NOISE SOURCE CIRCUITS FOR MICROWAVE

Depends. If they make a lot of noise (~30dB ENR) they are usually noise diodes. If they make only a little bit of noise, they may be temperature controlled termination resistors. If they make very little noise, they may be cooled terminations.

For quick and dirty noise sources, I've used RF transistor BE junctions in reverse breakdown. I got a spectrum that was flat out to about 1GHz. I suppose it's possible go higher, but that was good enough for me. I also have a calibrated HP346B somewhere, and a Dewar with an SMA terminator that I fill with LN2 when I use it.

Jeroen Belleman

Pseudorandom shift registers make reageant-grade flat noise; AoE has a section on that.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

Breaking down the BE junction of a 2N3904 with a series 1Meg resistor from

12Vdc used to yield something like 1 Vpp of very flat noise over the audio range. Cheap noise source for electronic games, sounds like applause in the background. However, did NOT know the technique would get something flat out to 1GHz! What transistor?

I used a 2n918 --this was over 15 years ago-- with 20kOhm to +18V and the load impedance must be low: 50 Ohms, since we're doing RF. Bare ENR is in the 30dB ballpark. For serious measurements it's good practice to sacrifice a bit of noise in exchange for good reverse matching by adding a 12 to 15dB attenuator.

The upper cutoff is set by the impedance at the emitter node, basically the BE junction capacitance in parallel with the load impedance. I never tried to find the impedance of a BE junction in reverse breakdown. I should take a look, some day.

As I learned how to make lower noise amplifiers, I 'upgraded' to quieter noise sources. These days, I measure noise by switching back and forth between a termination resistor at room temperature and another in LN2 at 77K. The bandwidth over which those are flat should extend to many GHz. (Albeit surely not as far as Boltzmann and Planck would say.)

Jeroen Belleman

00, Jeroen Belleman > wrote: > >> On 2013-09-17 11:0 3, RealInfo wrote: >>> Hi all >>> >>> How broadband noise sources are built from the circuit point of view? >>>

you can use a small "grain of wheat" incansescent light bulb.

Mark

That noise generator was circa 1974, went into coin-operated games, where you had 0-3 months to design and product 'lived' for maybe 1 year, before a new one was required.

If it's any help towards modeling the impedance of the 'reverse' biased junction; the impedance of a 'forward' biased junction is a modified Bessel Function of the first kind. I worked it out long hand circa 1968. ...took two weeks. The effort was to show that there was little improvement in performance by biasing the RF detection diode more than slightly ON. The RF Detection diode went into HP's Ku-band RF Source, the

12-18GHz RF Supply with Automatic Leveling (ALC circuit). The diode detector was part of the control loop so had to analyze it a bit. Whew! Those were the days, eh?

How well does that work?

Up to several GHz? Wow.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com

AFAIK, noise diodes are purpose-built zener diodes that are then calibrated.

I've used a regular ol' 6.8V zener as the source for a noise bridge for amateur-radio antenna tuning and receiver work. It works great, as long as it doesn't need to be well calibrated.

I've also wondered about how well comparing an LN-2 resistor with a room- temperature one would work. It sounds as if you feel it works well.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com

The op said "broadband", whatever that means.

How about zenering the b-e junction of a BFT25?

--

John Larkin         Highland Technology, Inc 

jlarkin at highlandtechnology dot com 
http://www.highlandtechnology.com 

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom laser drivers and controllers 
Photonics and fiberoptic TTL data links 
VME thermocouple, LVDT, synchro   acquisition and simulation

The title says "microwave". Whatever _that_ means, it probably means that you state it in GHz.

--
Tim Wescott 
Control system and signal processing consulting 
www.wescottdesign.com

How broadband? What frequency range? How flat a response do you need? Do you care about cost? Numbers please, not vague descriptions.

Well, you could look at what's inside commercial noise sources and see what you can steal. This article might help: or look at what others are doing for making noise: etc... Amazing what you can find with Google search.

