any detail "inside" the cable. While you can't change your scope, it should be possible to use jellybean CMOS logic to generate a much faster edge than the AWG can.

With your equipment the best that could be achieved would be to "extend" your cable by adding another length of your coax in series with the suspect cable. Several measurements could be done, with the good cable before/after the suspect cable, and with the suspect cable in "forwards"/"backwards".

[gf]

That is so gross that I would suspect there is fault in a connector or the core is broken somewhere along its length. Have you tried measuring the DC continuity between the two centre pins of the connectors, and the two connector shields? Both should be a fraction of a ohm.

DC resistance is OK. I also don't notice any loose contact (when connected, the signal is stable, even if cables are moved).
And I got three of these cables, each of them having the same deficiency.
They are brand-new. I recently bought them at eBay.
https://www.ebay.de/itm/112883482965?ul_noapp=true

As you can see from your pictures, the short cables plus long risetimes means you can bearly see any detail "inside" the cable. While you can't change your scope, it should be possible to use jellybean CMOS logic to generate a much faster edge than the AWG can.

I'm aware of the limitation, still one can seen the presence or absence of reflections. Maybe I build such a pulse generator if I find some time.

With your equipment the best that could be achieved would be to "extend" your cable by adding another length of your coax in series with the suspect cable. Several measurements could be done, with the good cable before/after the suspect cable, and with the suspect cable in "forwards"/"backwards".

Thanks, that's certainly worth a try 

gf

[T3sl4co1l]

It's just lossy as shit.

It's probably still legit "RG-58", it's a terrible spec.  RG-8/U is a lot better if you're curious, or RG-316/U if you want to get spendy(?).

Doesn't look to be an impedance mismatch issue.  But it would be helpful if the cable were a few times longer, or the pulse a few times sharper. 

The output with a higher termination looks cleaner because more high frequency content is getting through, to make up for the attenuation in the forward direction.  It doesn't seem to bounce because the subsequent reflections are attenuated twice (or more) over.

This is a neat tip for signal quality on PCBs and cables, by the way: if you expect you'll be dealing with HF losses, a somewhat lower source termination, or somewhat higher load termination, can compensate for cable "drool".

Tim

[gf]

Doesn't look to be an impedance mismatch issue [...] The output with a higher termination looks cleaner because more high frequency content is getting through, to make up for the attenuation in the forward direction.  It doesn't seem to bounce because the subsequent reflections are attenuated twice (or more) over.

At least the 75 Ohm SAT cable response (which is mis-termianted either, when 50Ohm are attached) looks indeed different and settles clearly after the 2nd or 3rd reflection (particularly visible in the shorted response), while the bad cable response has this long, smeared tail.

Sorry again for the limited equipment. For non-professional use, I don't want to spend so much. In fact my intention was not to do cable testing or RF work @GHz. I just stumbled into this topic when I got the feeling that something is wrong with the new cables - so I had to start some investigations...

gf

[tggzzz]

Sorry again for the limited equipment. For non-professional use, I don't want to spend so much. In fact my intention was not to do cable testing or RF work @GHz. I just stumbled into this topic when I got the feeling that something is wrong with the new cables - so I had to start some investigations...

No apologies necessary.

It is always pleasant to see someone checking, thinking, investigating, asking for help, listening, understanding. The most exciting words in science aren't "eureka", but "that's strange...".

[T3sl4co1l]

Indeed, doubt everything.  Measure everything, the best you can, with what you have.

It's quite convenient really that you have a pulse generator juuust fast enough to resolve that length of cable!

Likewise, be careful not to read too far into measurements that you "can't" make -- this is why knowing electronics theory is equally as important as knowing electronics practice.

For sharper pulses, just some jellybean logic will do very nicely actually (74LVC family and such), or an avalanche pulse generator is quite easy to build (there are many threads here).   That will quite easily make your scope the limiting factor (these need upwards of 500MHz bandwidth to resolve well).  The instrument is still the limiting budget factor even today, I'd say, although if you don't mind working in the frequency domain, there are some amazingly affordable spec and generator tools out there these days.

Tim

[radiolistener]

I have several 1 meter LMR195 pigtails from aliexpress and they all have bad SWR=1.3 at 28 MHz.
One of them has clear marking LMR195, so there is no mistake with cable type.

I tried to heat soldering points on connectors with iron in order to improve soldering quality, but it doesn't help.
Then I tried to measure cable capacitance and inductance.
Capacitance was within LMR195 specification.
But inductance was twice more higher than specification.
So I think this is just defective Chinese coax cable with wrong impedance. 

I also bought several 1 meter pigtails with RG58 cable from aliexpress.
But they works ok and have SWR ~ 1 up to 200 MHz.


Just connect 50 Ohm RF dummy load on the end of cable and try to connect it to transceiver output through SWR meter. Try TX with different frequencies at 10-100W and see what is SWR. Also try to sweep it with vector analyzer. In such way you can check if your cable is usable.