Author Topic: Another ebay noise source  (Read 6916 times)

0 Members and 2 Guests are viewing this topic.

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Another ebay noise source
« on: April 13, 2020, 08:46:53 pm »
I saw this in an EDN article a few days ago and ordered one for $27.83 including tax.  The screen shot showed it fairly flat and it came in a nice aluminum case.  It  arrived today.

969132-0

969136-1

When I first tested it at the specified 12 V, I was mehh.

969140-2

But when I tried some lower voltages it improved.  And after a bit of fiddle at 9.3-9.4 V I got this.

969144-3

So I switched to 2 dB/div for this.

969148-4

I'm not sure my 8560A is flat from 1 MHz to 2.9 GHz to within 1 dB.  I'll be very interested to look at this on my 8566B when I get it set up.

It's rather surprising that it's specified to 2 GHz, but performs well to 2.9 GHz.  The only improvement I can think of is an adjustable voltage regulator.

Reg

Edit:  Major FAIL by Reg!

I was running just above the noise floor of the 8560A.  The BG7TBL required a 40 dB attenuator to avoid blowing out my SDRplay RSP2.  This was before I bought the 8560A.  So I was being very careful not to blow out the 8560A front end.

I looked more closely at the circuit.  It's got an LM317 set to 7.2 V driving the diode, but the 3 transistor amplifier is being fed the full input voltage which explains the large changes in output level with input voltage.  In fact it only goes to 1.35 GHz (-3 dBm), but after lining the case with RF absorber at 10 V in it's flat to 0.5 dBm up to 1 GHz.
« Last Edit: April 26, 2020, 05:35:47 pm by rhb »
 
The following users thanked this post: AF6LJ, Neomys Sapiens, spongella

Offline spongella

  • Contributor
  • Posts: 26
  • Country: us
Re: Another ebay noise source
« Reply #1 on: April 14, 2020, 12:45:00 am »
Thanks for the info,just got one myself, want to try it for return loss testing.  There is a YouTube Video on using the noise source as a substitute for a tracking generator that gave me the idea.
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #2 on: April 14, 2020, 02:43:37 am »
Glad you found it useful.

This is the article that induced me to buy one:

https://www.edn.com/use-a-low-cost-noise-source-as-a-replacement-for-a-tracking-generator/

There are a lot of YouTube videos on the subject of varying quality.

I have one of the BG7TBL units which is rather mediocre plus a couple of others from someone in the UK which cost a good bit more, but are quite good.

The BG7TBL is OK if you are using an SDR as a spectrum analyzer, but if you have something with a greater span, they are not very useful.

I was impressed by the performance of these, but a bit disappointed that the designer didn't put in a voltage regulator that could be adjusted for a flat spectrum and set  at the factory.  It would easily be worth an extra $10-15 to have a voltage regulator trimmed at the factory for flattest response usnig a calibrated SA with a plot included for each device.  Let's hope that someone will do that.

A properly calibrated noise source is too expensive (>$1000) for most  people.  But a spectrum amplitude or power plot from a calibrated SA  would suffice for hobby work.

After my adventures with the BG7TBL "nose source' this was a a real delight.

Have Fun!
Reg
 
The following users thanked this post: sab, I wanted a rude username, spongella

Offline profdc9

  • Frequent Contributor
  • **
  • Posts: 318
  • Country: us
Re: Another ebay noise source
« Reply #3 on: April 23, 2020, 11:49:36 pm »
Do you have a schematic or know what kind of diode it uses for the noise?
 

Offline spongella

  • Contributor
  • Posts: 26
  • Country: us
Re: Another ebay noise source
« Reply #4 on: April 24, 2020, 11:35:21 am »
rhb your results concur with mine, I see a flatter response at approx. 9VDC. My spec analyzer goes up to 1.8 Gigs though. In conjunction with a directional coupler  I hope to conduct some return loss measurements on my different antennas with and without an antenna tuner.  I would only need to look at frequencies from about 1 MHz - 148 MHz  Am in no hurry though. 

profd79, no documentation comes with the noise source but I may open it up and see.  Perhaps Googling the net will yield what diode is used.
« Last Edit: April 24, 2020, 11:36:58 am by spongella »
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #5 on: April 25, 2020, 12:58:14 am »
I had an HP 8970B noise figure meter and 346B noise source arrive yesterday.  It was a bit of an adventure as it was throwing various errors when I powered it up, but I eventually got it so it appears to work properly.

