50 Ohms coax - lots of thanks to everybody !

[AlexResch]

I made another cable from a length somebody kindly gifted me and I used two better BNC connectors. I checked  the agreement between the sig gen and the Siglent SA and seems good enough for my modest needs. I get 1.2dB error at max ampl from the sig gen ( 21dbm ). 2-3 clicks under that and it's all around 0.015dbm give or take. Right now, freq is 156Mhz, level -100dmb and the SA shows -100dbm. I will try another 3642 sig gen in a day or two and report back. The one I am using now is unfortunately a used unit and I don't think I can quite trust it. Many thanks again !

[PA0PBZ]

I get 1.2dB error at max ampl from the sig gen ( 21dbm ). 2-3 clicks under that and it's all around 0.015dbm give or take.

So that is clearly not the cable and most likely the signal generator.

[AlexResch]

I get 1.2dB error at max ampl from the sig gen ( 21dbm ). 2-3 clicks under that and it's all around 0.015dbm give or take.

So that is clearly not the cable and most likely the signal generator.

Well, it could be but then I am not clever enough to tell why. All I see is that with the previous cables I get much more of a difference. With this cable the difference is practically meaningless for my needs. Unfortunately I have no other way to measure power but with the SA. Maybe I should get some sort of power meter from eBay. If you have any suggestions how could I somehow check on the sig gen I would appreciate it A LOT.

[PA0PBZ]

All I see is that with the previous cables I get much more of a difference. With this cable the difference is practically meaningless for my needs. Unfortunately I have no other way to measure power but with the SA. Maybe I should get some sort of power meter from eBay. If you have any suggestions how could I somehow check on the sig gen I would appreciate it A LOT.

I assume that by difference you mean the difference between the level set on the generator and the level displayed on the SA? If so that makes sense, the more losses in the cable the bigger the difference will be (the lower the reading on the SA in relation to the setting on the generator). Each cable has a certain loss per length, and this loss increases when the frequency increases, there's no way around that. But, the cable doesn't care what amplitude the signal is, if it loses 1dB over 100 meter at 100Mhz it will do that on +20dB and on -140dB. If you keep increasing the output you will eventually start to melt the cable and then the former is no longer true, but your generator is far from able to output that much power 

Almost all generators struggle to output the maximum they specify, certainly on the higher frequencies. If you do the same test on 10 MHz the difference could/should be less.

What is your goal? I can't really make any recommendations if I don't know what you are trying to do…

[AlexResch]

1. I assume that by difference you mean the difference between the level set on the generator and the level displayed on the SA?

2. If so that makes sense, the more losses in the cable the bigger the difference will be (the lower the reading on the SA in relation to the setting on the generator). Each cable has a certain loss per length, and this loss increases when the frequency increases, there's no way around that. But, the cable doesn't care what amplitude the signal is, if it loses 1dB over 100 meter at 100Mhz it will do that on +20dB and on -140dB. If you keep increasing the output you will eventually start to melt the cable and then the former is no longer true, but your generator is far from able to output that much power 

3. Almost all generators struggle to output the maximum they specify, certainly on the higher frequencies. If you do the same test on 10 MHz the difference could/should be less.

4. What is your goal? I can't really make any recommendations if I don't know what you are trying to do…

1. Yes

2. I understand and was prepared for that. BUT the cables were inconsistent amongst themselves within the freq band I was looking at - up to around 150Mhz. Freq being low, I was not expecting that.

3. I did most of the tests between 0dbm and -80dbm. The sig gen shows around 0.5dbm error under 140dbm and over 17dbm. At 17dbm ( 50Mhz ) from the sig gen the SA shows 17dbm. Over that I lose around 0.2dbm. ( sig gen 22, SA 21,82). In general the two agree closely ( around 0.03dbm ) and it gets way better if the signal is AM modulated.

4. My only goal is to build simple SW superhet radios and tweak the daylights out of them. I did the same for some years with audio amps but lost interest. Time to learn new things.

[PA0PBZ]

4. My only goal is to build simple SW superhet radios and tweak the daylights out of them. I did the same for some years with audio amps but lost interest. Time to learn new things.

Then don't worry about any differences between the sig gen and the SA, certainly not with signal levels that your radio will never see!

[David Hess]

Make the measurement without the cable and record the results and then make the measurement with the cable inserted into the test loop and measure the change in results.  The change represents the response of the cable.

[G0HZU]

If you want to formally measure the insertion loss of a 50R coax cable by swapping the cable in and out and measuring the change then you have to make sure the source/load VSWR of your test gear is very low. Otherwise you could end up with false and confusing results. You could even end up with a cable that appears to have gain when inserted inline...

[AlexResch]

Make the measurement without the cable and record the results and then make the measurement with the cable inserted into the test loop and measure the change in results.  The change represents the response of the cable.

What does "without the cable" mean ? Don't I need the cable to connect the two together ?

[AlexResch]

If you want to formally measure the insertion loss of a 50R coax cable by swapping the cable in and out and measuring the change then you have to make sure the source/load VSWR of your test gear is very low. Otherwise you could end up with false and confusing results. You could even end up with a cable that appears to have gain when inserted inline...

I wouldn't know how to do that but then, once I am less incompetent I will definitely ask. I'm pretty sure you know a slick trick for that one....

I did learn something interesting though from all these : out of 4 signal gens, the old Marconi 2022 seems the best at these low frequencies. The Anritsu is the worst AND has a serious issue with noise from the LED display(s). Pity the kbd on the Marconi seems to be going bad.

[David Hess]

Make the measurement without the cable and record the results and then make the measurement with the cable inserted into the test loop and measure the change in results.  The change represents the response of the cable.

What does "without the cable" mean ? Don't I need the cable to connect the two together ?

It means you first measure the response from your signal source to your receiver and then add the cable you want to test between them which probably also means adding one adapter.