R.F. probes

[Skywave]

There appears to be two approaches to the design of R.F. probes.

One is where there is a diode (or diodes) in the head of the probe so that the R.F. signal is first rectified, then followed by d.c. amplification; the other uses a resistive attenuator in the head, followed by broad-band R.F. amplification (to compensate for the loss in the resistive probe), then rectification to d.c., usually with consequent d.c. amplification.

I've always been a bit puzzled as to the relative merits of each method. Can someone enlighten me, please?

Al.

[Julesomega]

A diode probe has very wide bandwidth but limited sensitivity and poor linearity. An RF millivoltmeter can have high sensitivity and low noise, bit like a 'scope, really

[Argus25]

The main reason (or perhaps advantages and disadvantages) of each method relate to the dynamic range and level of the voltages that can be measured.

Very low level signals are difficult to rectify with good linearity, its not impossible, I posted the circuit of a two diode probe where one shunt diode compensates for the non linearity at low levels. Still, even with germanium or schottky signal diodes, much below 200mV linearity is an issue.

So, if you want a good results for low signals, its best to amplify the signal up with a broadband amp, then rectify/detect it for processing before the meter.

Diodes are much better with large dynamic range and higher RF voltages, for simple hand held probes, only limited by the diode's voltage ratings. Unlike the case with an amplifier where its output saturates and you can only use it over a much more limited range of amplitudes, unless of course you have a very good broad band calibrated & adjustable attentuator on its inputs to control the levels to avoid over-driving it.

So for the ideal RF probe, a very good broad band attentuator (of the quality you see in an oscilloscope) followed by a broad band amplifier (of the quality you see in a scope) then rectification and metering would be pretty hard to beat, only beaten of course by a good scope itself (like a Tek 2465B which has a 400MHz bandwidth).

One of the better methods I have seen is outlined in H&H where the input voltage of interest is converted into a current source. This drives the rectifiers, so that their non linearity vanishes. Their circuit would need a Fet added to its front end (much like in a scope's input section) and an attenuator and it would then make an excellent broad band RF probe without too much circuitry.

[Bazz4CQJ]

IIRC, the (successful) Marconi and Booton millivoltmeters both had the diodes at the front, followed by quite complex circuitry to achieve the DC amplification. Whether their designs were the outcome of profound theoretical assessments or "suck it and see" experiments, I guess we'll never know, or for that matter what their rationale was when they chose the final design?

The question as to exactly what diodes they used never gets answered!

B

[MrBungle]

I’ve taken to sticking my fingers in my ears and using an AD8307. This has a chain of amplifiers and defectors that I don’t have to care about. Put RF in, out pops very nice log response of 25mV/dB. Need more dynamic range? Switch in an amplifier. Going to blow it up? External attenuator. Did blow it up? (This did happen). Solder new IC in, no crying into an HP service manual wondering where the hell you’re going to find some new diodes.

[Skywave]

Thanks for your replies: things are clearer now.

Al.

[G0HZU_JMR]

The other thing to consider is the loading effect of the probe.


Passive Diode Probes
Generally speaking, the loading effect of a decent (passive unbiased) diode probe will vary with repect to signal level and frequency.

At levels of a few volts rms I'd expect a decent diode probe to look like a couple of MΩ (Rp) in parallel with about 1-2pF (Cp) and this would hold true up towards VHF. However, as the signal level is reduced the loading of the probe will gradually increase. Once down to levels of 10-100mV the probe might look like 100kΩ Rp in parallel with about 2pF Cp.

Once the frequency starts to climb into VHF and through to UHF the Rp figure will drop rapidly due the internal resistance of the diode(s) in the detector. So the numbers above will get much worse for both high and low drive levels. At 100MHz with a low drive level of <100mV, the Rp figure could drop to below 10kΩ. A lot depends on the diode type and the design. It's possible to get Rp and Cp figures much worse than the above if the wrong diode types are used.


Active Probes
There are lots of different types and the (loading) performance will depend on the design. A lot of the magazine based active probes I've seen on here use a JFET buffer followed by a diode detector. A typical example is the Haigh probe as in the image below.

In the low gain mode the JFET is configured as a source follower and at frequencies above a few MHz the probe will generate negative resistance at the input because of the capacitance of the diode detector at the source. This may never cause any problems in typical use, but it could cause instability or act as a regenerative q multiplier when probing certain circuits.

