R.F. voltmeter: your thoughts, please

[jimmc101]

Originally Posted by mhennessy

...
Note that the Farnell TM6 is very different to the Levell TM6...

Sorry about that, a bit of a senior moment on my part.

Jim

[Argus25]

I can't think of any circuit that would beat the one published in Horowitz & Hill, just a few transistors and you can easily hang a fet on the front of it to get a high Z input. The idea of it is very clever & simple. It runs off +12 and -18V. Instead of driving the detector diodes with a voltage, they are driven with a current. For even the smallest signal input the voltage rises very high, overcoming the diode drops and making the diode non-linearity vanish. Figure 13.30 page 890 second edn.

[Skywave]

Well as we all know, there is no such thing as a perfect semi-conductor diode, especially at low voltages. The appropriate laws of physics tells us that. But in the category of semi-conductor diodes for the purpose under discussion here, I've found that the OA47 to be one of the best. It's main drawback is its relatively low P.I.V.

Al.

[Argus25]

Originally Posted by Skywave

Well as we all know, there is no such thing as a perfect semi-conductor diode, especially at low voltages.

In the case of the circuit cited, this is completely unimportant.

The diode doesn't have to perfect or linear at low voltages at all because its current source drive rather than voltage drive. This means that for any small voltage applied at the circuit input (well below the diode's conduction level) due to the way a current source works, the voltage rises very high in an attempt to maintain a current, overcoming the diode drops and controlling the diode current instead. So, for the most part, the diode's low voltage and square law properties etc vanish from the equation with this circuit.

I have not built the circuit in this version, the only version of this concept I have trialed is one where the transistor current source was replaced by an inductor.

I have attached the circuit, you could put a resistor at the input and use that for the source resistor of an input fet.

[Bazz4CQJ]

Hi Argus

thanks for posting that circuit, which I think is a new one, at least to me and I'm not familiar with Horowitz and Hill, sounds like something you pro's would have? Could you let us know what kind of bandwidth they report for that circuit and what kind of sensitivity they claim? The hi values of the caps in that circuit suggest it was aimed at AF applications (?) whereas we are more aimed at 1-150MHz.


I have no pre-conceived ideas with regards to this project, but now that a couple of published designs have been tried and found wanting, I am trying to be a bit more critical.

It seems that some companies claimed remarkable sensitivity for diode probes using no pre-amp, but intuitively , me thanks that some form of pre-amp would be useful / necessary.

B

[Argus25]

On page 889 it says "it will operate to 100MHz or more"

But they don't specify or graph the sensitivity, the text says exactly:

"With no input signal, the amplifier's output is decoupled from the rectifier network, producing very high voltage gain (with its current-sink load); thus,only a very small input signal is needed to turn on the diodes. At that point the voltage gain drops to Gv=RL/(RE +re) (in this case Gv=3), preventing saturation"

[Bazz4CQJ]

I think I have seen a circuit very much like that before, except that it's upside down (cf. Collins to Armstrong, "Someone's upside down!".)

Easier to look at if drawn right way up I think; I'll re-draw it. Am I right in saying that the input impedance is "lowish"? Some of these circuits were designed to go right down to low AF frequencies; guess we could reduce those capacitor values somewhat for this application and construct in a "RF-friendly way".

B

[Argus25]

Originally Posted by Bazz4CQJ

Am I right in saying that the input impedance is "lowish"?

Yes it would be low about (hfe +1)43 ohms, where hfe for the pnp input transistor could be taken to be about 50 roughly (but hfe is a poorly controlled transistor parameter), excluding any base to ground resistor you have that would lower it. I would say the 43R resistor could be increased to 75R without problems.

So with a 75R emitter resistor for the pnp input transistor I would guess the input impedance is about 3k8 very roughly.

I think the best move is to put a resistor there (base to ground at the input) say 2k2 to 4k7, and make that the source resistor for a jfet. Connect the Jfet's drain to +12V via a 100R and a filter cap.(or just connect it to where the 100uF cap connects) Then you have a source follower to drive this circuit. The fet's gate can be biased as you have a -Ve supply and you can make a divider along the lines of other probes with fets in the front end. An MPF102 is a good fet to start with.

The 100uF should have a 0.1uF ceramic across it as should the other electros, but I think also that their values could be reduced anyway and simply use 1uF monolithic ceramic caps throughout for an RF probe application.

[KeithsTV]

Here's the Radcom article requested by Bazz4CQJ via PM.