I have no idea if I can actually do this, but it seems to me I should be able to calibrate the 8970B using the table on the 346B and then measure the ENR of the Chinese units at various voltages to +/- 0.25 dB for comparison to my 8560A traces.  If that is the case, I have 4 budget noise sources I'll examine in detail.

I'm not sure when that will happen, but as it does I'll post  updates here.  IIRC I did not have the range of test gear I have now available when I tested the BG7TBL "nose source".  And I bought a couple of low cost sources from a UK firm after that.

I'm off to play guitar.

Have Fun!
Reg
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #6 on: April 25, 2020, 09:34:17 pm »
I found some HPAK app notes about noise source calibration.  Done properly it requires a VNA and a noise figure meter or spectrum analyzer.  I've got all that, but I'd like to find a low budget approach.  I think that a nanoVNA and an 8-12 bit  SDR ought to be able to do a decent job if one works out the correct procedure.  The HPAK recommended procedure requires a calibrated noise source.  That's a bit too steep for an EE grad student.  But it might be possible to use the school noise source.  Better yest would be a first principles solution.

cf. Keysight 5988-7229 &  5089-8056 app notes

I've got an SDRplay RSP2, but in the spirit of minimum budget T&M kit, I think I'll get an RTL-SDR.

RTL-SDR  $30
LCR/transistor Tester $20
nanoVNA $45
noise source  $30
Total $125

It will take some DIY bits and serious thought to get good results from that, but it's probably much better than what the first people working on the devices had available with the resources of large corporations.

Add a  DSO (e.g. Rigol DS1202Z-E) , a cheap AWG (e.g F***Tech FY6800) and a DMM and it's a rather complete EE lab for around $500.  I would have gladly bled for that when I was in grad school in the late 80's.

So what can be done with the above list of stuff?  How do you do it? Enquiring minds want to know.

Have Fun!
Reg
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #7 on: April 26, 2020, 06:16:44 pm »
I blew my initial eval *really* badly.  Sigh.  I was just above the noise floor of the 8560A when I got the flat looking response but failed to notice that which I should have.

The circuit is pretty standard, a diode, 3 transistor amplifier and an attenuator.  It has an LM317G regulating the diode supply which on mine is set to 7.2 V.  The transistor supply is the input voltage which explains the variation in behavior with input voltage.

I lined mine with RF absorbing foam.  At 10 Vdc input it's flat to within 0.5 dB up to 1 GHz and then rolls off steeply reaching -3 dB at 1.35 GHz and -11 dB at 2 GHz.  It rolls off smoothly up to 2.9 GHz which is the limit of the 8560A

First the foam lining job.  I'm not sure this made much difference, but I had the material so I put it in.  I should have taken spectra at 10, 11 & 12 V before and after but I didn't think to do that. 

978514-0

978518-1

Next the output spectrum at 1 dB/div at various input voltages.

12 V input:

978522-2

11 V input:

978526-3

10 V input:

978530-4

For under $30 it's actually pretty good.  But something of a design fail to not have included a 2nd regulator and to specify it as 12 V input.  The transistors are clearly getting rather non-linear above 10 V.  And in typical cheap Chinese tradition, the spec is bogus and should read 1 GHz instead of 2 GHz.

I could not see a marking on the diode and the transistor markings had been sanded off.

I ordered a couple of RTL-SDR dongles, a V 2 and a V 3.  So more fun when those arrive.

Have Fun!
Reg
 
The following users thanked this post: N9XYP

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #8 on: April 26, 2020, 11:49:02 pm »
This has been a considerable education in using the 8560A.  There are enough parameter options to make it difficult to anticipate what will work best in conveying what is going on.

