Measuring RMS noise - UK Vintage Radio Repair and Restoration Discussion Forum

ms660 · 3rd Jul 2016, 7:20 pm

With an oscilloscope:

https://www.youtube.com/watch?v=zHU9xdU51FY

Not sure if I've put that in the right section here?

Anyone tried this method?

All comments welcome.

Lawrence.

Radio Wrangler · 3rd Jul 2016, 7:56 pm

Hmm

Tek's applications note isn't bad, but it does make the assumption that the probability density function of the noise is Gaussian. It usually is, and the central limit theorem makes this likely, unless the individual noise contributions are related by say avalanche mechanisms.

If you try this to assess the level of noise out of a receiver, you'll find the noise is not as smooth as he has, and unless you slow the scope down to a slow speed, it'll look jumpy, and it'll be too slow to do the eyeball comparison. His demo is using a much broader bandwidth of noise... a factor of about 10,000 times.

Where he comes unstuck is right at the beginning by saying what the power level of the noise is. With a noise signal, power is distributed over frequency and so the total amount of power you see depends not only on the noise signal, but also on the bandwidth of what you're using to look at it with. So if you have a wider bandwidth scope, you'll think the noise level is greater. You cannot state noise powers without also stating what bandwidth that noise power is in.

David

ms660 · 3rd Jul 2016, 8:02 pm

Have you got or can you point me to tek's application note, I'd be interested to read it.

Lawrence.

Radio Wrangler · 3rd Jul 2016, 11:12 pm

It was something I read back as a student. I used to build oscilloscopes as a schoolkid, so I read everything in the electronics engineering department I could get my hands on. I don't have a copy. Most of the stuff I've kept copies of date to my time in R&D at HP, so Tek stuff was a bit thin on the ground. We sold proper true RMS meters, real thermal converters!

David

kalee20 · 3rd Jul 2016, 11:58 pm

I seem to remember, assessing noise using a 2 channel scope was an idea of someone 'out there' who submitted it to Tektronix in response to a contest for novel scope applications.

It would hardly work with a digital storage scope, of course, with a uniformly bright readout.

G0HZU_JMR · 4th Jul 2016, 1:11 am

Quote:

Anyone tried this method?

Yes, I saw this video via another forum a few years ago.

I tried it using my Tek 465 and a wideband(ish) noise source that goes from 200kHz to 180MHz. This produces an extremely flat noise response across the whole BW at about -86dBm/Hz. I tried using various LPFs after the noise source to limit the BW of the noise into the scope to sensible bandwidths where the scope response was flat.

I got good results when cross checked against a 100kHz to 5GHz Anritsu thermocouple power meter and a Racal true rms meter. However, I didn't copy his test setup in the youtube video because I didn't like the way he had used 'tee' connections to the Hi Z scope inputs. They will look like shunt capacitors (and eventually stubs) once you get well into the HF>VHF region. So the load impedance won't be a constant 50R across such a wide frequency range.

So I used a different method to terminate it all correctly and this helps to minimise the overall measurement uncertainty.

G0HZU_JMR · 4th Jul 2016, 1:21 am

One thing that is probably worth mentioning is what a high level wideband (180MHz) noise source looks like when fed into a scope like the 100MHz Tek 465. i.e. when the full 180MHz noise BW is fed into the scope without any band limiting LPF after it.

This is a precision noise source I made many years ago that is flat to (probably) less than 0.15dB across the full 180MHz range. It uses a NoiseCOM noise source and a chain of carefully selected amplifiers and pads and band limiting filters to keep a +15dB margin for P1dB against the average power of the noise in the given bandwidth. This preserves the integrity of the noise source so it can be used for making valid measurements.

When fed into the scope with such a wide bandwidth, the trace doesn't look like flickery noise anymore. It just looks like a smooth 'fog' across the scope display with the expected probability density. Almost like the focus has developed a major fault and spread itself across the screen or as if the trace has been airbrushed onto the screen with such fine granularity that it looks perfectly smooth.
It looks quite pretty as the foggy look smoothly dims towards the edges of the display.

mhennessy · 4th Jul 2016, 9:32 am

Quote:

Originally Posted by G0HZU_JMR

I didn't like the way he had used 'tee' connections to the Hi Z scope inputs. They will look like shunt capacitors (and eventually stubs) once you get well into the HF>VHF region. So the load impedance won't be a constant 50R across such a wide frequency range.

