square law RF detector - choice of diodes in commercial design

Hello Stepan,

You could carefully measure one but if you find out it's indeed GE then you may not be able to source any. In the old days they did use GE point contact diodes here but these were fickle (easily damaged on impact, like a probe falling down) and they probably had to select the heck out of them before binning them for production. Nowadays Schottky and GaAs diodes are used. More reliable.

Here is an interesting paper about this stuff:

Regards, Joerg

Hi Joerg, From the link you gave:

------------------------------- Today, zero bias Schottky diodes produced on a silicon substrate or GaAs diodes are being mostly used. Their electrical characteristics are similar to those of the germanium point-contact diodes, but their long-term stability is as high as that of thermocouples.

------------------------------

Hmm, I wonder where I can find datasheets for the above mentioned very special diodes. Clearly no *ordinary* schottky or GaAs diode has a usefull a square law curve at microvolt levels, like a germanium diode does. Such data sheets elude me on google.

The boonton probe I referenced is '80s vintage and the direct descendant of that probe being sold today by Boonton/Ballantine, has the same specs and is compatible with the same meters, so it must not be a germanium diode, even though my measurements at 10uA and 100uA and 1mA at 20 degrees C do indicate germanium. I could measure the reverse breakdown voltage too but I'm afraid to do that without a datasheet. I have not measured thermal response. The probe runs without dc bias to the diodes.

...Stepan

I should clarify that it is the special GaAs diodes that I'm having trouble finding the datasheets for.

...Stepan

The ubiquitous 1N2x diodes were specifically designed to be square law detectors:

At least one of the Radiation Lab books discusses them.

I read in sci.electronics.design that snipped-for-privacy@hotmail.com wrote (in ) about 'square law RF detector - choice of diodes in commercial design', on Sun, 6 Nov 2005:

Are you sure about that? I think the square law relates to the action of the diode 'peak' detector *circuit* on an r.f. input, not to the d.c. characteristics of the diode, which are exponential.

I've measured '+2 dB out for +1 dB in' on detectors using 1N4148s.

In that probe circuit, you do need 'matched' diodes. How to 'match' them, though?

Look at the graph comparing the germanium diode to the silicon diode:

The slope of the V-I line is zero for silicon around the microvolt level. I believe that means the signal is overcome by noise and thermal drift so that you can't get a reliable reading.

...Stepan

I read in sci.electronics.design that snipped-for-privacy@hotmail.com wrote (in ) about 'square law RF detector - choice of diodes in commercial design', on Sun, 6 Nov 2005:

It only looks like zero because of the particular graph scales the author chose.

You do need careful design, but it isn't impossible. Remember that on the d.c. side you only need enough bandwidth to follow any changes you are interested in, so it can be quite low, thus minimizing noise. Why not take up the author's suggestion of a biased Schottky diode?

LOL the 1N21 is older than I am! But I think it's interesting and usefull. I wonder if 1N21 derivative is what's in here:

And below is partial datasheet for the 1N21:

Maximum Ratings: Power Dissipation = 100mW Derating above 25 grdC. : 8 mW/grdC. Top = Tstg = -55 to 150 grdC.

Characteristics: Frequency Range: 2 GHz to 4 GHz, S- Band Test Frequency : 3.1 GHz L.O.Power : 0.5 mW Z if : 1N21C : min 300 Ohm, max 500 Ohm 1N21D : min 325 Ohm, max 425 Ohm 1N21E : min 350 Ohm, max 450 Ohm 1N21F : min 350 Ohm, max 450 Ohm 1N21G : min 350 Ohm, max 450 Ohm VSWR : 1N21D : max 1.5 1N21E,F,G : max 1.3 Proof Burnout : 5 Ergs Noise Figure at N?= 1.5 dB: 1N21C : max 8.3 dB 1N21D : max 7.3 dB 1N21E : max 7 dB 1N21F : max 6 dB 1N21G : max 5.5 dB

Suffix: R = Reverse, M = Matched Pair, W = Reversible ______________________________________ Without guaranty! Mistakes are possible. Please ask via email if you find a strange value!

______________________________________

Maximum Ratings: Power Dissipation = 100mW Derating above 25 grdC. : 8 mW/grdC. Top = Tstg = -55 to 150 grdC.

Characteristics: Frequency Range: 8 GHz to 12 GHz, X- Band Test Frequency : 9.4 GHz L.O.Power : 1 mW Z if : 1N23D : min 350 Ohm, max 450 Ohm 1N23E : min 335 Ohm, max 475 Ohm 1N23F : min 335 Ohm, max 465 Ohm 1N23G : min 335 Ohm, max 465 Ohm 1N23H : min 335 Ohm, max 465 Ohm VSWR : max 1.3 Proof Burnout : 2 Ergs Noise Figure at N?= 1.5 dB (Index leider nicht lesbar): 1N23D : max 7.8 dB 1N23E : max 7.5 dB 1N23F : max 7 dB 1N23G : max 6.5 dB 1N23H : max 6 dB

Suffix: R = Reverse, M = Matched Pair, W = Reversible

______________________________________ Without guaranty! Mistakes are possible. Please ask via email if you find a strange value!

______________________________________

That's the million dollar question. Especially since the folks at Boonton and at Ballantine Labs chose not to use external bias in their industry-standard meters. I own the Boonton 9200A and it gives readings down to 50uV without using any kind of bias.

I wonder if it's to keep the probe input capacitance or impedance suffering. ???

...Stepan

Look on the Agilent website for the HSMS-2852 or similar. I'd post a link, but they made the URL about 200 characters long, and it probably contains all the information they've gathered about me for the past 5 years...

There are lots of application notes on there too.

Chris

Hello Stepan,

Just wondering: If this one doesn't fit could the OA91 be similar? I had used it a lot for detectors when I was young. It was in the skinny glass housing, not the big one like many other OA diodes.

As to where to get GaAs diodes that may be tough in small quantities. Ixys makes only the big ones IIRC and those will likely have too much capacitance for you and be too large.

Regards, Joerg

Hello Chris,

Are they engaging in cookie-mania? This diode looks like a good deal at

Regards, Joerg

I read in sci.electronics.design that Chris Jones wrote (in ) about 'square law RF detector - choice of diodes in commercial design', on Sun, 6 Nov 2005:

This URL is manageable:

Since this thread will live forever on Google, here is a summary of detector diode possibilities for use in RF millivoltmeter microwattmeter probes:

We note that these are used in RFID tags and that we google the like shown below: zero-bias schottky detector diode

We also note that there are many application notes available and that we google for: square-law detector schottky diode

Zero Bias Silicon Schottky Detectors:

Zero Bias GaAs Schottky Detector:

Low Barrier Schottky Detector:

Old Germanium Point Contact 1N23 (fragile and large) 1N21 (fragile and large) 1N34A (high capacitive loading) OA91 (high capacitive loading)

You should look at HP AN 986 "Square Law and Linear Detection" and HP AN 987 "Is Bias Current Necessary?"

But another thought would be: Why not use comtemporary diodes and maybe improve the performance of the probe?

Good luck, Dave

I searched for [AN 987 "Is Bias Current Necessary?"] and the first hit was:

Try searching for [986 "Square Law and Linear Detection"].

I read in sci.electronics.design that snipped-for-privacy@juno.com wrote (in ) about 'square law RF detector - choice of diodes in commercial design', on Mon, 7 Nov

Are they on the web? The numbers are now used for some chipsets, so Google gives 3 million wrong hits.

I read in sci.electronics.design that John Popelish wrote (in ) about 'square law RF detector - choice of diodes in commercial design', on Mon, 7 Nov

Thank you. The AN 986 is there as well, and can be downloaded, but the DiodeData directory is not accessible. Not very sensible.