I *think* the noise spike at 13 kHz is the LM317, but have not verified that yet.

Here are a few more SA sweeps all at 5 dB/div and 10 V input.  The only thing changing is the start and stop frequencies:

978792-0

978796-1

978800-2

978804-3

978808-4

But fun with cheap toys.

Reg

Edit: I have now verified that the 13 kHz spike is the LM317.
« Last Edit: April 28, 2020, 01:56:30 am by rhb »
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #9 on: April 27, 2020, 02:41:58 am »
I made this sweep for a friend who is into MW DXing.  Since it took 2 hours to collect, I thought I'd add it here. 

978846-0
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #10 on: April 28, 2020, 01:45:16 am »
For convenience and to make the comparison as close as possible, here is the BG7TBL "nose" source at 10 V input with the SA parameters as close as possible to post #8.  I don't see any point to reproducing post #9.

979886-0

979890-1

979894-2

979898-3

979902-4
« Last Edit: April 28, 2020, 02:00:52 am by rhb »
 

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 3282
  • Country: ua
Re: Another ebay noise source
« Reply #11 on: April 28, 2020, 02:28:54 pm »
Some time ago I bought this noise source with discount coupon:
https://www.aliexpress.com/item/32845101232.html

it cost me about 9 USD and works pretty good.

There is a small issue below 1.5 MHz, but above that it works very well
« Last Edit: April 28, 2020, 02:40:15 pm by radiolistener »
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #12 on: April 28, 2020, 03:56:54 pm »
Well, for that price I'd like  to get one.  Err, *strictly* for academic research purposes.  I do *not* have a TEA problem.   I *really* needed the HP 4145B & 16058A semiconductor test set I bought yesterday.  Honest. ;-) 

Unfortunately, they don't accept PayPal.

I'm really at the limits of what I can test with the 8560A on this unit. Here are  1000 trace averages from 200 kHz to 1 GHz at 1 dB/div.

First, the noise floor of the 8560A:

980336-0

The new noise source over the same range:

980346-1

And finally from 200 kHz to 1.5 GH:

980340-2

The 8560A has not had a cal in a long time so I have no idea what the true deviation from flat is.  That's actually one of my goals,  automated full cals every year of all my gear from a minimum number of externally calibrated devices.

The next step will be to calibrate the 8970B and repeat the measurements with it as well as determine what the amplitude test and calibration procedure is for the 8560A.
« Last Edit: April 28, 2020, 04:06:29 pm by rhb »
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #13 on: April 28, 2020, 04:28:08 pm »
A couple more photos at 5 dB/div and 10 trace average with default RBW to see what the ENR is across the 1 MHz to 2.9 GHz range.

The noise floor with power off:

980370-0

And then power on at 10 V:

980374-1
 

Offline 0culus

  • Super Contributor
  • ***
  • Posts: 3032
  • Country: us
  • Electronics, RF, and TEA Hobbyist
Re: Another ebay noise source
« Reply #14 on: April 29, 2020, 02:32:51 am »
Reg,

Have you tried using the noise marker functionality available on your spectrum analyzer? This will show you the noise level at the marker, normalized to a 1 Hz bandwidth and is a more accurate characterization of the actual noise floor.

https://www.keysight.com/main/editorial.jspx?cc=US&lc=eng&ckey=703435&nid=-35183.0.00&id=703435
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #15 on: April 29, 2020, 03:12:35 pm »

Having bought an 8970B & 346B I thought my time would be better spent learning to use it than using the 8560A for detailed measurements.

With a full cal, the 8970B is supposed to be accurate to +/- 0.15 dB which is rather better than  the +/- 1 dB accuracy specified for the 8560A.  Even without a full cal it's supposed to be accurate to 0.25 dB.

I have used the marker some, but mostly just to determine frequencies such as the presumed LM317 noise at 13 kHz.
 
I've built a GPIB controller based on WaveyDipole's AR488.  So I think automating calibration and characterization of noise sources with the 8970B has more benefit long term.  That should allow me to check the 8560A.

