FET selection

[G0HZU_JMR]

In case anyone is wondering why I keep mentioning stuff about negative resistance/impedance etc then this is because the probe can (in theory) go unstable at a certain frequency if you probe a certain load impedance if there is negative resistance in the input stage of the probe/buffer amp.

A whole branch of oscillator design and research is based upon creating a circuit that generates a negative resistance at its input and then you attach a tuned circuit to this negative resistance and you should get an oscillator if you get the design right.

So obviously, you don't really want a probe that has a negative resistance as part of its input impedance. The other reason is that you can get slightly fake results even if the probe doesn't go unstable. There is the possibility of probing circuits in a certain impedance range at a certain frequency range that will actually give regenerative 'gain' when the probe is attached. So the probe can in some rare circumstances make the signal look slightly bigger than it really is once the probe touches the circuit. However, this effect is probably not as significant as the possibility of probe oscillation when probing certain circuits.

[G0HZU_JMR]

Just a quickie for me for now. I found the Marconi probe cct. on the 'Net just few days ago: it is now in my list of ccts. to try. Incidentally, on the 'Net, one datasheet for the E300 F.E.T. stated that the E300 can be replaced by a BF256A.

Thanks! I might look into that to see if I can improve the input capacitance of my TK2374. I think the J310 is causing it to go out of spec for input capacitance. I suspect that I may have to 'select on test' a few of them to find one that drops in with minimal tweaking though

[kalee20]

Originally Posted by G0HZU_JMR

In case anyone is wondering why I keep mentioning stuff about negative resistance/impedance etc then this is because the probe can (in theory) go unstable at a certain frequency if you probe a certain load impedance if there is negative resistance in the input stage of the probe/buffer amp.

Absolutely. Even a simple source follower or emitter follower can show a negative resistance at its input, if it drives a capacitive load on its output. It's analogous to a simple earthed-cathode triode amplifier, driving a capacitive load, looking resistive at its input and damping a tuned-circuit connected to the input.

This thread is great, I'm really interested in the reports on the highly-bootstrapped input JFET source follower (bootstrapped back to drain, as well as to gate bias network)!

[G0HZU_JMR]

I did a LF analysis of the classic bootstrapped FET in post #25 and I optimised the setting of the 2.2Meg resistor to get lowest (parallel) input capacitance and highest (parallel) input resistance. This setting of the 2.2Meg resistor will vary from JFET to JFET but I managed to get a very impressive result on the simulation. It's difficult to test this on a real circuit but it managed 0.23pF and over 100Meg ohm. Note that to get it to work well below 1kHz I had to increase the output cap to 1uF and also the bootstrap cap to the drain was increased to 1uF. I also used a 1k ohm output load and not 50R. I also used a jellybean NPN transistor and not a UHF BFS17 model. The noise contribution when probing high impedance circuits was negligible. Very impressive! However, the probe will have a high noise figure when probing 50R circuits up in the RF region.

But with this setting of the 2.2Meg resistor the circuit shows an extreme level of negative resistance up in the RF region. eg the negative resistance kicks in above a few hundred kilohertz. So this isn't the same setting of the 2.2Meg resistor that I would use if I wanted to use the probe up at RF.

I think this probe is best suited for use up to maybe 200kHz if you want/expect ultra low capacitance and very high input resistance. It might be possible to hold off the onset of negative resistance by using something like a ferrite within the circuit but that's just a guess. Otherwise, the 2.2Meg resistor will have to be set such that there is no negative real part in the input impedance. That's the way I've set the probe up to get it to be very flat up to several hundred MHz but the input capacitance is more like 2pF in this case.

[G0HZU_JMR]

Absolutely. Even a simple source follower or emitter follower can show a negative resistance at its input, if it drives a capacitive load on its output. It's analogous to a simple earthed-cathode triode amplifier, driving a capacitive load, looking resistive at its input and damping a tuned-circuit connected to the input.

Yes, to show how easy it is to make this probe oscillate up at VHF I measured the real circuit on the VNA and exported the data as an s2p file. I can then look at it to look at the amount of negative resistance on an RF simulator. In the instability.doc attachment below you can see in the top right corner that there is about -50R resistance at 190MHz when the buffer is analysed at its input and I adjust the 2.2Meg resistor to show a sniff of negative resistance here.

In the lower circuit in the doc you can see I've added a shunt inductor to the buffer that measures 290nH at 190MHz on a VNA. It's really more like 230nH with a fraction of a pF self capacitance but this is equivalent to a pure 290nH inductor at 190MHz.

In the lower left graph this shows the input reflection coefficient going very positive at 190MHz once the inductor is inline. The phase angle needs to be at either 180degrees or 0 degrees to satisfy the oscillation requirements for a 1 port negative resistance analysis. You can see it is bang on 180degrees.