Drivel: Back in the dark ages, I shoved a small fluorescent tube into a section of waveguide as a noise source. The tube ran on DC to avoid conducted radiation from the 117VAC power lines. I had RF noise to about 13 GHz, where the tube would no longer fit into the waveguide. However, if I were to build it today, I would use a much smaller CCFL backlighting tube from an old LCD computer monitor. The necessary shielding was too messy, so I simply embalmed everything in aluminum foil.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

So build a pseudorandom sequence generator out of Hittite flipflops.

--

John Larkin         Highland Technology, Inc 

jlarkin at highlandtechnology dot com 
http://www.highlandtechnology.com 

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom laser drivers and controllers 
Photonics and fiberoptic TTL data links 
VME thermocouple, LVDT, synchro   acquisition and simulation

Yep, that's how high-frequency noise is often generated; the negative-resistance characteristic of a lightly conducting discharge tube stuffed inside a waveguide. It's often nice to have a solid-state source, but gassy tubes win this one time!

I'll take the solid state source. The fluorescent tube RF noise source sounds easy, but has a few problems. The gas discharge tube output is affected by everything. Temperature will change the conduction threshold. Negative resistance also makes a dandy amplifier, mixer, or oscillator often simultaneously. Any signals that come down the power supply leads will appear in the output spectrum. Or, they can mix with the noise forming noisy comb lines. Anything you can do with an NE-2 neon lamp, you can do as well with a gas discharge tube. The broadband isolation amplifier was another challenge. I had to work on the RF noise generator in the dark as turning on the lights would change the threshold voltage. Enclosing everything in a waveguide worked nicely until I discovered that reflections from the right angle waveguide joint and the closed back end formed a dandy cavity oscillator.

Eventually, all the challenges were overcome and my noise source worked reliably for about 5 years, when someone decided that it needed to be "calibrated" and broke the tube. Rather than replace the $2.00 hardware store fluorescent tube, someone burned $1,000+ on an HP 346B noise source. That's when they discovered that the fluorescent tube can easily handle accidentally transmit power, while the diode noise source would blow out the noise cartridge every time the xmit stage accidentally belched RF. Oh well.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Indeed I do. For any resistor just sitting there, the noise is known from first principles. With two resistors at known temperatures, I know the ratio of noise powers. I put that through my amplifier and measure the ratio at the output with a spectrum analyzer. Solving a pair of equations then gets me the noise of the amplifier.

It's called the 'Y method'. It doesn't require accurate knowledge of the amplifier's gain, and the spectrum analyzer doesn't need to be calibrated for bandwidth or absolute accuracy, as long as its scaling factor is good. No worries about using peak, average or rms detection either. It all drops out.

To get reasonably accurate results, the amplifier's noise has to be in the same ballpark as the source's noise levels, and its gain must be comfortably larger than the spectrum analyzer's noise figure. That's usually easy.

Jeroen Belleman

How many GHz do you need ? How flat does the spectrum be ? How large amplitude do you need ?

How about a 50 ohm termination resistor followed by a few MAR/ERA style MMICs ? If the first amplifier has, say, 3 dB noise figure, then half of the noise is from the resistor and half from the first amplifier.

This circuit should be good depending on amplifier types and frequency compensation.

Since the noise from the resistor as well as the gain of the amplifier is temperature dependent, these should in a constant temperature enclosure. This enclosure should also be well protected against external RF ingress.

Jeroen, How do you deal with the cable (capacitance) that goes down into the LN2?

50 ohm coax and resistor? Do you put equal lengths of coax on each 'arm'?

Thanks, George H.

I was at Bell Labs, Holmdel, (1949) when this was "invented". I forget who there found this, but he spent a whole afternoon buying every fluorescent tube he could find. They all worked about the same when put on a slant into a waveguide.

--
Virg Wall, P.E.