R.F. voltmeter: your thoughts, please

[Al (astral highway)]

Al, please remind me/ us -- what do you have in mind for this probe in particular? There seem to be so many excellent designs suited for dizzyingly above as well below 100Mhz!

[G0HZU_JMR]

Can anyone suggest what the downsides to that approach are?

I guess that the most immediate downside would be that the meter would become prone to damage through operator misuse. The description below is for the N channel JFET version as it's easier to interpret.

If you fed in 10V rms to the meter then the detector output would be about -28V DC. If the operator then set the range switch to the mV range by mistake then the JFET gate would presumably be fed with -28V DC directly rather than via the resistive divider. There will be typically about +2V on the JFET source. This is a bit marginal in terms of Vgs breakdown voltage but some devices can withstand this. But a typical opamp (or an instrumentation amplifier?) running a single rail would get zapped by this huge negative voltage at one of its inputs.
I think you may have to use a split supply opamp anyway to get the offset trimming to work well but you would have to provide some form of limiting to stop the opamp from being fed (detector output) voltages below the negative supply. Some opamps can't tolerate more than about a volt below.

[Bazz4CQJ]

Thanks for the comments on the issues with the op-amp.

In post #15, Al gave a link to the original article in '73 magazine' inspiring this thread, and I'm just in the process (53MB) of downloading that. I encourage others to do likewise, if only to see the front cover (December 1973), which is wonderfully incorrect and well worth $1 .

The circuit shown in the OP is interesting in that it squeezes in no less that 4 detector diodes in to the probe in what still looks to be a doubler, whereas the Marconi circuits, which drew attention in previous threads, were voltage doubler circuits using two diodes. I don't know whether Al was intending to put one of the Marconi probe designs in front of the amp circuit shown, or whether he was interested in keeping the probe and the amp just as shown.

I'm sure I had picked up the suggestion somewhere, that while using more complex rectification circuits in RF probes does increase the detected voltage, there can be downsides to these circuits in the form of "anomalous" responses - resonance effects?

B

[G0HZU_JMR]

There are 4 diodes in the detector head but only two of the diodes are used as the doubler/detector. The bottom two diodes are just there as (passive) thermal tracking diodes to help with the temperature compensation. I think both sets of diodes will be biased at a couple of uA and the idea is that any bias change in the 'real' detector pair due to temperature changes will be cancelled by the passive pair of diodes in the head.

I'm sure I had picked up the suggestion somewhere, that while using more complex rectification circuits in RF probes does increase the detected voltage, there can be downsides to these circuits in the form of "anomalous" responses - resonance effects?

For one thing you get two diodes effectively in parallel with the doubler circuit. If the diodes are biased then the small signal or 'dynamic' resistance of the two diodes will be in parallel so you will see heavier circuit loading. There's also the effect of the (series) spreading resistance in each diode so the effective Rp can get quite low up at VHF for very small signals. Maybe a couple of k ohm for very small signals in the square law region for a doubler/detector like this.

[Bazz4CQJ]

Mmm..I'm going to have to plead guilty to having been slow to look at the 73 article, as having done that now, I appreciate Al's interest in the design.

All we need now is some suitably expert individual to come up with the n-channel design; perhaps he's doing that right now .

Wasn't 73 good in the day?

B

[Radio Wrangler]

There was also the American 'Ham Radio' magazine which was excellent and technically oriented for people who built their gear. This eventually got taken over by 'CQ' mgazine and turned into a sister publication carrying the more technical stuff under the name 'Communication Quarterly' And then that got taken over by the ARRL and rolled into their own 'QEX' technical magazine, the sister to QST.

The ARRL do a CDROM of past QEX which includes all of Communications Quarterly and a lot of Ham Radio.

Also there's that American magazine archive site which I expect to have the whole lot.