This predicts the circuit will oscillate at about 190MHz with this inductor fitted. It can also be predicted with an even simpler analysis from the 2.4pF input capacitance of the JFET buffer. This is shown in the top left corner and the VNA shows about 2.4pF input capacitance for the buffer. This 2.4pF forms a resonant tank circuit with the 290nH coil and the -50R negative resistance at 190MHz keeps it topped up with energy and so it can keep oscillating

To demonstrate that it really does oscillate here I connected the shunt inductor to the buffer and you can see on the analyser plot below that it does indeed oscillate at about 190MHz. Stuff like this does take some of the mystery out of instability and it's nice to see it all work out as expected

[Bazz4CQJ]

Great work Jeremy; we used have a VNA at work and it was something which I never mastered. The thread has come along way; Al's OP simply sought information on FET's, but where we're at now shows that even the right FET needs a really well-sorted circuit. Construction hasn't been discussed in any great detail, but I'd guess that's going to be critical as well.

Are we now at a stage where Al/others need to spec the probe in terms of upper and lower frequency range (I got the impression from the OP that LF was not needed), perhaps 1-150MHz would be a useful target? Does the work to-date suggest that the Marconi circuit could be the most promising candidate?

B

[G0HZU_JMR]

I guess so, but I've been a bit distracted because I think I've found the Bob Pease probe circuit proposed by MrBungle earlier in the thread. See the image below. I couldn't resist having a play with this on the simulator and it is a very interesting circuit. I think I'm going to make one because it looks like it is designed to be an active scope probe. My initial fumblings with it on the simulator show that it promises about 0.5pF input capacitance and crazy high input Z without the input shunt resistor when analysed up to and just above the AF region.

But above this it dives strongly into negative impedance where the real part becomes strongly negative. However, I've played with some faster BJTs as suggested by MrBungle in his post and I think the negative resistance can be managed with a fairly large series resistor at the input. Maybe 1500R?

If I had to choose which one to use as an active scope probe then this one is a clear winner as it works down to DC and looks to have the lowest capacitance over the RF frequency range and it offers 1:1 transfer in terms of voltage. It does need a dual supply which is a bit of a downer but I think the performance it offers makes it worthwhile. I think the bandwidth will be limited to about 100MHz because of the series 1500R resistor I've added to tame the neg R issues on the simulator. However, there may be other ways to improve it here. This would be a great active probe to use with something like a Tek 465 scope when aligning solid state Ham/SW/CB radios because it will not pull the circuit as much as a x10 scope probe and you can attach the counter to the CH1 out at the back of the 465 scope. So you could see the waveform on the scope as well as measure the frequency on the counter.

Not sure when I will have a go at making one, I've got to buy some fast PNP transistors and I think I'm going to make it such that it could be used as a real (but ugly) active scope probe.

I think the Marconi circuit is a good /safe/predictable circuit if you want to make a basic HiZ RF probe for a 50R spectrum analyser.

You can also make x10 and x100 voltage range extender tips for it just like the ones in the official probe kit allowing it to work with higher RF voltages without damage. But the Bob Pease probe looks to be (easily) the best probe if you want to use something with a scope. It will probably work well with a 50R spectrum analyser as well with maybe an extra buffer stage?

[Skywave]

Originally Posted by Bazz4CQJ

Are we now at a stage where Al / others need to spec the probe in terms of upper and lower frequency range (I got the impression from the OP that LF was not needed), perhaps 1-150MHz would be a useful target?

Yes: I think we are. I'll explain: here's the full story.

First, although this cct. has now acquired the status of a 'probe' - which it could certainly form the basis of - that was, and is not, my intention, but I can see that for what I have in mind (as follows), it is a distinct possibility.

Second, the initial general idea was to see if I could construct a broadband a.c. only amplifier for use with either an oscilloscope or the 'front end' of an R.F. voltmeter. In either case, the use of a conventional X10 'scope probe sprang to mind, to be fitted at the buffer's input. (Now thinking along the lines as per my first para. above). With these two possible applications in mind, I have a number of commercially-built broadband amplifier modules, all of which have gains of 20 dB or more, but have input and output impedances of 50 Ω. Which, of course, is what one would expect. (N.B. The noise figure of any of them is not known). So a high input-Z buffer, between probe and such an amplifier is needed. For 'scope use, the input of the buffer will probably need the addition of a trimmer capacitor, so that the probe 'thinks' it is 'seeing' the 1MΩ input plus a few pF of the 'scope. For a R.F. voltmeter, a similar approach may be required.

