Russian Noise diode/tubes, anyone use these?

[rwgast_lowlevellogicdesin]

Ive been looking in to either making a homebrew tracking gen, or a really nice flat broadband noise source. Each needs to work to 2ghz. So my thinking on a noise source is to reverse breakdown a zener or an RF transistor and then through a few era5 mmics after it. Looking on ebay I found this kind of stuff

http://www.ebay.com/itm/4x-2D3B-USSR-Noise-Diode-Miniature-Tubes-NEW-/201434297863?hash=item2ee66b7607:g:nesAAOSwWnFWAAW4
http://www.ebay.com/itm/2D2S-Russian-noise-diode-tube-NOS-Lot-of-2-/262075471491?hash=item3d04ea5683:g:YSkAAOxyrUVRnKA3

Anyone know much about this stuff? I cant find ny data sheets, projects etc etc. They look cool but im not even sure
 they are what I think they are and if they would make a good noise source.

[T3sl4co1l]

The noise level is quite low, essentially shot noise at anode current.

They're suitable for testing receiver NF but not so much a stochastic TG, not without a lot of amplification (and adjusting for flatness).

Here's a typical European example:
http://www.mif.pg.gda.pl/homepages/frank/sheets/009/k/K81A.pdf

Still, not much data.  You're expected to set Va to a typical value, and adjust Vf/If to obtain the desired noise power as measured by a meter.

Tim

[rwgast_lowlevellogicdesin]

Thank you ill definately check out noisecom for sure!

As far as the second option I have read up on a lot of that stuff and in the end I will most likely go down that path, unless something easier and more esoteric pops up either from noisecom, russian tubes (that have some documentation..), or anything else I may find. I made a few active VLF/HF antennas and still have a few of these metal 2n5109 transistors left they are 1 watt with a 1.2ghz Ft, I figured I would experiment using them as the reverse bias noise source, but to tell you the truth I really do not understand reverse break down and its relation to white noise so im not sure if using one over a 2n3906 is just a 4 dollar waste.

Lastly if anyone knows anything about these russian tubes please chime in, there so cheap I would just pull the trigger but in reality with no data on them im 90% sure they will just end up in my RF junk box and never get used when I could have spent the money on some low loss semi rigid coax or a GPS module that will definately get used.

[uncle_bob]

Hi

If you just wire up the tubes or the Noisecom diodes, they will be flat into the 30 to 100 MHz range. You need to get fairly tricky to get them anywhere near 2 GHz.

We went over all this in detail in another one of your threads on the same topic I believe ...

Bob

[IconicPCB]

http://www.russiantubes.com/prop.php?t=12&p=32

[uncle_bob]

http://www.russiantubes.com/prop.php?t=12&p=32

Hi

You most certainly are not going to get to 2 GHz with that tube. The microwave tubes are *much* different than that package.

Bob

[LaserSteve]

Microwave hams use a rf transistor in a stripline package that is biased to breakdown. With the right device and matching, followed by a MMIC, you can get considerable bandwidth.

Steve

[uncle_bob]

Microwave hams use a rf transistor in a stripline package that is biased to breakdown. With the right device and matching, followed by a MMIC, you can get considerable bandwidth.

Steve

Hi

The OP has at least three threads going on that sort of approach

Bob

[chris_leyson]

Noise Factor(dB) for a temperature limited thermionic diode is given by 10log(20*Ia*Rload) where Ia is the anode current in amps and Rload is typically 50 . I've used MOV A2087 (CV2171) in the past as a noise source, good for 10dB excess noise and maybe good up to 50MHz depending on layout. CV2341 is a coaxial 70 noise source, 20dB excess noise and maybe good to 2.5GHz according to the data sheet. Even if you managed to build a noise source using a CV2341 exactly as per the data sheet you would have to measure the excess noise as there is a good deal of uncertainty above 200MHz, and even more uncertainty in a 50 system. I wouldn't trust it at 2GHz.

[uncle_bob]

Hi

Having built and tested a number of these circuits .... There are no shortcuts past VHF. Even with a well equipped lab, it is a hard thing to build and get right.  Getting to 2 or 3 GHz with tight flatness is not easy. It is one of those things you buy rather than build if you only need one.

