Alan, enjoy your circuits. I have a couple of question on the RF sniffer. On the photo of the board, does the main sniffer coil go through a mutual coupling? Or does the coil connect direct to the circuit?

Also is that a ferrite bead on the transistor?

Chris VK3ABU

Bert,

Got the schematic, sorry for not replying sooner.

The general approach seems OK. I don't think it needs to be that complicated. I like bridge-style circuits, they tend to be sensitive yet robust against thermal variations if both bridge arms are isothermally connected.

I'd use the dual diodes of the NJQRP sniffer with the op-amp differential amplifier. You could bias the other side of the meter with some diodes like you've shown or perhaps a follower from the reference diode so. Offset voltages might be a problem and how stable they would be with temperature changes once trimmed would limit the DC gain you could apply.

A rather more complex solution with no baseline drift would be Dicke-switching arrangement with a synchronous integrating detector. It might be PIN-diode switched. If such an insanely sensitive and stable device would actually be required is doubtful.

RF-side wise I'd be tempted to use a MMIC if current consumption isn't a consideration. They are ultra-simple to use. Otherwise a common emitter/source amplifier is useful at lower frequencies. Common base/gate or cascode might be easier to keep stable over a large range of frequencies.

Thanks and I will send the image through email shortly.

Bert,

Email me the file and I'll attach it to this discussion.

It has occurred to me in the past that it would be handy to allow small attachments of MIME type image/* (etc) to the comments section. You can of course link content, but I'd prefer to store it locally so it won't become a broken link over time. While there is an ugly method for me to attach objects to a comment, everyone should be able to do it as a basic feature. Pictures do indeed help communicate ideas.

I'll add it to the feature list and support it in the near future... I should probably add MathML support too, and some kind of feed on the comments. I get comments on very old articles which are often quite valuable and should be pushed to the homepage as some kind of "recent comments" call-out.

Regards,

Alan

Currently I designed another RF sniffer circuit, based on your circuit. I want to share it and discuss it with you, so can I post the schematic diagram here and how?

The main thing I want to discuss is about the common mode input(differential amplifier mode) of an op-amp, since I am new to single supply op amp circuits.

Thank you very much.

After playing with this detector topology, I can conclude that the detector diode is not operating very good when biased at the "knee" of the diode curve. So, the best way to increase its sensitivity greatly is to use a RF amplifier first before detecting, right?

Currently I am working on a RF amplifier, hope that it will have better sensitivity.

Bert,

Yeah an amplifier in front should help with the sensitivity. You could use a MMIC or a transistor. A coarsely tuned-collector CE stage might be a way to cover a particular band. It becomes more like a TRF receiver as you improve the selectivity and sensitivity.

One option, especially for VHF and higher, might be to use a super-regenerative detector. They can be set up to consume a current approximately proportional to the log of signal strength, so it would give you a logarithmic displaying detector as well as extremely good sensitivity. My tone dipper is somewhat like this, but with an audio frequency output instead of a moving needle. I've long thought about using this feature of the super-regenerative detector to implement a log-IF detector for an ultra-simple toy spectrum analyser or a selective field strength meter.

A completely different and superior broadband approach would be to use a log-detector IC like the AD8307. It has at least 90 dB of dynamic range and can see down to the thermal noise floor of its 500 MHz bandwidth. It can be used in narrow-band service too and has about 1 dB linearity and accuracy - pretty much the ideal device for small-signal RF power measurement. The device is about $7 US, but works wonderfully, I've built a (so far undocumented) power-measurement head with one.

Regards,

Alan

Thank you very much for the info you gave!

Nevertheless, I admit that I am crazy because I still want to push the sniffer to its limit, by making it able to detect "field strength" of radio stations.

I think a simple BJT common emitter amplifier made out of BF199 is good enough to amplify RF by capacitive coupling from antenna, right?(What I mean is RF is coupled capacitively into the amplifier, then send to the tank circuit to select a frequency, then to detector)

Bert,

The easiest way to make it more sensitive is to tune the front end so it is resonant at the desired frequency. Even fairly low-Q tuning will make is much more sensitive - at the cost of improved selectivity of course. My wave meter version is almost too sensitive but has to be tuned to locate the signal. It gives you more information, but it doesn't tell you is something is just making RF or not.