Third: as is the case with many R&D ideas, initially it was simply a question of quickly connecting circuit 'blocks' together and then seeing how far the idea developed. (Run it up the flag-pole and see if it flaps! ) When I started the work on this buffer, I had no idea at all that its performance would fall short and that it would develop into such a saga, although I do appreciate the assistance provided here. I was anticipating a fairly straight-forward 'build, test and confirm' type of exercise. All the available literature about it, although not very detailed, certainly gave the impression that its B/W would be 'flat' up to about 100 MHz - or more. And that is my requirement at H.F.: 'flat' to 100 MHz. At L.F., a 'flat' response down to 1 MHz (approx.) should be adequate.

Aside. This project is being 'time-multiplexed' with a few home improvement jobs, so responses from me are not to the usual rapidity that I normally try to adhere to. If applicable, your patience will be appreciated: thank you.

Al.

[Bazz4CQJ]

The project I started with in this context was an RF voltmeter supposed to work from '100kHz to (at least) 150MHz', but that used nothing more than a humble 2N3819, with no source follower ahead of the DC amplifier, strung out on a fairly "casual" PCB (see attached PDF).

That design seemed to have some "heritage", but it didn't work for me, and after following this thread, I wonder if it's worked for anyone .

Anyhow, the project has been dormant for some time, so I'm more than happy just to 'look and learn' as the thread evolves .

B

[Radio Wrangler]

I think that circuit is somewhat optimistic. I don't see it making the mentioned frequency range, nor is there anything to do the RMS bit of the RF to RMS marked section.

With a lot of published circuits, even in fairly respectable magazines, it takes roughly as much knowledge to determine whether something is worth building as it would take to design your own.

David

[Skywave]

Originally Posted by Radio Wrangler

With a lot of published circuits, even in fairly respectable magazines, it takes roughly as much knowledge to determine whether something is worth building as it would take to design your own.

Although I dislike 'negativity' such as that, since it tends to inhibit creativity, simple experience leads me to agree with David. And yes, I do realise that in most books and publications, space is limited. However what is often lacking is not only a detailed spec. of what a project will (and won't) do, but how its claimed specs. were determined - a list of test equipment used and procedures followed would often be useful.
The only 'plus' side I can think of that results from finding that a completed project fails to perform as claimed (or implied) is that it can cause the builder of it to investigate 'why?', and thus hopefully expand that builder's knowledge. And that situation illustrates the value of a forum such as this.

Al.

[Radio Wrangler]

Indeed!

Life was a lot harder in the days before the internet. Dodgy circuits in magazines affected builders in either if two ways:

1) I must be useless! it doesn't work. I'll take up stamp-collecting.
2) It doesn't work!, now why? read, read, read, learn, learn, lean. Ah, that's why! I'll throttle that author/editor if I ever meet him! it could never have worked.

By a process of elimination, I think those of us on here who learned how to be effective at electronics must have come from group (2). But now we can provide help to those at risk of going down path (1) and save them from a horrible fate

David

[Bazz4CQJ]

But surely the circuits appearing in magazines were subject to some editorial review, whereas circuits and designs now appear on websites and blogs which are just down to one individual, so the risks trying one of those are possibly greater?

Highly experienced professionals are capable of designing their own circuits, but we amateurs, who are butchers and bakers and candlestick makers by day, used to depend on the likes of F.G Rayer for designs in PW etc, but those days are long gone, and the gulf between pros and amateurs is now enormous, as evidenced by the utilisation of a VNA. What was that line about "things we don't know that we know about, and things we don't know and don't know about" ?

B

[julie_m]

Magazine "quickies" (and some longer articles; but especially the sort consisting of a few short paragraphs and a credit-card-size diagram) depend on actual circuits constructed in real life by readers. And the simpler a circuit is, the more likely it is to depend on actual values of individual components; for instance, a transistor having a certain hFE under biasing conditions that themselves depend on the tolerance of a resistor and the state of the battery. So just because the prototype worked, doesn't mean that anyone else's build will work! An experienced design engineer working in the industry and aware of typical component tolerances probably would include some extra components just to make the circuit less dependent on individual components (and probably have to justify this to their manager; but that is a separate story in its own right). But when you have only built one, and it worked, you don't always then go on to investigate the effects of individual components. Again, if one was working on a safety-critical system, an analysis of the consequences of individual components being out of specification would be a standard part of the design process. But this probably would not be the case for (say, for argument's sake) a simple circuit cobbled together on a postage-stamp-sized piece of copper-strip breadboard to act as a low-pass filter to improve the sound quality when recording some old, worn records onto tape, and which worked well enough in practice to be worth sending in to a magazine to share with other readers (and maybe win £10).

As I believe I have mentioned elsewhere, there is a very fine line between making the best use of the materials at one's disposal, and a crude and ugly bodge.