Keith

[Bazz4CQJ]

Many thanks for your help once again Keith. I think that's the only tuneable design I've seen for homebrew, though I think there were commercial instruments with that feature.

Interesting, though, that even with the FET's at the front, he's getting nowhere near the sensitivity claimed for some of the commercial designs (Marconi, Boonton), where they input directly on to the diodes.

B

[Skywave]

That RadCom design: aren't C3 and C4 erroneously transposed? And the one of that pair that decouples the +12 v. PSU rail: shouldn't there be an additional and higher value capacitance in parallel with it - and possibly a low value one as well - simply to ensure good decoupling over the freq. range that this circuit is designed to operate over?

Having said that, the idea of the unused L/C cct. having a 100 Ω resistor switched across it, so that the input Z approximates 50 Ω (plus some small inductive reactance) by transformer action, irrespective of which L/C cct. has been selected, is quite a smart technique. Or is my analysis of that arrangement incorrect?

Al.

[Skywave]

Originally Posted by Bazz4CQJ

Many thanks for your help once again Keith. I think that's the only tuneable design I've seen for home-brew, though I think there were commercial instruments with that feature.

Indeed; R.F. tunable voltmeters: the one that immediately springs to mind is the Rohde & Schwarz HFH. I recall using one of those in my early R&D days: superb instrument - but large and very heavy!

Al.

[Bazz4CQJ]

Unfortunately, I don't think we can apply the adjective "superb" to this meter. The sensitivity is just not there and the input impedance is low. The Search goes on ...and on...

[Skywave]

I built the circuit from the '73 magazine'. The results were very disappointing, despite a very careful construction and the matching of various components as descibed and as required. The freq. range was poor; the linearity was not only poor but was inconsistent from range to range; the switched ranges did not accurately correspond to the values of the resistors in that range switch.

Originally Posted by Bazz4CQJ

The Search goes on ...and on...

Well for myself, maybe. I've spent may hours investigating many approaches but none have resulted in a completely successful design. Overall, for me, the results have been an example of a Curate's Egg. I haven't totally ceased my research, but there other things in my life that I must now catch up on and attend to. For the time I have available, this project has now lost its priority status.

Al.

[Argus25]

Originally Posted by Skywave

I've spent may hours investigating many approaches but none have resulted in a completely successful design.

Curate's Egg ?? I had to look that up:

"In its original context, the term refers to something that is obviously and essentially bad, but is euphemistically described as nonetheless having good features credited with undue redeeming power"

But have you actually built and evaluated the current source drive system from Horowitz & Hill ? It might be a case of the assumed Curate's Egg and the solution is right there, in front of you.

One example of this sort of thing, it happened to me: It was said by many experienced commentators on a number of websites and forum's (including Tek's) that ferroelectric ram could not replace the battery backed up Sram in Tek 2465b oscilloscopes, because of differences in the addressing systems. But nobody checked the address scheme used by the actual firmware in the scope. I realized that not a single person, despite their remarks, had physically attempted to program an Fram, put it into the scope and see if it actually worked before coming to their "solid conclusions" that it wouldn't. It did work and the rest is history.

People get put off conducting their own experiments by other people's remarks and the solution can be right there all the time.

[Argus25]

...to add something to the last post above: If you look at the output part of Horowitz & Hill's circuit: On the face of it, it might look like a standard capacitively coupled two diode peak to peak AC voltage rectifier.... but it in fact isn't.
Notice the low value of 100R on its output. That value is far far too low for the rectifiers to work properly for the usual voltage detecting mode. The 100R is actually a "current sensing" resistor. This is also part of the genius of the design.

[Radio Wrangler]

Quote:
Originally Posted by Skywave

I've spent may hours investigating many approaches but none have resulted in a completely successful design.

Al.

Mankind has been developing instruments for measuring the level of RF signals for over a century and in that time many different operating principles have evolved, and many different ones are still in use today.

The reason is that no one type does everything you might want. True in the past, still true now. Looking for the one that does everything will lock you into a condition which has been adequately described in legends (Flying Dutchman, Holy Grail - even the Monty Python version).

R&S, HP, Boonton, Marconi Instruments, Siemens, Tektronix and dozens of smaller companies have made lots of types of RF level measuring instruments. None tick all the boxes, all have their uses and some are very specialised indeed (E.G. HP 3745A SLMS) some are just a diode and a meter. ALL cover a limited frequency range. ALL cover a limited amplitude range. ALL give limited information about the nature of the signal.

