R.F. voltmeter: your thoughts, please

[Bazz4CQJ]

Originally Posted by G0HZU_JMR

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.

Yes, as someone who did build it , my analysis is that your analysis that it wasn't worth analysing, is "not unreasonable", at least with regard to the front end. I am only an amateur!

B

[G0HZU_JMR]

I think the G0CWA circuit is an example of how it's possible to take individual circuit snippets that have the potential for good performance and then spoiling that performance by connecting them badly in terms of the completed circuit of snippets (if that makes sense).

For example, I've seen HP use a similar (Schottky) diode doubler circuit in the past for a precision meter that measures noise. It too had a sniff of forward bias at the base of the first shunt diode and at first glance it all looks very familiar. But HP used a current source to drive the circuit at a fairly high level and also the bias circuit was very carefully compensated for temperature. It also only worked over a narrow IF frequency band and it achieved remarkably good linearity when measuring noise signals over something like a 20dB range. eg less than 0.05dB error in linearity over the whole range. It looked like a really elegant circuit and I presume that HP corrected the (already tiny) linearity errors in a lookup table in memory somewhere. So suddenly a very similar circuit can achieve instrumentation levels of performance over amplitude and temperature

I am only an amateur!

Under the skin we are all just amateurs. I learned that lesson very quickly when I started my design career at my current workplace in 1990. I was rubbing shoulders with some incredibly talented engineers who started the company (they were ex Racal) and I learned an awful lot from them. But even these gods can have holes in their knowledge. Sometime the holes can be surprising and I don't think it's possible to be a guru or expert in all aspects of RF design. But those guys were pretty close! Obviously I have my own potentially embarrassing holes in my knowledge base. We are all just human

[G0HZU_JMR]

One of my concerns for the original VM in this thread in the 73 mag is that it probably won't deliver the same bandwidth for very small signals compared to large signals. It may well deliver decent VHF performance up at several volts rms but I'm not convinced that it will do the same on the bottom mV range. The bandwidth could be a lot less here. I don't have any 1N914 diodes here and certainly not any 1N914 diodes from that era but I'm not sure this diode type will deliver a wide detection BW down at the mV drive level. I don't think the original 1N914 diodes were designed to be used as small signal HF>VHF detection diodes? Has anyone here got any?

[Skywave]

The write-up for the G0CWA R.F. voltmeter contains the following extract:

Connect a DVM to the junction of D1, C5 and C6 gradually increase the setting of VR1 until a very low voltage is measured on the DVM: the lower the better. This has set the DC bias of the diodes to its optimum value.

Which has me puzzled. "A very low voltage - the lower the better" surely includes zero volts. Now presumably that is not what the author means. What is obviously missing is a specified voltage (or a voltage range, even if approximate) that should be aimed at by the adjustment of VR1 or some other criterion that is relatively easy to arrange and verify.

Suggestions, anyone?

Also - and this is a small point - "the DC bias of the diodes" (note the plural) presumably should state "the d.c. bias of the signal rectifying diode".

Al.

[Skywave]

Originally Posted by G0HZU_JMR

Under the skin we are all just amateurs.

That depends on your definition of an 'amateur'. There are many members of this forum who are employed in electrical / electronic firms and their knowledge and expertise provides them with a living. That is my understanding of a 'professional'. An 'amateur' may indeed have comparable skills, etc., but unlike a 'professional' that member does not use those skills etc. to produce an income for him / herself. So on that basis, it would seem that we (members of this forum) are not "all just amateurs".

But I will agree (with what I suspect was in your mind) that even accomplished and recognised experts occasionally make mistakes and that their knowledge and expertise does indeed have boundaries.

Al.

[G0HZU_JMR]

But I will agree (with what I suspect was in your mind) that even accomplished and recognised experts occasionally make mistakes and that their knowledge and expertise does indeed have boundaries.

Agreed. One of my favourite sayings about RF is that every day should be a school day if you can find the time. It is such a diverse subject and maybe I should have said 'under the surface we are all still at RF school'.

[G0HZU_JMR]

Originally Posted by Skywave

Suggestions, anyone?

Normally you would bias a Schottky diode (or a Schottky diode doubler) in this case with a few uA in order to reduce its video resistance (Rv) such that it can efficiently drive the 17k load resistance at the detector output.

The G0CWA circuit uses a fairly low load resistance at the output of the detector (17k ohm?) so I think you would have to push a current in the order of several hundred nA (at least) through the two detector diodes to get the Rv down. A ballpark guesstimate at Rv for each diode would be Rv = 0.026/Ibias so if you went for 0.5uA bias then Rv would be 52k ohm in each diode.

So the 'standing' dc voltage at the output of the detector would presumably be = 0.5e-6*17k = 8.5mV for 0.5uA through the diodes.

This bias level should give a much better detector efficiency than no bias although you might want to aim for a couple of uA. This would need the bias pot to be set for about 34mV standing DC (with no RF drive) at the three 50k detector output pots that are in parallel. Hope I've got all that correct. it's late but I could dig out some 1N5711 diodes and measure some on a VNA with varying bias levels tomorrow.
Even with several uA bias I'd still expect the detector to show a square law response for very small RF drive levels up to maybe 25mV so you won't suddenly be getting 'linear' detector performance because of this bias. You will just be pushing the detector efficiency sideways in your favour. Or at least that's what I think should happen. The original designer of this detector claimed it would be linear even at 1mV of RF drive once biased but that isn't realistic. I think it will tend towards 'reasonably linear' at RF signals of a volt or so but even then there will be some Vdrop loss in the diode. So it will never do what the original designer claimed and produce Vdet = 2*Vpk at the detector capacitor.

http://www.zen22142.zen.co.uk/Circui...tgear/rfmv.htm

Also - and this is a small point - "the DC bias of the diodes" (note the plural) presumably should state "the d.c. bias of the signal rectifying diode".

Not sure what you mean. The detector uses two diodes arranged as a voltage doubler. So they are both detector diodes and the bias current will be the same through both of them as they are in series. Both diodes detect and contribute towards the overall detector voltage.

[Bazz4CQJ]

Originally Posted by Skywave

The write-up for the G0CWA R.F. voltmeter contains the following extract: Connect a DVM to the junction of D1, C5 and C6 gradually increase the setting of VR1 until a very low voltage is measured on the DVM: the lower the better. This has set the DC bias of the diodes to its optimum value. Suggestions, anyone?

Regarding the biasing of the diodes, my reading was that supposing you put in a lowish level of RF (10-100uV), then increase the bias on the diodes, presumably there is a knee at which the output voltage rises (sharply?), and there is no benefit to biasing past the knee?

Regarding amateurs in radio, if you think of the days when the British Army was fielding No 19 sets, any experienced amateur could master its operation and servicing. Whatever the Army is using now, I guess that most amateurs might struggle to switch it on. I was licenced back in '69, then left the hobby for a really long time, and when I came back the gulf between hams and pros had grown massively. Properly understanding modern solid-state circuits can be hard. Perhaps this is one reason why so many are attracted to working on Vintage models?


B

[Skywave]

Originally Posted by G0HZU_JMR

Not sure what you mean. The detector uses two diodes arranged as a voltage doubler. So they are both detector diodes and the bias current will be the same through both of them as they are in series. Both diodes detect and contribute towards the overall detector voltage.

Ah! I missed the fact that D1 is acting as a rectifier, so the two diodes produce a voltage doubler, as you have said. I saw D1 as a half-wave. rect. with D2 simply feeding in the bias current to D1.
All clear now: thanks.

Al.

[G0HZU_JMR]

Originally Posted by Skywave

Ah! I missed the fact that D1 is acting as a rectifier, so the two diodes produce a voltage doubler, as you have said. I saw D1 as a half-wave. rect. with D2 simply feeding in the bias current to D1.

In order to describe how this doubler/detector works I guess a lot depends on how you define 'detector' or 'rectifier' or 'level shifter'.

If the diode doubler/detector on its own is considered (no JFET buffer) then for very large signals (several volts rms) the shunt diode could be viewed as a level shifter that clamps the bottom of the waveform to a fraction of a volt below 0V. The (detected? rectified?) DC offset of the waveform will then be shifted up to approx Vpk. Then the series diode will detect the overall Vpk of the level shifted waveform. So the DC output will be approx 2Vpk. In reality there will be Vdrop losses and the detector won't be linear for small signals.

[Bazz4CQJ]

It seems to be a common feature of some (all?) of the commercial RF voltmeters that the probes have internal heating which I assume avoids the diodes swinging up from room temperature to somewhere near body temperature when the probe is clasped in someone's hand.

I've just done a rough and ready test to see how big this effect is on Ge point contact diodes over the range 20-40°C and the results are attached.

For a professional grade instrument, it looks like they had done their homework.

B

[Bazz4CQJ]

Guys, some time a go I mentioned a design which had orginally appeared in Radcom and, thanks to Keith's help, we finally have a copy of it, see Post#2 https://www.vintage-radio.net/forum/...d.php?t=132584

There is a derivative circuit using npn transistors at http://www.mw0llo.com/milliwatt.aspx.

It's just arrived; have not yet dug in to it.

B

[Skywave]

The thread entitled Re: Simple RF probe for DMM has been closed and is now linked to this thread. That closed thread refers to a R.F. voltmeter by GW3JGA. For continuity, here is a copy of the relevant link:

https://billingtonrepairs.files.word.../voltmeter.pdf

Al.

[Skywave]

A close exam. of the GW3JGA voltmeter shows that it is basically a high-input Z D.C. voltmeter, with the optional provision of an A.C. probe for the measurement of A.C. voltages up to at least 100 MHz.

However, for the purposes of such H.F. measurements, it seems that its use will be strictly limited to low-impedance circuits. That is because at the input of the probe there is a diode between that input point and 0v. / ground. So the diode is acting as a shunt half-wave rectifier, thus presenting a very low impedance to the circuit under measurement.

Or am I mistaken?

Edit.
Upon review, I see that the previously-discussed R.F. voltmeter cct. that appeared in '73 magazine' uses a similar technique, leading to the same limitation.

Al.

[Bazz4CQJ]

Hi Al, well would it be any better if the probe was re-arranged as a series arrangment?

Alternatively, what happens if you put your HF-VHF amp in front of the diode probe?

Have you taken a look at the Radcom circuit I referred to a couple of posts above? I have started construction of that; I bought a bag of 100 off 2N3906's a while ago, and have to do something with them .

B

[Skywave]

Originally Posted by Bazz4CQJ

Alternatively, what happens if you put your HF-VHF amp in front of the diode probe?

That is a very relevant Q. Up to now, the general idea of doing that with any detector / amplifier / metering arrangement has always been on the drawing board. However, there is one problem with that design for that purpose: its dynamic range is of the order of 45 dB and to cover a useful and sensible range of input signals, it really need to be a good deal more. Now I've mentioned that dynamic range, the logical Q. is how did I measure it? In a word, only approximately, since I don't have the kit to make an accurate assessment: if anything my method was rather crude, but better that nothing. Essentially it consisted of using a sig. gen. for the amp.'s input and a 'scope on the amp.'s output and examining the waveform.
The idea of using fixed value attenuators at the amp.'s input will probably be problematical, since the amp.'s input impedance will vary over quite a large range with frequency.

Originally Posted by Bazz4CQJ

Have you taken a look at the RadCom circuit I referred to in a couple of posts above?

Not yet.

Al.

[Skywave]

Originally Posted by Bazz4CQJ

Hi Al, well would it be any better if the probe was re-arranged as a series arrangement?

I'm not sure what you mean by that: can you clarify, please?
Thanks.

Al.

[Argus25]

Skywave,

I have attached a circuit that might help on a wide dynamic range RF probe, these, now very vintage, were once very popular due to the simplicity and wide band response.The basic circuit appeared in hobby magazines etc.

The attached circuit shows how the RF probe is made very wide band +/- 3dB at 10kHz and 700MHz (flat in between) and converts the signal to its rms value. It works over 0.25V rms to 30V rms input range, and can handle 200V DC.

The operating principle is based on the non linearity of CR1 being cancelled by CR2 for low level signals. I think most likely they are schottky diodes, but I don't have any exact information on them. The circuit results in 0.7071 of the peak value appearing at the junction of R3 and CR2.

You could always build something like this, then for signals lower than 0.25V rms simply switch in say a x 10 or x 100 wide band amplifier at the input (like a mini-circuits job), depending on how low you wanted it to go. The output could be fed to a meter via an OP amp buffer set up with a 10meg input impedance. Any amplifier added in front of this could simply have its input protected by back to back diodes.

[Skywave]

Thank you Argus - that is worthy of investigation which I will do when time permits.

In the meantime, I have just reviewed the two broadband amplifiers I mentioned earlier. One uses 2 x BFY90 and the other uses 2 x BFR91A; each use a 2N4416 JFET. (These designs were derived from the Marconi TK2374 R.F. probe). With a few changes to the designs and corresponding adjustments, I was able to extend the dynamic range of each from 45 dB to 60 dB (approx. figures). Although the BFR91A variant has a higher freq. -3 dB corner, an initial exam. reveals that the BFY90 version gives a better dynamic linearity over the freq. range of interest: < 1MHz to 100 MHz (or thereabouts). So at present, the BFY90 version will now be subjected to more detailed & searching measurements.

Al.

[Skywave]

For reference: my above post refers to post #22 in this thread, q.v.:

https://www.vintage-radio.net/forum/...d.php?t=131572

which I have requested be closed. That leaves only this thread here that I will be posting to and is effectively the thread where other threads that I have produced, related to this topic, all meet. That keeps everything nice 'n' tidy and easy to manage - for myself, other members & the moderating team.

Al.