UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Closed Thread
 
Thread Tools
Old 8th Feb 2020, 12:33 pm   #21
Malcolm T
Hexode
 
Join Date: Dec 2018
Location: Lugo, Spain
Posts: 477
Default Re: Analyser required suggestions and rev bias diode noise gen.

Thats a nice lot of very interesting reading there , interesting about Mr Gordon Edge at PYE , learn something new everyday here .

Thank you
Malcolm T is offline  
Old 8th Feb 2020, 5:29 pm   #22
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Analyser required suggestions and rev bias diode noise gen.

Quote:
Originally Posted by G0HZU_JMR View Post
Note that the technical description for ENR in the first article doesn't look right to me. But this doesn't really matter for higher ENR numbers like 31dB. The correct definition/calculation matters a lot more if the ENR is as low as 5.6dB for example.
Indeed. The definition that used to be used wasn't really the excess noise above actual thermal noise. It was the excess noise above a standard temperature thermal noise. This introduced an error in temperature correcton. Small for high ENRs and temps close to 290K.

When introducing new noise sources with built-in thermometers and more stable diode current control, we switched ENR definition to mean excess nois ABOVE actual temp thermal, where the excess is quoted as dB above 290Kelvin kT.

This allowed slightly better correction for noise source body temperature. This change went into the new ENR calibration rigs which were an associated product. (Noise/Com bought one so the change is in their diodes too)

Well spotted!

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 11th Feb 2020, 1:50 am   #23
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Analyser required suggestions and rev bias diode noise gen.

Yes, it's easy to get in a muddle with ENR figures...

If the ultimate aim is to produce a boosted noise level from a noise diode then it's worth doing a few calculations that predict the average noise power over the bandwidth of the noise source otherwise it's possible to degrade the quality of the noise source if the amplifier stages introduce too much compression on the noise peaks.

At work, the old rule of thumb was to make sure that the average power of the noise over the full bandwidth of the noise source should be at least 15dB lower than the output 1dB compression point of the amplifier(s).

My old high level noise source has a 50 ohm output impedance and 180MHz bandwidth. I'm pretty sure it has an ERA5-SM output amplifier and this has a P1dB of about 18.5dBm. Following this is an internal 3dB attenuator at the output. This gives a very good source VSWR.

To meet the above rule the average output power of the noise source should not be greater than (18.5 - 3 - 15) = 0.5dBm.

It actually puts out about -3dBm when measured on a thermocouple based power meter although this is adjustable over a range of a couple of dB. If the bandwidth of the noise source is about 180MHz then the noise level in a 1 Hz bandwidth should be -3 - 10*log(180e6) = -85.6dBm/Hz when measured on a spectrum analyser that has a noise marker feature.

On my Tek spectrum analyser the noise level typically measures -85.7dBm/Hz as the marker is moved across the band which is very close to the predicted level. It measured -3.1dBm on the power meter and this has a 100kHz to 5.5GHz sensor head.

I had a go at measuring the crest factor of the noise source using an old HP 500MHz digital scope and it indicated the crest factor was about 6.3. I'm not sure how well this old scope can perform this measurement but this result of 6.3 (= 16dB) is probably average for a general purpose lab noise source.

A precision lab noise source might manage a crest factor of 8. It is really easy to spoil the crest factor of a noise source especially when trying for a high noise level. If the noise source is only going to be used to measure filter responses then this won't matter much although things become more critical when doing BER testing on systems using digital modulation.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 11th Feb 2020, 9:02 am   #24
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Analyser required suggestions and rev bias diode noise gen.

Theoretically, all noise levels will have peaks which can reach infinite voltage, in both directions. This means that ALL real-world noise has limited crest factor, and it also means that you cannot assess the level of a noise signal by looking at the peak voltage. Only statistical methods can get a grip on it.

Where the noise creation mechanism is the sum of a great many small, similarly sized contributors, the central limit theorem says the probability density function will tend to Gaussian. So, everyone always assumes noise is Gaussian, which leads to the rather interesting circumstances when it isn't, being overlooked.

Human perception of white noise in analogue systems seems to follow the RMS level.

As Jeremy says, in digital systems, it's the peak values which do the greatest damage. In something like a QAM modulated transmission, if peaks aren't big enough to cross into the next state, you get no errors, so crest factor limitation makes the Eb/N0 curve stop dead to zero errors instead of the theoretical decline. Noise peaks, if not clipped can generate multi-bit errors on occasions and the classic text books on digital radio give the standard analyses... but limited crest factor makes things massively different.

The HP 3708A is one of their more obscure instruments. It would take the IF of a microwave radio link receiver, measure its level (true RMS thermal converter) and create band limited noise of a level that tracked it, giving a servo-controlled signal to noise ratio for assessing the liability of demodulators to create errors versus noise level. In designing it, we had to be very, very careful to always maintain good crest factor. I had to design filters which maintained constant noise-bandwidth over temperature for this.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 11th Feb 2020, 12:23 pm   #25
peter_scott
Dekatron
 
peter_scott's Avatar
 
Join Date: May 2003
Location: Edinburgh, UK.
Posts: 3,273
Default Re: Analyser required suggestions and rev bias diode noise gen.

This Spectrum Analyser works well. I'm using Windows. I haven't checked if there is a Linux version but I guess you could always use Wine.