Bob

[rwgast_lowlevellogicdesin]

@uncle_bob

Sorry if my post is irritating you.... It is true I did post something about this MONTHS ago, but it was a bit more broad this post was acually about these russian diodes themselves and not meant to turn in to a general thread on noise sources...

So why the interest MONTHS later again, well they sell cheap noise sources on e-bay and I can find very little info on them but what I can find is not promising at all. I bought a Ham It Up, HF up converter, not because I care about HF but the version 3 has a built on noise source and I read that someone wished they had bought a Ham It Up instead of the 20 dollar e-bay special. Problem is there is no information on the damn noise source, I dont even know who to e-mail about it, it is OSH and NoElec makes the boards but did not design it and they have not answered any e-mails on the specs. It seems like a POS to me from what I can tell it loses power every decade or so and then plummets around 300mhz or so. I am really not sure if it the RTL SDR or the Noise source since I can get no info from NoElec... I just ordered an AirSpy and besides having a nice low noise floor with out any spurs or imaging, it can do a spectrum sweep in fo 24mhz-1.8ghz in 1 to two seconds depending on settings, with 12 bits of resolution and great intermod rejection and low noise floor I would say this is pretty competitive with a real spectrum analyzer if not better than an older piece of gear green screen gear that still costs 100+

The deal is I am really in to L Band right now, and even some hydrogen line stuff. I really want to start designing filters and other stuff for 70cm up. That alone is a challenge and honestly I would rather not waist time building a noise source or tracking generator. Sadly I am in no finacial position to buy a tracking gen/2ghz rf signal gen :/. Basically I am hoping to find something I can either have OSHPark make a PCB and soldier together or even some schematics. I was actually hoping the "Russian Noise Tubes" were some weird broadband thing invented during the cold war to jam the whole spectrum or something... I should also correct myself, I dont need a perfectly flat source, a gradual power slope is fine or even a few humps with in a few db, just as long as it is uniform enough to work with....

As a side note I actually have completed Sciene Work Shop spectrum analyzer kit, it uses TV tuners to get from 1-500mhz, then another tuner to get to 900mhz. When it was donated to me it was not finished and had no documentation I hardly knew what I had. Anyways I ordered the book from the guy and it turned out it was just a bunch of articles compiled by hams on different designs including one using the same parts as there kit. The site has a 100 tracking gen option and I am waiting for him to get back to me but I will bet it is only 500mhz. I would love to use this thing so I could have an instantaneous bandwidth of 900mhz, and two SpecAnns for certain tasks. The issue is the software I payed 50 dollars for that uses your sound card and PC as a display does not work on modern PC's and he contacted the author who refuses to release the source or even update it. Basically it was intended to use with an analog scope in X,Y mode my issues is I use a DSO and X,Y shouldn't even be built im mode it so awful. I have an HP 500KHZ analog in the shed that works, the biggest issue besides a 3 inch screen is it is a freakin 75lb rack mount unit that has no place in my work area, or good use for that matter... Would anyone know of some software that will read signals formatted in X,Y for vector displays and render them on your PC, X,Y is the only reason I dont use a curve tracer too!

[IconicPCB]

To add a bit more to the cacophony of the thread...

I would give BFR93 base emitter junction a go.
A microwave tranny ( Ft 6GHz, emitter base capacitance in the picofarad region...), just what is needed and best of all ...BFR93 WILL NOT CHALLENGE THE WALLET

[uncle_bob]

Hi

The problem with starting a new thread rather than cintinuing the old one is that all the same stuf is gone over again and again. Ultimately all the same questions get asked. Eventually the whole project will be explained again. Similar threads here that go on for years work very well. People who are new to the thread can go back through the previous 20 pages and see what is what. With the same topic split into many fragments it is of use only to you. Others will have a very hard time with it....

So, yes, there is a reason for my comments.

Bob

[CJay]

http://www.rfdesignuk.com/

[rwgast_lowlevellogicdesin]

@Bob, I do understand your point, and will try to stick to keeping similar posts in the same thread from here on out. I think I am just desperate and I justed want Russian noise Diode to stick out to anyone who has ever had experience with them!

@Post above..... I need to design L Band filters and equipment, your tool is only for 800-900mhzish, secondly its 900 dollars!!!! That is hardly in my budget and if it were I am sure I could pick up a second hand sweeping gen for about that much, which would be much more usefull.