A major effort of mine is a system of switches and attenuators under GPIB control to do an automated annual lab cal.  I've got most of the relays, bulkhead connectors and such.  I'll probably use something like a 16501A chassis to hold all of it so it looks at home with all the other HP gear.  Goal is to use a minimum set of standards with external calibration either commercial, the USA Cal Club or some mix of the two.

The main reason I've been using the SA for this is it's set up for use.  And it also provides a good display with no extra effort.

I'd not looked up the 8560A amplitude spec until a few minutes ago, but obviously there is no way to tell if the apparent variation in output level is the noise source or the SA.  That's quite remarkable for a $28 ebay gadget. It just needs a 2nd LM317 to regulate the input and a barrel plug to BNC-F so I can connect it to the 28 V pulsed output of the 8970B.  Probably easiest to put that in a separate box.  I don't have any PCB design tool skills and thru hole would be somewhat difficult to fit in the available space.  But perhaps someone sells a little LM317 based adjustable regulator board.

Have Fun!
Reg

 
The following users thanked this post: 0culus

Offline 0culus

  • Super Contributor
  • ***
  • Posts: 3032
  • Country: us
  • Electronics, RF, and TEA Hobbyist
Re: Another ebay noise source
« Reply #16 on: May 01, 2020, 05:16:24 am »
That's fair. If you have the right tool for the job, better to use it. Thanks to this thread, I'm now working on falling down yet another TEA rabbit hole.  :-DD
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #17 on: May 01, 2020, 03:00:23 pm »
These are my objectives for this project:

1) calibrate a full span SA sweep using an RTL-SDR or nanoVNA

2) measure noise figure using an RTL-SDR or nanoVNA

3) calibrate full span amplitude performance of the nanoVNA

4) calibrate cable attenuation measurements with the RTL-SDR or nanoVNA

with comparisons to HP kit performing the same tasks and detailed evaluation of the noise source performance. Flatness, stability, temperature coefficient, etc.

I think with modest effort and changes this diode noise source should be capable of 0.5 dB or better accuracy from 200 kHz to 1 GHz. I'm hoping to persuade the OEM to add voltage regulation for the transistors and a calibration table and stay under a $50 price point. Perhaps add a thermistor and an FET or two to temperature compensate the voltage regulators if needed. I suspect that is not actually needed though and that it will be stable enough once warmed up.  Unlike the BG7TBL unit, this one does not get hot.

Have Fun!
Reg
 

Offline spongella

  • Contributor
  • Posts: 26
  • Country: us
Re: Another ebay noise source
« Reply #18 on: May 03, 2020, 12:37:35 pm »
Well I have attempted using the noise source in lieu of a tracking generator to make return loss measurements.  No luck, even though a YouTube video shows differently.  I connected the noise source to a return loss bridge, HF antenna and used a spectrum analyzer and a SDR-dongle to view the response.  The noise source is definitely working but I see no differences when I attach the antenna.  I am using a 50 Ohm dummy load as a reference. 

I will try to take a picture of how I attached all the components together and will not rule out the fact that I messed up hi hi.  Thanks.
 

Offline OwO

  • Super Contributor
  • ***
  • Posts: 1250
  • Country: cn
  • RF Engineer.
Re: Another ebay noise source
« Reply #19 on: May 03, 2020, 12:57:32 pm »
Well I have attempted using the noise source in lieu of a tracking generator to make return loss measurements.  No luck, even though a YouTube video shows differently.  I connected the noise source to a return loss bridge, HF antenna and used a spectrum analyzer and a SDR-dongle to view the response.  The noise source is definitely working but I see no differences when I attach the antenna.  I am using a 50 Ohm dummy load as a reference. 
I just don't see the point of this approach (or even a tracking generator) when you can simply build a one port VNA for under $20 in parts (if you forgo having a display and port 2). ADF4350 x2, resistive bridge, IAM-81008 mixer, and STM32F103/GD32F103 with built in ADC will do the job far better than even a decent SA + TG setup.
Email: OwOwOwOwO123@outlook.com
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #20 on: May 03, 2020, 06:18:20 pm »
The real point is to master the physics, mathematics and technological limitations of various techniques.  There are a lot of "obsolete" methods which are very much worth understanding and being able to apply.  You can almost always buy a solution, but knowing how to use what you already have is much more fun.  And can be critical if your time frame or budget doesn't allow buying an instrument designed for the task.

