Diode RF probe issues

[wizard1]

Howdy,

Being cheap and without a scope, I decided I would build a simple diode RF probe to check power output from one of my radios. I used a kit from http://www.qrpkits.com/rfprobe.html. For a 5W CW signal into a 51.5 ohm dummy load on one of my DMMs (keysight U1252b) I get 12.6V out. I expect something in the 16V range. This does seem to vary with frequency, so that number was measured around 14 MHZ. Using an old Radioshack 22-802 DMM, I see the expected ~16V out.

Today I dragged both meters and the probe into work where we have more advanced diagnostic equipment (scopes, signal generators, etc). If I look at a sine wave @ 20 kHZ, 3V peak-peak I get 0.9821V out of several meters I tried (including the two above). That seems about right, and agrees with the scopes measurements for Vac RMS.

My question is, why am I not seeing this behavior carry over to the shack and my transceiver? I do notice that the output of the probe is not a level DC voltage, it's more of a sine wave which strikes me as a bit odd (see picture with blue trace below. Blue = probe output, Yellow = signal I'm measuring).

Two photos of the probe in use and corresponding readings are below.

[wizard1]

Here is a brief followup to this. I still haven't found the issue, but I've started to suspect the Keysight U1252B has a problem with the input signal. Perhaps it is subject to RF noise? Still trying to track this down. I'll post with updates as I have them.

[ChristofferB]

I have no concrete explanation to offer, but it seems it's just a AC coupling cap, shunt diode and series resistor from the link you provided. It's not so surprising that the output of the probe is still an AC waveform! - A lot of designs has a cap to ground after the diode to smoothen out the signal a bit more, it seems. I think your keysight DMM is 'too good'.
It's a nice looking kit, but with that long unshielded spear as probe tip, I'd think it would pick up a fair bit of noise!

In lack of better test gear, I've found this HUGELY useful: https://www.ebay.com/itm/1-8000MHz-AD8318-70dB-radio-frequency-log-detector-RSSI-measuring-power-meter/142139056479?hash=item211825e15f:g:dxMAAOSwPCVX9GfW

it's a tiny little log detector that'll give a linear DC voltage per dbm - and from 1MHz-8GHz! Mind you, it'll only do up to +5dbm i think, but that could be fixed with a attenuator.
 

[Andrew]

qrpkits have another document on the probe page called "Using the RF Probe"

http://www.qrpkits.com/files/RF_Probe_How_To.pdf

it mentions the required input impedance of the dmm and how to alter the series resistance to suit. Could that be an issue?

Also, given the series 0.01uF cap, I would test the probe between 1 and 30 MHz rather than at 20kHz.

I have a similar probe which was used with an old analog meter, now where did I put it...

[CJay]

I think what you have there is a demodulator probe?

The diode and cap need to be swapped if you're going to measure power.

I'm confused, that doesn't make sense to me but QRPKits...

My dummy load with diode power measurement isn't configured that way.

Using various meters, Fluke, TTi, some other make I can't remember right now, I get repeatable, accurate results with a dummy load with built in diode detector plus 100nF cap to ground on the output.

The following page is what I based it on and the formulae there work nicely too

http://www.k4eaa.com/dummy.html


 

[wizard1]

All,

Thank you for the comments. I've got a small update. After some testing with different DMMs and a scope I've confirmed the Keysight U1252b is showing the wrong reading. An older U1252a reports the correct value. I also have additional data that at least one other U1252b reports the same wrong reading. What I'm looking for now is a way to test this without the probe so that I can have Keysight take a look. If I find a signal generator that is capable of putting out something that is similar to the probe output that should be enough right?

[bd139]

I wouldn't throw this at keysight. This is a poorly designed RF probe or mismatched meter. Two things to check:

1. Check the input impedance of the meter in the range in question. Only manual range. Some of them have 100Gohm impedance (configurable). At least my U1241C does. Needs a constant 10M impedance presented or it'll go mental due to the RF heading into it and read all sorts of weird shit. The RMS reading is a voltage divider between the 4.7M and the 10M input impedance.
2. My probe, homebrew, is the same but it has a 10n cap across the output of the 4.7M resistor and ground. This was required to get a stable reading on a couple of DMMs I used.
3. Don't expect a particularly accurate reading under a 5v p-p or so. This is a guestimate tool.
4. You have a scope - that's a far better RF probe within its bandwidth.

I built my own high Z multimeter and probe for this task. I will detail this elsewhere when I get some time.

[wizard1]

bd139 is on to something!

It turns out the input impedance in the U1252B manual is wrong! For DC voltages over 3V the input impedance is actually 10Mohm + 10Mohm in parallel (5Mohm). Doh! 

So after all that I found my issue. The capacitor you mentioned in #2 is something I haven't tried yet. Was that helpful for the Keysight you have?

On 3-4, good points. As you said the scope is a much better tool for this!

[bd139]

Re, point 2, yes that was helpful. It acts as a crude low pass filter that gets rid of a big chunk of the RF energy.

I built an RF power sensor instead in the end based on this as my probe was only useful for peaking: http://www.qsl.net/sz1a/download/build%20an%20rf%20power%20meter.pdf

This is only good for substitution based measurements but is accurate to about 1% from -72 to +13dBm (!). Ugly but it works well!



AD8307 was from random Chinese supplier.

[wizard1]

I'll have to add a cap to see if that helps any for me as well.

Wow, that looks like a neat project. Thanks for the photo as well.

Have you noticed any issues with the AD8307 from China?

[bd139]

The Chinese ones seem absolutely fine. No problems at all.

Building a better power meter at the moment. That one worked but the OCD person hiding in me wasn't happy with the chaotic design