Can someone tell me why something that perfectly fits the bill for me as far as a noise source is so much!!

$1000 2ghz Noise source box http://www.ebay.com/itm/Noise-Com-NC6111-2GHz-Noise-Generator-/331651589355?hash=item4d37f9a4eb:g:z8gAAOSw9r1V8fPH

Then there are a lot of big sma tubes stamped with the big Agilent/Hp/Keysight etc that range from 10mhz-18/26ghz and they are like $2500!! While I could pick up a very clean and nice 2GHZ Sweeping RF SigGen for $950

http://www.ebay.com/itm/Keysight-Agilent-E4420B-ESG-2-GHz-RF-Signal-Generator-/302012337616?hash=item4651568dd0:g:WuMAAOSw5VFWLWLo

I honestly dont care how I don it im just trying to string together a poor mans SNA that can get to 2ghz, So I can test filters and antennas.. but im baffled where a noise source would be more valuable than a full on quality keysight signal generator... The only reason I am looking in to noise sources is because I figure it will be quicker, cheaper, and easier to build than some kind of crazy microwave sweeping gen, Working on audio circuits using a nice 24bit 196khz sound interface, I can either sweep or run white noise from an output to DUT and then to an input, with very clean resolution and lovely dynamic range, and in my experience, tracking with white noise works fine but a real frequency sweep is a much better picture of the DUT. 

This also got me thinking, and I have not seen it so Im betting its just not feasible... but could you run a soundcard output in to a PCB that could do a bunch of frequency multiplication with rf chips or something? Or maybe you can just fine free sig gens in a test equiptment landfill:)

[uncle_bob]

Hi

The reason what you are looking for costs so much is that it is not easy to do *and* not many of them are sold. The practical answer is to spend significant time shopping and pick up a surplus gizmo at auction. You may not win every auction, you may not get a good one every time you win an auction. You will spend $100 on a $2,500 gizmo each time you try.

Bob

[G0HZU]

Looking at your requirements, I don't think a noise source is what you need.

However, if you want to know why the Noisecom 2GHz noise source is on sale at $1000 on ebay then it's a case of supply and demand. If you 'need' one of these they are in limited supply on the used market. So prices will remain firm compared to general purpose signal generators that are on sale every day in large qty.

Also, the spectral density of that NS is about -80dBm/Hz. So across a 1GHz BW that means the average power across its 1 to 2GHz BW will be 10*log(1e9) dB higher = 90dB higher.
So it probably churns out +10dBm of total average noise power. This needs a very special output amplifier if the goal is to produce 'proper' noise because you need a 14dB margin from P1dB in the amp. Note that it is only spec'd to produce noise across 1GHz to 2GHz so that particular model has limited use. No good for 433MHz for example.

So the output amp will need to be rated at several hundred mW (probably 1W) and it will have an attenuator after it to help improve the source match. It will also have some expensive RF step attenuators and careful design of the signal path to keep the noise floor level across the band.

But for your use, -80dBm/Hz is only going to give you a limited dynamic range with a typical low budget spectrum analyser. You would be much better off buying a used VNA. Something like an old HP871x. The prices of these vary a lot but they are getting cheaper and cheaper.

Or you could take a chance on one of the various miniature USB VNAs out there. These should be good enough for general design work but they won't be as good as a proper old school HP VNA in terms of accuracy or dynamic range. But the noise source and spectrum analyser approach will be even cruder with fairly limited performance over a limited dynamic range.

[uncle_bob]

Hi

Maybe another way to look at it:

None of the major instrument manufacturers start with a noise source to build an analyzer. The reason is cost. It is cheaper to do other ways.

Bob

[rwgast_lowlevellogicdesin]

Well maybe one of you has a ham it up, this is an RTL-SDR sweep at unity gain, blue bar is 75ohm term, green is the noise source. BTW is there anyway to convert the graph to peek output in DBm i dont understand the conversion from root hertz, im more familiar with that being a measurement used as the noise floor of an opamp



As you can see it is almost impossible to measure an FM trap.... thats a bit ridiculous, im not sure why they added this as a feature....

[uncle_bob]

Hi

Given the way the RTL-SDR works, you are not going to get very good performance out of it in this sort of application. It simply is not designed to do this sort of thing.