To paraphrase someone's sig with reference to sail planes.  "There is no substitute for span."  "Yes, there is.  Skill and imagination.  But you can buy span."

I very much cherish my knowledge of machining set up techniques from over 120 years ago. They may be too time consuming for regular  industrial use today, but they also provide the key to achieving accuracies which are hard to get even today.

I rather doubt that the ADF4350 mashup will equal an HP 8560A.  Needs some low pass filters.  I'd love it if that were true.  That's somethingt I want to see on the market.

Show everyone and post all the data to reproduce it and I'll pay you twice the BoM via PayPal for something that simply works and 5x if it beats an 8560A w/ TG.  Even more if it gets cloned to the extent of the $20 LCR/transistor tester and the nanoVNA.

Have Fun!
Reg
 
The following users thanked this post: spongella

Offline profdc9

  • Frequent Contributor
  • **
  • Posts: 318
  • Country: us
Re: Another ebay noise source
« Reply #21 on: May 03, 2020, 07:08:01 pm »
 
The following users thanked this post: spongella

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #22 on: May 03, 2020, 09:12:01 pm »
Unlikely.  The ADF435x are square wave sources.  So you need two with low pass filters and a mixer to produce a clean sine wave output as OwO hinted at, but should have stated more explicitly.

The fact that unit  starts at 33 MHz says it's a single ADF4351.  A dual ADF4351 unit would start at a few kHz at most.

I've not kept up with SaabFan's thread, but a low cost (<$100) 10 kHz - 2.2 GHz synthesizer would be a huge boon to a lot of hobbyists.  It's really the next major step after the Arduino based LCR/transistor tester and the nanoVNA.

So far as I know a good RF source is a "sine qua non" for calibrating a noise source on a DIY budget.  It's necessary, but not sufficient.  A decent power sensor is also required.

I don't have time to do this myself, but if someone has the time and skills I'm interested in discussing providing support for an OSSW/OSHW RF synthesizer development.

Have Fun!
Reg
 
The following users thanked this post: spongella

Offline profdc9

  • Frequent Contributor
  • **
  • Posts: 318
  • Country: us
Re: Another ebay noise source
« Reply #23 on: May 03, 2020, 10:01:34 pm »
The device does have two ADF4531. 

https://om0a.cq.sk/en/amaterske-radio/jtgp-1033-d6-spektralny-analyzator-s-vf-generatorom-32-4400-mhz/

It generates a IF signal that goes to a log detector and is demodulated.  It filters harmonics in the same way that the VNAs do because these generate harmonics of the IF.  It seems to be very much like a heterodyne VNA must using log detection rather that detecting I and Q.
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: Another ebay noise source
« Reply #24 on: May 04, 2020, 12:32:13 am »
Interesting.  I'd not seen that.  Have you found  the schematic?  While interesting,  I don't think it is mixing 2.2-4.4 GHz from 2x ADF451 devices to get ~DC to 2.2 GHz as the mixer output.  That's what is needed for a good synthesizer.

I think it's using one ADF4351 for the source and one for the LO.  While that works as demonstrated by the nanoVNA, VNWA and others,  it also leads to spurious responses.  I noticed in one YouTube video the narrator was calling them "harmonics" which they are not.

I'd really like to see the RF output spectrum at 33 MHz on a lab grade SA.  I've got a moRFeus and the output spectrum is awful at low frequencies because of all the harmonics.

This is the MoRFeus spectrum at 100 MHz.  It's lowest output frequency is 85 MHz.

983682-0

The MoRFeus is an ADF4351 and mixer.  With external filters it's a useful device, but I'd really like to see a proper synthesizer with spurious outputs - 60dB or better.

Reg
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf