Ever built a working RF noise source?

[T3sl4co1l]

Incandescent lamp + suitable amplifier.

Stupid idea.  But if it works, it ain't stupid.

Seriously, though: a high temperature resistor is a good source of noise, and the only thing hotter than white-hot tungsten is UV-hot plasma.  And, plasma being an electromagnetically active material, it can exhibit complex excitations and resonances with geometry of the apparatus (accidental cavities, stubs and whatnot).

The only difficulty might be finding a bulb that matches well with its surroundings...

We're looking at:
https://www.google.com/search?q=sqrt%284+*+boltzmann+constant+*+2400+kelvin+*+50+ohms+*+1Hz%29
2.57 nV/rtHz is pretty low noise (most op-amps do considerably better..), but since we're contending with 0.9 nV/rtHz (50 ohms at 293K), it's not horrible.  I mean, 9 dB must be good for something, right?

A 50 ohm bulb might be 6.3V at 125mA, or 12.6V at 250mA.  Not so hard to find, I think.  Use a bias tee (L from +V, coupling C to load) to power it.  (Note that the noise power is the same regardless of size, as long as the temperature is there -- so it's to your advantage to use as small a lamp as possible!)

This looks reasonable:
http://www.mouser.com/ProductDetail/JKL-Components/2203/?qs=sGAEpiMZZMtj1PWpWL2%252bvDwicDUvKr1eDUSCVRwqUUA%3d

A bulb this small should be good into the GHz.  The glass will give a somewhat lower impedance around the source, but that's probably an advantage given the wire leads, and wide spacing (twin-lead transmission line always has a highish impedance).  Soldering it into a PCB, or poking it into PCB-mount sockets, should give reasonable SWR.  (And you can even measure the SWR by transmitting into it!  Just remember to do it while it's hot -- when unbiased and cold, it'll be a not-terrible short circuit instead!)

If you want the second hottest noise source on the EEVBlog Forum, I suggest doing it.

Tim

[bson]

I am curious about this B-C junction method. How exactly do you build that? Do you leave the emitter floating? What voltage do you apply? It it in avalanche mode or something like that?

Ask and ye shall receive.  This pops up every so often.



I put it on a small board with a THS3202DGN based wideband amplifier inside an EMI shield.  Works great; the band is flat from DC out to about 500MHz, then the THS3202 ripples and falls off.  (Like many CFB amplifiers it produces a single hump before hitting the wall.  Mine is about +2dB, so no big deal and for frequency domain measurements it's easily normalized anyway.)

What I find interesting is that some circuits that are very clean when swept show oddball wavy frequency response when hit with white noise.  IMD is my guess.

[LaserSteve]

Festoon lamps have been used as loads, power meters,  and noise sources.  Because they are cylindrical, its fairly easy to slip copper or brass  tubing around them and make a coaxial device.  The old RSGB UHF/VHF handbook has a few examples.  I'm just not aware of it being carried out to 1 Ghz, although it is documented to high VHF.  In the olden days, they were used as fairly accurate power meters. One lamp on DC, one lamp on RF, and mounted on either side of a oiled piece of paper.  When the spot illumination was nearly equal, the RF power was close to its DC equivalent. I'm told this can hit accuracies of 2.5% or better with the human eye...



Steve

[jh15]

I did the oiled paper photometry in high school physics in the '60s was precise. Then Polaroid, with the Swinger camera reminded me of it, comparing a grain of wheat bulb with the rangefinder view.

Miss Dr. Land, he didn't seem evil like present companies. Always wanted to work for them, knew some, they loved it.

He did hire the blind. If for profit or kindness, it was unique. Signted people would go nuts loading film in total darkness all day.

[Timeok]

I made a series of noise sources using a BFR90A. The basic features are:
Power supply: 28V
Freq. 1MHz - 2GHz
ENR about 15dB
Flatness +/- 0.6dB
And most importantly, they are able to calibrate them individually.
I have several prototypes and a final version.

[ionospheric]

@Timeok Can you please share the schematic?
Many thanks

[jonpaul]

see Microwave Noise lamp:

An ionized plasma emits a wideband noise sectra.

For audio/LF we used GenRad noise gens with a Thyratron or noise diode (special Zener)

The resistor thermal noise is calculated by the usual formula and depends on the Tdeg k and resistance only, very low in uV range ususally.

Finally in 1970s Nat Semi made a rerpeating sequence (shift reg) noise source IC for hifi acoustic eq.

Enjoy,

j

[iw5bsf]

hello i'm new in the forum i saw a great work i also have a noisecom diode anti-noise head to match the hp8970a but i can't set the parameters for the hp8970a could you give me some pointers thanks
Robert

[TimFox]

Has anyone tried using a 5722 vacuum diode?
It is designed for use in "saturated" emission (not space-charge limited) so that the shot noise output can be computed directly from the DC plate current.
I'm not sure how to optimize the connections to the filament/cathode; the plate goes to a 50 ohm load.
The circuit in the data sheet (below) shows RFCs in the filament leads, but the three filament connections are bypassed to ground with 500 pF.
If the cathode is supposed to be at AC ground, why are the RFCs needed?
I thought it would be straightforward to regulate the emission by servoing the filament current, but I was hoping to use it a lower frequencies than the usual RF circuit.
https://frank.pocnet.net/sheets/201/5/5722.pdf