Peter
peter_scott is offline  
Old 11th Feb 2020, 2:59 pm   #26
IanG3XYV
Pentode
 
Join Date: Sep 2014
Location: Bedford, Bedfordshire, UK.
Posts: 172
Default Re: Analyser required suggestions and rev bias diode noise gen.

I've been enjoying this discussion and the contributions from our RF experts.

I have an HP8970a Noise Figure Meter but it didn't come with an HP346 noise source so I built a quick source using an RF transistor in avalanche mode which was good enough to prove that the NFM was working. I found a reasonably priced noise source pcb here:

https://g8fek.com/noise-gen.html

I've got the BBgen which should give a reasonable flat output up to 5Ghz. I've only just got it housed in a diecast box and feel I should do something more serious with that and the 8970 combination. I shall be rereading the thread carefully. I'd like to categorise the BBgen ENR levels across the frequency range. There are also some good application notes on the website above.

Ian
IanG3XYV is offline  
Old 13th Feb 2020, 12:39 am   #27
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Analyser required suggestions and rev bias diode noise gen.

Interesting development alert...

I managed to find some BAT-17 Schottky diodes and had a go at making the noise source suggested by John back in post #5.

I cut out a small piece of double sided PCB and then cut the required tracks/pads with a scalpel. I used an SMA end launcher connector for the RF output. I must say I'm quite surprised how high the noise level is from this diode! I biased it at just over 1mA and it produced -133dBm/Hz with a fairly flat response from LF up to 1GHz. That is a lot of noise for a basic and cheap diode!

At the moment there is no matching pad at the output so the source VSWR will be quite poor. It should be a lot better with an attenuator at the output but this will obviously be at the expense of the noise level. The performance is so good I really ought to make the PCB 'properly' with an accurate milled PCB with modern SMD parts to try for the highest bandwidth. At the moment I'm using 0805 SMD parts salvaged from a 30 year old dev PCB from work. These could be smaller or better if I swap across to modern SMD parts. I think the response could be made to be quite flat from LF through to UHF.

I'll try and do a few tests in the coming days to see how stable the noise level is over temperature and bias current. I feel that the bias current setting will be quite critical. I have no idea how rugged this diode will be when biased into avalanche like this but these diodes probably only cost about 30p each.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 17th Feb 2020, 10:36 pm   #28
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Analyser required suggestions and rev bias diode noise gen.

I finally managed to get some free time to look at this BAT-17 noise source again and I did a few quick tests over temperature at 28MHz.

To measure the noise level I used a real time spectrum analyser with the noise input fed direct to the digital IF section and I took measurements at 28MHz. This analyser has a decent internal self calibration system and there is no RF conversion involved anywhere. So the noise level readings should be very reliable over time.

Here at home I don't have anything exotic like a Thermotron chamber but I do have a fairly reasonable test chamber for testing small circuits over a temperature range of about 15degC to 40degC. This has a fan assisted heat source to blow air around the test chamber and this is controlled via a PC and a thermocouple temperature probe inside the test chamber. The heat source is adjusted by the PC as required to keep the air temperature constant at the desired set point temperature. It can usually hold the circulating air temperature within +/- 0.3degC for long periods and the temperature can be adjusted in 0.5degC increments if required.

I tested at 15 to 35degC in 2.5 degree steps and I adjusted the bias current to be 1.50mA at every temperature step. This only needed a small change in the bias voltage. I didn't use quite the same resistors as John mainly because I just fitted what I could salvage from an old PCB. I used 1k ohm for the bias feed to the diode and 220R from the PSU. I also used 100nF caps for the coupling and the decoupling.

At the end of the tests (up at 35degC) I turned the test chamber back down to 25degC as a system check and it gave the same -132.4dBm/Hz reading as it did earlier. This is a good confidence check against any slow drift that might be happening in the test gear over time.

The results are given in the image below and the stability is quite good across 20-25degC. What isn't so good is the response vs frequency. There was over 2dB of ripple from LF through to 40MHz when tested at room temperaure. Above this the response is fairly flat out to 1GHz. It's a shame it has a poor response below about 40MHz. At some point I'll try measuring the stability vs temperature up at higher test frequencies.
Attached Thumbnails
Click image for larger version

Name:	BAT17_tests.gif
Views:	68
Size:	7.6 KB
ID:	199205  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 18th Feb 2020, 12:11 am   #29
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Analyser required suggestions and rev bias diode noise gen.

As a comparison, I tried testing my 'raw' Noisecom noise source. This is just a 28V biased Noisecom diode followed by an attenuator and I tried testing the noise level across a range of 22.5 to 32.5degC. The result was that I couldn't measure any change in noise level across this 10 degree change in temperature although this is really pushing the limits of my test gear.

It does help to prove that my test method with the real time analyser and the temperature chamber is reliable because I think the spec for this Noisecom diode is <0.01dB change per degC so it should be very stable across all typical room temperatures.

The spec for flatness was something like +/- 0.5dB to 500MHz but it seems to be a fair bit better than that. However, it is quite difficult to make a reliable measurement for something like this.

The BAT-17 noise source is probably at its best up at VHF and into UHF and I'll try and do some more tests on it. Overall, it is performing much better than I expected...
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 8:00 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.