I have a Boonton 91H and a 91-12F probe to go with it. Through my own stupidity, I damaged the probe diodes. Much has been written about these probes but I could not find anything that told me what diode to use to replace the proprietary Boonton device. In the process of fixing the probe I have collected a lot of stuff on diodes and the Boonton probes. After reading a lot about RF probes and Boonton probes, I ordered a number of diodes to try and today I got one that seems to do the trick, at least out to 500Mhz where my generator stops. The diode that works in my probe is a 1SS99. I was worried about this diode because it only has a 5 volt breakdown voltage. In the Boonton probe, there are two diodes connected in a full wave arrangement so the diodes each see the peak-to peak signal voltage. The meter top range is 3 Volts (RMS) which translates to about 8.5 volts peak to peak. I ordered 25 diodes so I went ahead and did the calibration at the full 3V and the diodes are fine. I guess time will tell how robust they really are.

I tried Schottky diodes BAT15-04W, Avago HSMS-2822. The Bat15 is a 4 volt diode and the 2822 is 15 volts. These are both usable but they don't calibrate as well and they both behave somewhat poorly on the 1mv scale, particularly below 500uv. The Bat15 was better but not by much. I bought an HSMS2812 but did not try it because it has lower output that the 2822.

I also tried 1N270. This calibrated OK at low frequency but rolled off pretty bad starting at about 20Mhz.

The 1SS99 I stumbled across on the auction site. The guy is advertising them as ISS99.

The last thing I will add is that the HSMS2822 had the flattest frequency response. It was in a SOT23 package and I bought some small PC boards for SOT23 breakout. I mounted a 1000pf 500V MLCC cap on there too to replace the cap that was in the probe. Don't ask what happened to that. Also, I suggest using film type resistors for the 150 ohms in series with the diodes (or SMT resistors ). The 1SS99 is a glass diode about the size of a 1n4148, smaller than the Boonton diodes. I see some swing in the frequency response but it stays within the spec for the meter.

Last thing, the Boonton diodes are rated for 10VAC. I doubt that I can apply that safely with the 1SS99. Also, the fact that it will handle 400VDC doesn't mean you can connect the probe to a hot 400V circuit. The 400V step will pass right through the coupling cap. I might do some experiments since I have 23 diodes left, to see how much DC it will handle hot. If you connect it to the circuit with power off and then bring it up "slowly", you can get away with big DC probing.

Hope that helps somebody.

Tony

Always interesting to see something brought back to life through cleverness. Nice job

I probably should have waited to post that. There are still some imperfections with the 1SS99.

1. The 1mv scale is not right when calibrated at 900uv per the manual. Calibrating at 800uv instead makes most of the scale good (200uv to 800uv is good) but it starts reading low above 800uv. 1mv will read out as 910uv.

2. The low end of the 3mv scale is off. 1mv reads as 1.2mv but 2mv and 3mv are good.

3. The 3v scale calibrates at 2.5v per the manual and is pretty good below 2.5v but reads low above 2.5v with 3v reading about 2.8v. This one doesn't surprise me because we are exceeding the reverse voltage spec on the diode.

I find the behavior between 800uv and 1.5mv on the 1mv and 3mv scales to be a little surprising. That doesn't seem to be a logical place to have problems.

Still, this diode is the best one so far.

Tony

Here's a PDF with some great info on Boonton probes -
http://ve2azx.net/technical/RF-Probes.pdf

I have seen that page but thanks for reminding me of it. This page says that a later boonton probe used a BAT62 diode but then it says BAT32. But BAT32 is not a SOT-23 while the picture in the article shows SOT-23. I forget why I decided not to try the BAT62. I'll go back and revisit that one.

From the same site:

http://ve2azx.net/technical/BoontonRFprobes.pdf. This is where I first saw the 1SS99 mentioned for the probe. It includes an update in January 2021. There is no mention of the BAT62 here. I'm glad I looked at this again though (because you mentioned the site) because it gives me yet another diode to try (1SS351).

I have a 1N830 on order. These are microwave detectors and they don't offer normal diode specs (like VF and VBR). I was curious about them and happened to stumble on a site that had 2 of them for sale.

Here is a picture of the calibration results for the 1SS99. There is room for improvement but it is not bad, for the most part.

I have done what I can so here is the final update on this from me. I ordered 1ss351 and BAT62 diodes from Mouser. I had high hopes for the BAT62 since it seems to be used in the later 952001B probe. With my 91H meter, these two diodes fail to calibrate properly on the 1mv scale of the meter. At least the 1SS99 works fairly well on that scale. The other diodes might be fine for a model 92 meter. The 91H was never actually sold with the 91-12F probe as near as I can tell from the old Boonton catalogs. The last probe offered with the 91H was the 91-12E. Maybe I will run across one of those some day. In the mean time, the 1SS99 will have to do for me.