Exactly

ms660 · 4th Jul 2016, 9:59 am

Had a Google around....I think this is the original article that the video in the link was referring to.....Tekscope, April 1969:

http://w140.com/tekwiki/images/e/e3/...2_Apr_1969.pdf

Lawrence.

3rd Jul 2016, 7:20 pm	#1
ms660 Dekatron Join Date: Apr 2011 Location: Cornwall, UK. Posts: 13,454	Measuring RMS noise With an oscilloscope: https://www.youtube.com/watch?v=zHU9xdU51FY Not sure if I've put that in the right section here? Anyone tried this method? All comments welcome. Lawrence.

3rd Jul 2016, 7:56 pm	#2
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,902	Re: Measuring RMS noise Hmm Tek's applications note isn't bad, but it does make the assumption that the probability density function of the noise is Gaussian. It usually is, and the central limit theorem makes this likely, unless the individual noise contributions are related by say avalanche mechanisms. If you try this to assess the level of noise out of a receiver, you'll find the noise is not as smooth as he has, and unless you slow the scope down to a slow speed, it'll look jumpy, and it'll be too slow to do the eyeball comparison. His demo is using a much broader bandwidth of noise... a factor of about 10,000 times. Where he comes unstuck is right at the beginning by saying what the power level of the noise is. With a noise signal, power is distributed over frequency and so the total amount of power you see depends not only on the noise signal, but also on the bandwidth of what you're using to look at it with. So if you have a wider bandwidth scope, you'll think the noise level is greater. You cannot state noise powers without also stating what bandwidth that noise power is in. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

3rd Jul 2016, 8:02 pm	#3
ms660 Dekatron Join Date: Apr 2011 Location: Cornwall, UK. Posts: 13,454	Re: Measuring RMS noise Have you got or can you point me to tek's application note, I'd be interested to read it. Lawrence.

3rd Jul 2016, 11:12 pm	#4
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,902	Re: Measuring RMS noise It was something I read back as a student. I used to build oscilloscopes as a schoolkid, so I read everything in the electronics engineering department I could get my hands on. I don't have a copy. Most of the stuff I've kept copies of date to my time in R&D at HP, so Tek stuff was a bit thin on the ground. We sold proper true RMS meters, real thermal converters! David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

3rd Jul 2016, 11:58 pm	#5
kalee20 Dekatron Join Date: Feb 2007 Location: Lynton, N. Devon, UK. Posts: 7,088	Re: Measuring RMS noise I seem to remember, assessing noise using a 2 channel scope was an idea of someone 'out there' who submitted it to Tektronix in response to a contest for novel scope applications. It would hardly work with a digital storage scope, of course, with a uniformly bright readout.

4th Jul 2016, 1:21 am	#7
G0HZU_JMR Dekatron Join Date: Sep 2010 Location: Cheltenham, Gloucestershire, UK. Posts: 3,077	Re: Measuring RMS noise One thing that is probably worth mentioning is what a high level wideband (180MHz) noise source looks like when fed into a scope like the 100MHz Tek 465. i.e. when the full 180MHz noise BW is fed into the scope without any band limiting LPF after it. This is a precision noise source I made many years ago that is flat to (probably) less than 0.15dB across the full 180MHz range. It uses a NoiseCOM noise source and a chain of carefully selected amplifiers and pads and band limiting filters to keep a +15dB margin for P1dB against the average power of the noise in the given bandwidth. This preserves the integrity of the noise source so it can be used for making valid measurements. When fed into the scope with such a wide bandwidth, the trace doesn't look like flickery noise anymore. It just looks like a smooth 'fog' across the scope display with the expected probability density. Almost like the focus has developed a major fault and spread itself across the screen or as if the trace has been airbrushed onto the screen with such fine granularity that it looks perfectly smooth. It looks quite pretty as the foggy look smoothly dims towards the edges of the display. __________________ Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 4th Jul 2016 at 1:31 am.

4th Jul 2016, 9:59 am	#9
ms660 Dekatron Join Date: Apr 2011 Location: Cornwall, UK. Posts: 13,454	Re: Measuring RMS noise Had a Google around....I think this is the original article that the video in the link was referring to.....Tekscope, April 1969: http://w140.com/tekwiki/images/e/e3/...2_Apr_1969.pdf Lawrence.