There is an awful lot of good amateur radio technical articles once you know where to look. These are not the usual 'here are the photos from our DXpedition' articles

David

[Bazz4CQJ]

In the article in 73, the author discusses his selection of Si over Ge diodes for the probe. He doesn't say whether the Ge ones were junction or point contact, but haven't the latter often been favoured for probes because of their low capacitance? His measured input cap was 4pF. Perhaps his requirement for an input range up to 10V excludes point contact? I wondered whether the author might still be around, but his callsign expired in 2004 .
B

[Skywave]

Originally Posted by Bazz4CQJ

I wondered whether the author might still be around, but his call sign expired in 2004

The '73 mag'. article gives his call sign and address. Just out of curiosity, I used Google Maps to locate that address and other Internet resources to discover that his wife (presumably ) sold that property in 2009 and to whom it was sold. Draw your own conclusions.
It's really quite amazing just how little privacy one has these days, even world-wide, when wandering around the 'Net.

Al.

[Skywave]

Originally Posted by astral highway

Al, please remind me / us -- what do you have in mind for this probe in particular?

I don't have a specific application in mind at present: broadly speaking, it has the potential to be just another piece of useful test gear. The overall objective is to investigate the properties of various different approaches to the measurement of R.F. voltages and to see where my conclusions lead me - hopefully to the construction of something useful.

This isn't the only design that is under consideration: there is also the design by G0WCA - which is under investigation. (But that's a topic for a different thread).

Aside.
As various members have pointed out, with any attempts to measure R.F. voltages - especially at high freqs. and from moderately high-Z sources - things can get a bit difficult: cct. loading, etc. But that doesn't mean that such a measuring device is not worth having: simply an awareness of its limitations when it is in use. An obvious example of that is the humble X10 passive 'scope probe with a typical 'tip' capacitance of something like 15 pF.

Al.

[Skywave]

Originally Posted by Bazz4CQJ

. . . if only to see the front cover ('73 mag.' December 1973), which is wonderfully incorrect . . .

Quite: I couldn't have phrased that better myself!

I don't know whether Al was intending to put one of the Marconi probe designs in front of the amp circuit shown, or whether he was interested in keeping the probe and the amp just as shown.

Hopefully, I've just answered that Q. as per above (post #49): many options are 'still on the table'.

Al.

[Bazz4CQJ]

Originally Posted by Skywave

This isn't the only design that is under consideration: there is also the design by G0WCA - which is under investigation. (But that's a topic for a different thread).

Al, you may not re-call, but I have built the G0WCA design (see post #49 "FET Selection" thread). I'm just in the process of taking a second look at it, as initially, the buffer amp seemed not to work. A number of criticisms were made in the thread about the front end of that design.

B

[Skywave]

Originally Posted by Bazz4CQJ

. . . initially, the buffer amp [in the G0WCA design] seemed not to work . . .

Barrie - the G0WCA design: right now, off the top of my head, no, I don't recall you mentioning it previously - I'll re-read your post in the thread you have referred to. But what is more to the point are the results I have had with that design: the FET buffer stage in mine works O.K., but after about 40 ~ 45 MHz or so, the output from it slowly drops off. But my 'investigation build' is a bit of a lash-up and the FET is a BF256A - which, to me, is the primary suspect for that fall in response. I intend to try a 2N4416 when I find the time (as is featured in my version of the Marconi probe). And if that doesn't improve things, a rebuild with due attention to established R.F. construction practices will then be tried.

What FET did you use - a 2N3819?

Al.

[Bazz4CQJ]

Yes, I used a 3819 as per the original. And because I thought the FET might be vulnerable to damage, I put the FET in a holder rather than solder it in, which was probably a poor trade-off. I used a PCB, again as per original.

It's around 2 years ago since I built it, and it got put in the pending tray when it seemed that it didn't achieve the high frequency response which was claimed (I wanted to go to 146MHz min). Since you mentioned it a couple of days ago, I've got it back on the bench .

The way he has the potentiometers for the 3 ranges in parallel - does that seem like an optimum arrangement?

B

[Skywave]

Originally Posted by Bazz4CQJ

I have built the G0WCA design (see post #49 "FET Selection" thread). I'm just in the process of taking a second look at it, as initially, the buffer amp seemed not to work.

Following on from my posts subsequent to that quote, I've made a few measurements on mine - which as I said, uses a BF256.

For a signal source, I used an HP8640B sig. gen., fed to the cct. via 50 Ohm coax, with a 50 Ohm termination at the end of that and then a very short piece of coax to the cct. under test. Measurements were made with a Tek. 2465 'scope via an HP X10 passive probe. Test freq. was 10 MHz; measurement results (as below) were subsequently found to be valid down to at least 1 MHz.

For 20 mV p/p, at the gate; 18 mV p/p at the source; a tiny change in the d.c. level at cathode of D2 (the rectifying diode).
For 200 mV p/p, at the gate; 180 mV p/p at the source; a change in the d.c. level at cathode of D2 (the rectifying diode) from 280 mV to (280 + 50) mV.

For the two diodes, I use HP 2800-5802 'hot carrier' diodes. For the bias diode, I had 0.7 v. at its anode. The load on D2 was a 50 k-Ohm pot. - as per the original design.

HTH,
Al.

[Bazz4CQJ]

OK, so it looks like your version with the BF256 is working as a unity-gain buffer to at least 10MHz, and from your earlier post, possibly up to ~40MHz?

For anyone who's interested, Skywave is working on the circuit attached, originally published by G0CWA on the net. In the original circuit, that feeds in to a CA3140 op-amp. In the original design, the FET is a 2N3819, and he claimed that it was good to at least 150MHz. My version was wasn't .

B

[Radio Wrangler]

"good" can mean anything between 'has a specification and meets it' and 'confirms that something is there'

David

[Bazz4CQJ]

I recall playing with 3819's in GDO circuits a very long time ago and forming the opinion that something better was desirable to work comfortably at 2m, though it seemed plausible that a unity-gain buffer was possible. As for the meaning of 'good', maybe G0CWA had some 'really good' 3819's?

B

[Skywave]

Originally Posted by Bazz4CQJ

As for the meaning of 'good', maybe G0CWA had some 'really good' 3819's?

Generally, it is quite well known and recognised that the mutual conductance and the pinch-off voltage can vary over a wide range with 2N3819 FETs. Perhaps G0CWA was lucky; perhaps he had several and picked the best one. However, the '3819 is an old design for a JFET: I'm sure that there must be more recent types with tighter controlled specs. and which will probably work to higher freqs.

Al.

[Radio Wrangler]

The 2N3819 was probably the breakthrough device for JFETS.

As Al says, the pinch-off voltage Gm and Idss can have huge variations within the device's registered spec.

Later devices weren't a lot better as made, but the manufacturers had different type numbers registered for different parts of the range so for any type number, the spread looked better at least.

David

[G0HZU_JMR]

I think the inspiration for the detector in the G0CWA circuit came from this web link
http://www.zen22142.zen.co.uk/Circui...tgear/rfmv.htm

The test results for the linearity of the detector for small signals don't appear realistic to me. I would say they are in fantasy land as the performance claim seems to indicate this thing has the linearity of a precision rectifier at tiny signals. eg at a few mV . I would expect this detector to work in square law up to a few tens of mV and then gradually trend towards linearity on the upper range. Up at the higher end, 1.5V rms seems a lot to feed into that circuit with the 2N3819 so I would expect to see some issues at the upper end too. The input JFET circuit is far from ideal and will have issues of its own wrt its performance as a 1:1 buffer and how well it can drive the detector stage.

The G0CWA meter also uses a linear scale and I don't think this detector/meter will be very linear. Also the design has no temperature compensation and also the JFET stage won't perform very well. The PCB layout is poor and the RF input connection and grounding doesn't look good. I remember this circuit from a recent thread and I thought it was so poor it wasn't worth analysing let alone building unless you just want something that works reasonably well as an RF indicator with a very casual dial calibration.