In the high gain mode the JFET is configured as a common source amplifier with a fair bit of voltage gain. Because of the drain-gate capacitance of the JFET, the loading of this circuit will be problematic even at a few MHz. The Rp could be as low as 3kΩ by 10.7MHz and the input capacitance will be high as well because of Miller Effect. It could easily have a Cp of 20pF for example. So this probe would be a poor choice for probing high Z tuned circuits unless it was used for LW or MW circuits only.

To improve things a resistive attenuator/divider has already been suggested by Argus25 but the other way to get around this is to use a capacitive divider ahead of a FET buffer. This should give high Rp and low Cp. Then amplify the signal before detecting it.

[Argus25]

I have seen a lot of transistorized/Fet scopes trashed for parts. The quickest way to build a good RF probe would be to commandeer the input attenuator and at least part of the vertical amplifier, including its input Fet and add the rectifier & meter circuit to that. You would then have a proper calibrated attenuator and you could use x10 probes and you would also know the bandwidth and loading effects. A vertical plug in style scope amplifier would also be good if you set it up with the extra parts, housing, power supply meter etc.

[G0HZU_JMR]

I think a lot depends on what the probe will be used for. If it is going to be used for demodulation then I guess a probe with a diode detector (somewhere in the circuit) will be a popular choice.

If basic level detection is all that is needed (rather than demodulation) then I would have thought that a capacitive divider followed by a JFET buffer followed by Mr Bungle's AD8307 log amp would work very well. However, the detector output will be approx. 25mV/dB so no good for demodulation.

In order to kill any negative resistance at the JFET gate the source would need to be terminated in a low resistance. This would sacrifice some sensitivity but the AD8307 has lots of sensitivity. With a 10:1 or a 20:1 capacitive divider ahead of the JFET the Rp of this system would be huge even up into the VHF region.

My old -3dB BW Q meter jig uses a capacitive divider followed by a JFET followed by the 50R input of my VNA. The input Rp at the capacitive divider is so high I can't measure it even up at VHF. It's probably three orders of magnitude higher than anything I can easily measure here. I think it uses a 25:1 capacitive divider and I did some tweaks to the JFET buffer to get the Rp at the JFET gate really high (but still avoiding negative resistance). I can't remember which JFET I used but it wasn't anything special.

My Marconi 2388 1GHz RF probe uses a capacitive divider at the input (followed by a FET buffer) and I think that the rotary attenuator at the front is just a variable capacitor in series. So the attenuator works by varying the division ratio at the front end.

[Argus25]

The problem is though, for a home project, where there might be limited test gear (hence the need for an RF probe in the first place) building a good wide band attenuator is the tricky task, the rest of it is relatively easy. Because it requires a combination of both electronic and mechanical design to get it right. And if the RF probe is to work well over a wide range of input voltages from tens of milli-volts to 10 volts or more, the attenuator is required. That is why it would be easier to scavenge one (which already has the multiple design parameters solved) from a defunct scope, than try to do that part of the design from scratch, especially if the probe needs to work up to 100MHz or more.

[Bazz4CQJ]

Originally Posted by Argus25

I have seen a lot of transistorized/Fet scopes trashed for parts. The quickest way to build a good RF probe would be to commandeer the input attenuator and at least part of the vertical amplifier, including its input Fet and add the rectifier & meter circuit to that.

Could you suggests some models that would fit the bill? Taking a look on ebay, scopes with say a 100MHz bandwith are still asking significant prices for "spares or repair" items.

B

[Argus25]

You have to wait for broken ones. There are a few Hitachi's on ebay USA, I'm not sure about your UK ebay. A 40 MHz one would be a good start and mostly you get two attenuators 5mV (straight through) to 5V, and they are all set up for a x10 probe.

https://www.ebay.com/itm/Hitachi-V-4...kAAOSwRW9cmTZh

[G0HZU_JMR]

Originally Posted by Argus25

The problem is though, for a home project, where there might be limited test gear (hence the need for an RF probe in the first place) building a good wide band attenuator is the tricky task, the rest of it is relatively easy. Because it requires a combination of both electronic and mechanical design to get it right. And if the RF probe is to work well over a wide range of input voltages from tens of milli-volts to 10 volts or more, the attenuator is required. That is why it would be easier to scavenge one (which already has the multiple design parameters solved) from a defunct scope, than try to do that part of the design from scratch, especially if the probe needs to work up to 100MHz or more.

I think that this has various issues:

High Cost
Difficulty in obtaining a suitable donor
Complexity and time to build
Really (really) poor performance in terms of circuit loading up at VHF (with or without the x10 scope probe)

Poor (unpredictable?) RF response with a x10 scope probe up through VHF

I'm tempted to ask how you would adjust the compensation on the x10 scope probe (without a scope) but I really don't think it matters because the other limitations are much more of a problem.

especially if the probe needs to work up to 100MHz or more.

The circuit loading of a scope input or a x10 scope probe is pretty grim once you get up to 100-150MHz. By 150MHz, a x10 scope probe could be the equivalent of attaching a 10pF capacitor in parallel with a 300Ω resistor. Contrast this to a decent passive diode probe that might manage 20kΩ in parallel with 2pF at 150MHz. An active probe could be even better than this, especially if it has a capacitive divider at the tip.

A home made passive diode probe with something like a 1N5711 diode might manage 20kΩ in parallel with 2pF at 150MHz but it will depend on the signal level driving it. Compare this to 300Ω in parallel with 10pF for the x10 scope probe. What's the point in trying for accurate attenuators if the probe itself is going to corrupt the measurement though excessive loading?

[Argus25]

Originally Posted by G0HZU_JMR

The circuit loading of a scope input or a x10 scope probe is pretty grim once you get up to 100-150MHz. By 150MHz, a x10 scope probe could be the equivalent of attaching a 10pF capacitor in parallel with a 300Ω resistor.

But, you don't have to use a x 10 probe. You can terminate the input at the attenuator with 50R, go to a 50R system at those frequencies, as you often do with a scope anyway for VHF RF work, then if you want a high Z input, do the usual with a pre-made Fet input probe or simply make a buffer with a fet and some transistors as an active probe head with your 2pF input. The attentuator allows a much greater range of signal amplitudes to be measured, than having an RF probe with no attenuator at all and you cannot condemn the utility of the attenuator because of the high frequency difficulties of the x10 scope probe .

[Argus25]

Also, Silicon Chip designed a really inexpensive to make high input Z differential probe head, made to drive what is a standard scope input attenuator. It is good to over 80MHz and has the advantage of the differential input with no earth lead inductance effects that you get with a single ended probe and if combined with a scope attenuator and broadband amp/rectifier/metering would have the makings of a good RF probe system for that frequency range. Plenty good enough for AM & FM radio work.

http://www.siliconchip.com.au/Issue/...e?res=nonflash

Handy too that its powered by a USB port.

[G0HZU_JMR]

OK but the thread is about attenuated/buffered or passive diode detector based probes. I'm not sure how your setup fits into this concept. Are you suggesting that for VHF one should use an external active probe followed by a switchable scope attenuator (and its FET buffer) followed by more gain ahead of the diode detector to get it into the linear region?

It seems to be getting more and more complicated and I think most people would be better off just repairing the donor scope.

[Argus25]

Originally Posted by G0HZU_JMR

It seems to be getting more and more complicated and I think most people would be better off just repairing the donor scope.

I couldn't agree more !

I think in reality, for small RF signal levels, that require amplification before any reasonable level of signal can be obtained for any linear rectification, it is better just to go straight for at least a 400MHz bandwidth scope as it contains all the required equipment/sub-circuits to measure the signal and you can also use a high Z probe head on it if you want.

Any peak RF voltage over a few hundred millivolts or more for a cheaper solution, just go for a diode RF probe, like the HP ones with two diodes where one diode compensates the non linearity of the other for the lower level signals. Since the probe has a DC output you can use any old scope or DVM with it.

[G0HZU_JMR]

Yes, I'd agree with that

My trusty old Racal 9300 true rms voltmeter is a bit like the setup you were describing as it has a BNC input with a 'scope like' input impedance and it has manually switched attenuators in the front end. But it uses some clever and complicated circuitry to create the true rms detection and it can work down to about 30uV and I think this is what makes this meter so special. It manages a 20MHz BW and I think you can get these for less than £40 these days.

So that is another option I guess. What is certain is that there is no 'one size fits all' solution for an RF probe. That's why I have so many of them here...

[Station X]

Thread reopened.

[Bazz4CQJ]

The "Golden Fleece" of a high-impedance RF probe has been coming up on the forum at regular intervals for some years. Numerous published designs claim wonderful solutions using a single J-FET, but a number of us have attempted to build them but without success, and when those with much more expertise than myself model the behaviour of such components, they tell us that "negative resistance" is going on and that won't do.

I just noticed that there's a thread on high-impedance RF probes going on over on the EEV forum https://www.eevblog.com/forum/testge...zer/?topicseen.

I don't feel competent to assess whether there are any ideas worth looking at there, but if anyone else wants to take a look, please report back and let us know whether a "Eureka" moment could be upon us .

B