I did fiddle around some with the resistor values in the probe when trying the BAT62 and it did not seem to help much.

Tony

It is interesting to read the section in:

http://ve2azx.net/technical/RF-Probes.pdf

...on square law detector diodes. However, they did not mention the "Back Diode". This is actually a Tunnel diode used in reverse and these make amazing square law detectors down to extremely low signal levels. This is explained with a graph from a Radio Astronomy lab that was published in Horowitz & Hill's textbook, The Art of Electronics.

Also, from that book there is a method that trumps all other methods for making an RF probe which can give accurate linear readings over a wide dynamic range and down to very low RF signal levels, by completely eliminating the non-linearity and square law properties of the diodes, no matter what it is.

The concept is simple, instead of driving the diode with a voltage source, it is driven by a current source.

When most see this circuit though, they are confused by it because it just looks like a peak to peak rectifier, driven by an RF amplifier.

But close inspection shows an enormous impedance mismatch with the high Z output of the current source ultimately being loaded by 100R via the diodes. It is this extreme mismatch that linearizes it. The input voltage to the diodes simply rises high enough to generate a current that is directly proportional to the input voltage, that current develops a voltage across the 100R load. It is very very clever.

Another remark I could add is, that over many years, it was customary in AM radio detectors, especially transistorized ones, to attempt to have a reasonable impedance match of the final IF transformer and the load on the AM detector. It turns out that this is exactly the wrong thing to do, if you wan't the detector to perform well with low signal levels, because the square law behavior of the diode steps in under a couple of hundred milli-volts into the detector.

This is one interesting reason why tube AM radio detectors often work better on weak stations, generally there is a greater mismatch of a relatively higher voltage source (out of the final IF which can sometimes be 10's of volts off load)emanating from a higher source resistance, more resembling a current source, with a relative higher detector load. So detector linearity is better maintained for lower level signals in a tube radio vs a transistor one. Oddly it is one of those scenarios where the "impedance match dream" is a bad idea.

I was able to significantly improve the AM detectors in my Eddystone EC-10 radios, for low level signals, by changing to a current source drive for the detector diode.

I got my hands on a Boonton 91-12E probe with the original Boonton diodes intact. It behaves the same as the BAT62 diodes did and will not calibrate on the 1Mv scale. Now I suspect that my 91H is sick.

The symptom is correct calibration at 900uv per the manual but applying smaller signals results in increasing too-low readings until the meter actually starts reading below zero when applying 300uv. It returns to zero when zero volts is applied. It acts like there is a small constant offset. My guess is that the chopper is the problem.

Tony

AHA! I just had an AHA! Moment when the darkness was lifted. I was about to toss my 91H in the corner.

The basic issue is that the 91H meter zeroing technique is sensitive to the polarity of the detected RF (a DC signal) from the probe. The chopper runs on 60hz AC and the injected zeroing signal is derived from 60Hz AC so the chopped DC signal from the probe will either be in-phase or anti-phase relative to the zeroing signal. One way works, the other doesn't. I changed the connector on my meter at one point and was not careful about how I connected the input leads to the connector. When I am inserting diodes into a probe, I don't pay much attention to which direction they are connected (in proper series yes, but which end to which probe wire, no).

Note that the 91CA and 91DA meters do not have this issue because they simply inject DC directly at the probe inputs so the zeroing offset and the probe signal are chopped together and cannot have the wrong phase.

The 91H zeroing circuit is too clever I think. First, they inject a positive DC into the meter circuit when the switch is in the 1mv position. Then they inject an AC zeroing signal into the middle of the amplifier chain controlled by the zero knob, and this gets rectified to a negative dc at the meter that cancels the DC offset they injected.
It seems that this AC zeroing signal is supposed to interact with the chopped signal from the probe as part of the "linearizing" process on the 1mv range. By hooking the probe leads backwards, I de-linearized it.

Interesting stuff but what a headache.

Tony

Just tried a BAT62 and now it works fine. I think this is the correct diode to use to repair old Boonton probes. I did a curve trace on all of the diodes I have, including the original Boonton diode and the BAT62 is by far the closest match in the square law region. I bought some from Mouser (BAT62E6327HTSA1). I also got some SOT23 adapter boards on amazon that are supposed to adapt sot23 to DIP. This diode is in a SOT143-4 package with two independent diodes (it is a 4 pin device) in opposite directions so it is easy to jumper two pins and get a series pair for the probe. It looks something like this when mounted (this one is actually the 1ss351 in a SOT23):

Here is the result of calibration with a BAT62 diode. I kept the vertical scale the same as the chart for the 1SS99.

Thanks for the detailed report. This will be a big help when I tackle my Boonton 91.

mark