Is this really a 75 Ohm cable?

[.Scott]

There have been a few times in my career as a SW engineer when I have gone to the EE's on the project and they have been unable to answer my EE question. My thought is that if I post these problems on this message board, they might make it into the EE curriculum much to the advantage of newly minted EE's.

This first one is pretty simple. I was given some 3-foot lengths of 75 Ohm "thinnet coax cable" and told to make sure it worked on our frame grabber boards. After some experiments, I came to the very strong suspicion that this cable was 75Ohms in name only.

So I asked a couple EE's on the project (and later several other EE's) to measure the characteristic impedance of these cables. So far, none of the half dozen or so EE's I have questioned on this had any ideas.

I was able to perform the measurement. It didn't take very long (an hour or two) and it didn't require any specialized equipment ... just the the sort of stuff (cables, scopes, connectors, etc.) we had in the video lab.

All of the cables in the batch were about 66 Ohms - and the decision was made to drop that product.

I will post my solution later.

 

[DaveE]

So far, none of the half dozen or so EE's I have questioned on this had any ideas.

It's a broad field and everyone ends up being a specialist. You can't learn everything. "I don't know, why don't you ask that guy" is often the very best answer.

To me the problem is simple, but not to EEs that don't do that stuff. I'm not a bit concerned with Engineers that can't answer my "simple" questions, I probably can't answer some of theirs either. Unless of course they claim to work in that field. Even so "I don't know" is still the best answer if it's true.

Also, I think you're asking about lab technique as much as theory. If a new grad is working in a good lab and says "ask the Rohde & Schwarz Network Analyzer", I'm not sure they're wrong. They just may never have had to solve it with an oscilloscope. It doesn't necessarily mean they don't know about reflection coefficients and such.

 

[.Scott]

Also, I think you're asking about lab technique as much as theory. If a new grad is working in a good lab and says "ask the Rohde & Schwarz Network Analyzer", I'm not sure they're wrong. They just may never have had to solve it with a signal generator, oscilloscope, and known resistor. It doesn't necessarily mean they don't know about reflection coefficients and such.

So perhaps I have been expecting too much out of the EE's.
But you clearly know the answer. Most of my time was making sure that I had those "reflection coefficients" right.
Here's my set-up:

 

[DaveE]

Maybe you could do it with just an oscilloscope, of the right flavor. Probe calibration output for the source, internal 50 ohm termination for the know R, and the display for measurements. OTOH, the network analyzer is quicker.

I bet there are a bunch of creative ways to do it. Like a DMM and a handful of 1N4148s for a detector maybe? They made radios long before they had oscilloscope and such.

 

[.Scott]

Of course, in this specific situation, I wasn't trying to be clever. I only needed to demonstrate the source of a problem.

About a third of the EE's I have tried this problem with have been pretty vague on exactly what "characteristic impedance" is. Perhaps if this was demo'ed when the topic came up in class, it would stick a bit better.

I have another case, there involves properly "cruising" a PLL during "sync-less" periods. It required a detailed understanding of what the PLL was doing - and would definitely fall under the "specialist" description provided by @DaveE above.

A third case is perhaps "clever", but it's more MacGyver Physics than Old EE ... How to erase a uv EPROM without a lamp or eraser - sunlight isn't enough.

 

[DaveE]

I have another case, there involves properly "cruising" a PLL during "sync-less" periods.

That was part my first real analog design job, in 1982. Part of a tracking antenna nutator drive system. If the reference signal was lost we wanted stay in phase as much as possible until it returned. I learned a lot about analog track and hold amplifiers and PLLs then. Of course now it would be digital; another entry into my bank of obsolete knowledge.

 

[yungman]

Yes, it's very simple to measure. Like you said you have scope and generator. Just use a high speed pulse ( or even square wave )with rise time within 1nS or so into one end of the long cable(3ft is plenty) and put the other end on the scope with different termination resistors and find the value you get a nice square pulse( no ringing or funny shape). If it is only in a few hundred MHz, I bet you can use a small single turn pot and it will work quite well already. More importantly, all connections have to be point to point. Don't leave long ground wire and stuff, that will introduce ringing.

I did not try to "find" the impedance, I did verification of impedance of striplines and stuff, I just terminate with the designed impedance and look at the waveform.

I never use any more fancy setup and it's very conclusive already.

 

[.Scott]

That was part my first real analog design job, in 1982. Part of a tracking antenna nutator drive system. If the reference signal was lost we wanted stay in phase as much as possible until it returned. I learned a lot about analog track and hold amplifiers and PLLs then. Of course now it would be digital; another entry into my bank of obsolete knowledge.

In the case I dealt with, the original design "froze" the VCO voltage near the sync event. But for precise results, you need to do it near the midpoint between two sync events.

 

[Baluncore]

I was given some 3-foot lengths of 75 Ohm "thinnet coax cable" and told to make sure it worked on our frame grabber boards. After some experiments, I came to the very strong suspicion that this cable was 75Ohms in name only.

Where did the spurious claim of 75 ohms come from?

Thinnet used 4.75 mm diameter, RG-58 coaxial cable, with an impedance of 50 ohm. It was used for Ethernet. RG-58 was thinner and more flexible than the lower-loss RG-213, 10.3 mm cable, used for longer cable runs in buildings. RG-59 was 75 ohm cable.

Ethernet was specified as 50 ohm. Standard BNC connectors are 50 ohms. There are a few special BNC connectors that are slightly different mechanically and electrically for 75 ohms.

In general: Solid dielectric cables are nominally 50 ohm. Foam dielectric cables are 75 ohms.

 

[.Scott]

Where did the spurious claim of 75 ohms come from?

It was in the data sheet.
The product was actually 4 cables joined together to carry 4 channels of video.
This was a couple of decades ago - I don't know if it claimed to meet RG-59 - or whether it predated RG-59.

 

[yungman]

One thing I am puzzled, usually manufacturer do do quality control, I would be really surprised it is off. I personally never get into qualify cables, but I never run into question the accuracy of the impedance of coax.

I looked at your drawing of the setup, it's quite a bit more complicate than what I usually do. Is your generator 75ohm source impedance? I guess I don't work on video, am my stuffs is 50ohm.

Also, you connect a known 75ohm coax to the cable under test, you sure the coax is really 75ohm?

It is very hard to talk by looking at the drawing. Have to look at the setup. My experience in RF is if you don't get the result you think you should get, over 1/2 of the time is something wrong with the setup. They have to be very exact.

Also, what is the rise time of your generator? Make sure the rise time is around 1nS and source impedance is 75ohm. Don't get offended, just make sure your scope is set to high impedance input! I just want to make sure because my scope has option of 50ohm termination. Just checking.

That's why I use a very simple setup and put the scope at the end of the line using a "T" and put the termination. I eliminate the extra coax and one disruption of the scope in the middle. This is so I have minimal disruption on the Tx line. Any mismatch will cause ringing. You can even do this with a known 75ohm coax first to look at the result, then put the coax in question to compare. You should get a very nice square wave if the line impedance matched.

Lastly, remember the characteristic impedance might vary, you might want to check the spec. Say if the spec says +/-5%, you can get as low as 71.5ohm. Then a little error in measure and it's not unthinkable to see about 66ohm. To be honest, one doesn't have to be EXACTLY 75ohm. If you are within +/-5%, I double you will notice the difference.

 

[DaveE]

usually manufacturer do do quality control

The good ones do. There's also a lot of crap out there (eBay, etc.).

 

[yungman]

The good ones do. There's also a lot of crap out there (eBay, etc.).

Ha ha, If this is on a job, I would hope OP buy it from a reputable store of a known brand!

I did talk about I bought wires and cables on ebay before, but those are for audio stuffs. If I do RF, it has to be known brand from reputable place, no exception. Ha ha, I am born cheap, but not that cheap.

 

[.Scott]

One thing I am puzzled, usually manufacturer do do quality control, I would be really surprised it is off. I personally never get into qualify cables, but I never run into question the accuracy of the impedance of coax.

I looked at your drawing of the setup, it's quite a bit more complicate than what I usually do. Is your generator 75ohm source impedance? I guess I don't work on video, am my stuffs is 50ohm.

Also, you connect a known 75ohm coax to the cable under test, you sure the coax is really 75ohm?

It is very hard to talk by looking at the drawing. Have to look at the setup. My experience in RF is if you don't get the result you think you should get, over 1/2 of the time is something wrong with the setup. They have to be very exact.

Also, what is the rise time of your generator? Make sure the rise time is around 1nS and source impedance is 75ohm. Don't get offended, just make sure your scope is set to high impedance input! I just want to make sure because my scope has option of 50ohm termination. Just checking.

That's why I use a very simple setup and put the scope at the end of the line using a "T" and put the termination. I eliminate the extra coax and one disruption of the scope in the middle. This is so I have minimal disruption on the Tx line. Any mismatch will cause ringing. You can even do this with a known 75ohm coax first to look at the result, then put the coax in question to compare. You should get a very nice square wave if the line impedance matched.

Lastly, remember the characteristic impedance might vary, you might want to check the spec. Say if the spec says +/-5%, you can get as low as 71.5ohm. Then a little error in measure and it's not unthinkable to see about 66ohm. To be honest, one doesn't have to be EXACTLY 75ohm. If you are within +/-5%, I double you will notice the difference.

When I collected this data, it was new territory for me, so I did quite a bit of validation.
I don't recall the rise time, but I could easily resolve a few inches of two-way signal travel time on the scope, so it must have been well less than 1ns. Also, the work we did commonly involved timings below 0.5ns, so I am sure that both the signal generator and scope could work in that range.

I was triggering off the initial rising edge and displayed roughly 20ns of data

I measured the 75 Ohm terminating resistor on a ohm-meter that was under current calibration.
Before measuring the thinnet cable (shown as red in my diagram), I performed these measurements:
1) Just the two orange cables and the terminating resistor. The terminating resistor could be seen on the scope - but it was very minimal and there was no net change in voltage level after the reflection. This demonstrated that second orange cable was, in fact, 75-Ohms or very nearly 75-Ohms. No reflection off the signal generator observed.
2) Just the two orange cables with a 50-Ohm terminating resistor. This was just to make sure that I ready to capture the reflection. As expected, the reflection showed a drop in the voltage level. This measurement also provided me with a check on the equations that convert the voltage drop to the cable impedance.
3) Three orange cables with the barrel connector and the 75-Ohm terminating resistor - so I had a regular 75-Ohm cable where the thinnet cable was going to go. With this set-up, the scope clearly showed the barrel connector and the terminating resistor - but there were no persistent voltage level changes.

4) Then finally, the setup shown above with thinnet cable. This clearly showed the voltage drop from the reflection at the barrel connector and the voltage rise from the reflection at the terminator.

As I recall, I had two assemblies of 4 thinnet cables each. I repeated this experiment on all four cables in the first assembly and some from the other. All of the measurements on the thinnet cables gave the 66 Ohm result.

The product we were developing was an analog video frame grabber board. Since I had developed the SW for measuring the quality of the grabbed images, I was assigned the task of evaluating this "new" cable. These frame grabber boards can easily resolve the correct 8-bit code used to generate an analog pixel value - so +/-5% cable impedance would not be adequate. Even before I performed these measurements, I had triple checked the impedance matching on the frame grabber board itself - and it was good. So, by the time I asked for a impedance test on the thinnet, I already had strong evidence in hand.

 

[tech99]

If a cable under test is very short then the method will not be able to see the mismatch. A classical method for finding the charachteristic impedance of a cable (or other 4 terminal network) is to measure the input impedance with a short circuit and then an open circuit on the far end. The characteristic impedance is then found from Zo = SQRT (Zoc x Zsc).

 

[Tom.G]

Or if it is not critical, simply Z_o = SQRT(L/C).

That's handy if you have an LCR meter, or some way to measure L and C.

I've used it when making long runs of video cables from a computer to a display. The long runs were often done with several shorter cables/extensions strung together.

Sometimes there was reflections/ghosting due to a cable having a different impedance. All you really needed was to measure the ratio of L/C in each cable to find the outlier.

Measure each cable for L with the far end Shorted, and for C with the far end Open.

Cheers,
Tom