[ionburn]

Things go wrong but I don't like to be pessimistic. Don't forget the people who designed that mobile phone which they recalled recently were professionals! They really blew it!

With this FET thread for instance, it is easy to cross reference all over the place on the internet and components are so cheap at the moment that it is easy to try a few approaches.

How do we learn but to explore!

[Radio Wrangler]

Originally Posted by ionburn

How do we learn but to explore!

Exactly, that's where the fun comes in... and there's always something more interesting around the corner.

I'm definitely on the professional side of that fence which others believe in somewhat more strongly than I do. I used to design VNAs, spectrum analysers, noise figure analysers and all sorts of oddball telecoms stuff. It was great fun and very interesting. I went through an educational process which stuffed me with info, but to be honest I learned more from working amongst some amazing people. So, I can do the maths, the formal analyses, the formal design processes, but I have learned that they aren't the bit where the greatest value comes in. The creativity bit you have to do yourself, without any formal crutches to lean on. That's the part that gives you the feeling of real achievement.

So, with all the tools in my head and on my bench, what goes wrong?

Well, as a professional, you get to design what the firm decides it wants, not what you want to design. This can be exceedingly frustrating.

You sometimes have to design things with limitations that you expect to cause bad results. I suspect that Samsung's designers had hard targets for battery life, output power, backlight intensity, mechanical dimensions allied to very demanding timescales. If designers are hemmed in from all sides with no manoeuvring room, then what you get isn't likely to be good. It can be very very frustrating to be caught up in something like that. Also anyone who says 'It isn't going to work' has just toasted their career. So people who really know whether something is going to be OK or not do not say before things do go wrong.

Lots of engineers get home and want nothing to do that even vaguely resembles work. Others get home and relish the opportunity to design something for themselves without their hands being tied.

The Bill and Dave school of management went something like "Find and hire the best people, then trust them to want to do a good job. Try not to get in their way while they do it. Get them help if they ask for it." It was the complete antithesis of micromanagement and it not only worked, it built one of the best firms in its field. After they'd gone, the management theorists started to circle and the golden eggs dried up.

The pros get the fancy tools and some bits of sheepskin. The amateurs get the freedom.

So I try for the best of both worlds. I've collected an array of fancy tools at home. The sheepskins are rolled up in an old chest of drawers in the spare bedroom, and I get to play with things I want to do. No-one is paying the piper, and the piper gets to play his own tune.

At this time, there is a lot of fancy test equipment available and affordable - especially if you keep your eyes open for opportunities.

The most important thing to remember is that if I did it, anyone can. Cultivate curiosity and a sense of fun and go go go go go!

With the internet, we can now all tap into the sort of advice that couldn't have been imagined in the seventies. I don't need much advice on how things work, or how to design a wotsit, but I do need advice like where to get a thingy.

David

[julie_m]

Originally Posted by ionburn

How do we learn but to explore!

Well, exactly! There is nothing to say that a mere hobbyist cannot do FMEA-type "what if" calculations, if they feel so inclined. In fact, some of the larger "multi-in-one" project kits from Tandy used to encourage this thinking (What happens if you replace the 47 kΩ resistor with a 39 kΩ one? What happens if you replace it with 56 kΩ? Can you think of a use for this?) depending, obviously, upon what parts were still available for further experimentation.

[Skywave]

"If you want creative people, give them time to play": John Cleese.

Al.

[Skywave]

Originally Posted by Skywave

Post #31 refers:
Right now, my stock of F.E.T.s (all types) is low: time to buy some more, especially 2N4416, and try again.
Nothing else to add for now.

And now there is!

I bought some 2N4416 and some BFR91A from a 'known reliable' source, fitted them to yet another breadboard prototype and evaluated the resultant freq. response. There was no significant change: 'flat' to about 55 MHz; -1 dB @ 60 MHz; -2 dB @ 70 MHz (all figures approx.). I tried varying the bias on the F.E.T, but as before, no improvement was obtained. With these transistors I was expecting a 'flat' response to 70 MHz, at least. So I can only conclude that the style of my build must be introducing unwanted 'invisible' reactance (probably capacitive) which is limiting the H.F. 'end' of the B/W.

I am now very close to accepting that H.F. limit and 'moving on' to develop the overall requirement concept further. Despite what my signature states, my patience (etc.) is not inexhaustible and there is nothing that is critically sacrosanct in a response that is 'flat' to only 50 MHz (or thereabouts). And within the "overall requirement concept", there are still unknowns that need to be explored and determined. But that's half the fun of amateur electronics R & D, yes?

Al.

[Bazz4CQJ]

Perhaps you could add 'pragmatism' to your toolbox - I think it's a virtue. And in a week when Leonard Cohen and Robert Vaughn have sadly left us, it's clear that time is limited so persistence should be too ?

B