If you want to do it:

      First sweep the green line with a short in the system. It will look something like the green line you show.

      Next sweep the green line with your DUT in the system. It also will be close to the green line you show.

      Now subtract one from the other. Don't worry about the root Hz stuff, it drops out.

Since the two lines are swing up and down by 10's of db's very often, the net result is not going to be pretty. The best you can do is get down to the blue line. The dynamic range is not going to be very great.

Bob

[rwgast_lowlevellogicdesin]

Do you think this is just an issue with cheap dongles, and when my airspy gets here (much better 12bit dynamic range 20db lower noise) it will work much better or will everything be flopping all over like the picture above?

Thanks for the tip on doing the subtraction tip to sweep a short and then a DUT and subtract the noise ill see how that works out.

[CJay]

@Post above..... I need to design L Band filters and equipment, your tool is only for 800-900mhzish, secondly its 900 dollars!!!! That is hardly in my budget and if it were I am sure I could pick up a second hand sweeping gen for about that much, which would be much more usefull.

If that was aimed at me, my bad, I got the wrong link (same guy, two sites with different target markets for products).

Try this one, BBGen+ 0-6GHz noise source for £59, the BBGEn is slightly lower spec and is cheaper still.

http://www.g8fek.com/BBGen+.htm

http://www.g8fek.com/BBGen.htm

[Dago]

Here is my implementation of a noise source from 0~3 GHz with an AT41486 transistor, maybe it is of help to you http://www.dgkelectronics.com/microwave-noise-source/

[uncle_bob]

Do you think this is just an issue with cheap dongles, and when my airspy gets here (much better 12bit dynamic range 20db lower noise) it will work much better or will everything be flopping all over like the picture above?

Thanks for the tip on doing the subtraction tip to sweep a short and then a DUT and subtract the noise ill see how that works out.

Hi

To be useful for normal work, you need at least 60 db of dynamic range across the entire span. Most commercial gear is in the 120 db range. You aren't going to get there with a dongle. There are a *lot* of surplus units out there. Get one of them.

Bob

[LaserSteve]

I have one of the Chinese EBAY noise boards.  It works, but the variant I have uses a switching voltage regulator that is a bit undersized and runs hot.  So the first thing you have to do, besides building a shielded  case for it, is upgrade the VR to external power if possible..... The one I have has very good bypassing at each stage and good RF layout.

http://www.ebay.com/itm/NEW-12V-RF-noise-source-Simple-spectrum-external-tracking-source-generator-noise-/231729180565?hash=item35f422af95:g:rmwAAOSw1KxXNVF2

Yes, its a noise source. Is it flat, who knows, does it work well to 1.5 or 2 Ghz, or so... Yes...

I'd rather use a comb generator  for sweeping filters with my cheap CATV Service grade SA.  Even my home  AVCOM PSA-37D, as ancient as it is, has far more dynamic range then the TV tuner SA kit or one of the RTL dongles....  They go for about 350$ on Ebay..   I'm not suggesting you buy a 37D unless its working and you can get it for a song. There is far better hardware out there.   

With the CATV Grade SA and some carefully chosen  downconverters, I can approach what I can do it work with the good SA.  However that comes from years of  practice  with the hardware.  I also have quite a selection of attenuators, comb generators , PTS synthesizers,  and filters at my disposal.

Rather then building  a noise source, consider a working PTS160 or PTS250 from Ebay, and use harmonics... The PTS tuning in one Hertz steps is VERY useful.  Unless you are making really sensitive receiver front ends or preamps, you will appreciate a harmonic source or comb generator,  over a noise source, when using low cost SAs... A PTS followed by a MMIC buffer driving a comb generator or a Minicircuits  mixer module wired as a doubler is amazingly useful.



http://www.edn.com/electronics-blogs/the-emc-blog/4402169/DIY-6-GHz-comb-generator (really works well to 3 Ghz...)

Even a cheap TTL clock oscillator driving a 1N914 or Schottkey Diode via 100 Ohm resistor radiates beautiful harmonics up to L band...

I have one of these on order, but you need to choose your initial frequency carefully...

http://www.hsmicrowave.com/CMG.html

I'd agree, buying the best commercial SA you can afford is a must.

Steve