My first decent measuring instrument was an old AVO CT38 valve voltmeter. It worked well for my needs on the usual multi-metery sorts of things. The meter face was small and the case was huge. But it had an AC detector which was removable in a cumbersome probe housing which worked on a short, 2-foot adaptor cable as an RF probe. It contained one EA50 diode valve. I can smell the tropicalisation stuff just thinking about it. I hardy ever used the RF voltage function, it was simply too insensitive.

Today, courtesy of affordable stuff on the second-hand market and getting dead instruments for free and fixing them, I'm awash with test equipment.... and not one RF voltmeter.

What I do have are lots of alternatives:

Precision RF Power Meter.... Does very accurate measurement of level. Some use thermal techniques to measure true RMS. The heads I have cover 1uW to 10mW over 10MHz to 10GHz and I have attenuators to take them to 200W. Rolling-off below 10MHz is their biggest limitation, and they're all 50 Ohm input impedance. I have some diode type heads that go down to 100pW full-scale and up to 40GHz but are still no good below 10MHz and are limited to 50 Ohms. They all just read the total RF power going into them. I use them when their limitations don't matter

Oscilloscopes... I have several of these covering DC to 1GHz. A 100MHz Tek 465B lives centre-stage on my bench. It is what I reach for first. The GHz job is 50 Ohms, but all the others tick the high impedance box. Signals need to be several mV or bigger, but you do get to see the waveshape and that gives a lot more clues than you get from a pointer and a scale. The professional market dropped the RF voltmeter a long time ago and switched to scopes.

Spectrum Analysers... Essentially a calibrated scanning receiver with switchable bandwidth and a graphical display. I have one which will let me see down to a few Hertz, and one which goes up to 22GHz. These things are very sensitive (-145dBm 0.000000000000003 Watts) and I have attenuators to cover up to 200W. Level measurement accuracy isn't as good as the precision RF power meters, but they do allow you to unravel multiple signals, to examine modulation etc. They don't measure relative phase so you can't convert the results to waveforms or to true RMS values.

Directional RF Power Meter... This is a specialised thing to live on the output of a transmitter. It reads the power sent to the antenna and it independantly reads the power which is reflected back due to imperfections in the feeder and antenna. I have a few Japanese ones which use coupled transmission lines and simple diode detectors giving display on a crossed-needle dual meter movement. I have a transformer-hybrid coupled one which I designed for the G-QRP club (Sprat 61 with wrong callsign printed) which goes down to 1 Watt full scale and is useful over 1.5 to >60MHz. I have a fancy one using the ADI log detector chips built into a monster (read 'White elephant') ATU I built. This reads power on log scales over a very wide range, and as the cherry on top, it reads phase too. Very rarely used!

One of these instruments cost most of 100k bucks when it was new. In hobby terms, what I paid for it would have bought a new mid-range DSLR camera, or a new middle-quality leather saddle for a horse. Even it doesn't tick all the boxes...

One design won't do all Al wants, but several alternatives in concert will.

Horowitz & Hill 'The Art of Electronics' is an outstanding book. If you don't know what it is, Bazz, you don't know what you're missing. Its size is intimidating. The breadth of electronics covered is intimidating. Being written by a Harvard prof and a very senior researcher might be seen as even more intimidating. Actually it's a nice book which takes you from the basics right through to all sorts of stuff. The writing style is friendly and chatty. It's digestible. It and the current ARRL handbook are the two tomes I tend to carry around with me.

David

[Argus25]

Originally Posted by Radio Wrangler

The professional market dropped the RF voltmeter a long time ago and switched to scopes

Yes you are quite right. My 2465B scope can measure mV into its internal 50 R load and up to hundreds of volts with the x 10 probe and has a bandwidth of 400MHz. Also it has cursors for exact voltage readouts. It is a testament to the design of the vertical amplifiers in these scopes and it does make the notion of an RF probe somewhat obsolete.

[Skywave]

Originally Posted by Argus25

Curate's Egg ?? I had to look that up

I do not know what source you used for a definition of that reference, but from Chambers Concise English Dictionary, we have the following definition:

Curate's Egg: anything of which only parts are excellent.

Which is exactly what I meant when I used that phrase - which is substantially different to your definition. Perhaps, in Australia, the term has a different definition.

Al.

[Skywave]

Originally Posted by Skywave

I've spent may hours investigating many approaches but none have resulted in a completely successful design.

Originally Posted by Argus25

But have you actually built and evaluated the current source drive system from Horowitz & Hill ?

Not yet. As I have explained previously, the amount of time I can devote to this subject - which is very large - is quite limited.

Al.