You won't be able to build a sensitive DC-daylight type sniffer without more amplification really, and making a broadband pick-up inductor is nearly impossible, so all practical sniffers are a bit selective. You can try fiddling with the inductor to make yours more sensitive without explicitly tuning it, or put a trimmer capacitor across the inductor and try tuning it. The improvement at resonance should be quite dramatic.

Op-Amp wise take a look at the njqrp sniffer. It uses a bridged arrangement to detect the difference in voltage across two diodes, one with RF applied the other not. It also has a tuned front-end, but you could try it untuned if you like.

All depends on what you are trying to achieve. Mine is built for a fairly broadband response, reasonable sensitivity and most of all very low quiescent current consumption (no off-switch required). Also for circuit simplicity I guess - read me being lazy :) Using an op-amp can make it more sensitive (and easy to vary the sensitivity, compensate for temperature changes, etc), but will pull a bit more current and probably have a more complex topology. I'd use an op-amp with a common-mode range that includes the -ve supply rail, as you'll likely be using input voltages close to "ground".

Regards,

Alan

Just now I thought about using op-amps to amplify the signal. Here comes the problem, that is how can I bias the detector diode without affecting the op-amp's input in a bad way? How much gain would I need? Would a JFET input op amp(I have TL072) make the biasing easier?

OK, done with the darlington transistor, now I found that it is not very sensitive, although I am using a 100uA meter. It only responds sensitive enough to VHF and lower frequencies. With the bias, the needle stays at about 20% FSD. It reaches full scale only when its coil is 1cm from my 100mW fm transmitter. How's about yours? What improvements can I make?

Bert,

Darlington's have two base-emitter drops between their "base" and "emitter", so you need about 1.2 V on their base with respect to the emitter before they will conduct significant collector current. Try few more dropper diodes to raise the base voltage, although three should have been enough... Maybe you need to stand more current through the bias string because you are operating far down on the non-linear "knee" of the diode curve at low currents.

You could try a LED or other device to drop more voltage, a red or orange LED should be about right, at a very feeble current that won't make it emit much light. If the general idea isn't to save current so there is no need for an off-switch then you can just run more current or even use fixed resistors.

Regards,

Alan

For your information, KSP13 itself is a darlington transistor. Sorry for the multiple posts.

Hello again.

I think I have some problems biasing darlington transistor. I used three 1N4148 and 1Mohm for the biasing, but it seems that it is not biased well,ie no deflection on the meter. What is going on? I am using KSP13 for the transistor.

Thanks.

Yes! I just get it to work. I used a BC549B, since my newly-bought BC550C is rubbish, which shows itself as a PNP transistor. How could I get such transistors?

Not suprisingly, the circuit could only detect a SMPS in a few inches. But suprisingly, the circuit seems like a better receiver for RF that come from a mobile phone(on a call) than the coil itself,why?

Thanks a lot, and i think I will not need the figures from you.

Warmest Regards,

Bert

Bert,

I didn't get around to it this weekend, but I'll measure a 1N5711 for you shortly and post the figures.

The oscillator took about half an hour I think, maybe more. Lots of fiddly holding of SMDs with the tip of my fingernails and tweezers. The two 39p caps at the transistor base were the hardest as they are very small indeed and so is the transistor.

Regards,

Alan

What a neat work! How much time did you spend to build the 1.5GHz Oscillator?

I tested its reverse leakage current at 12v by using microamp range of DMM, and it does not show any reading(less than 0.1uA. Also, forward voltage at 10uA is 200mV, at 1.8mA , it is 380mV.

I don't have other schottky diodes to be compared to this diode, as this is my first schottky diode. From the datasheet, all the specs are same as 1N5711, except the reverse breakdown voltage,is 60V , while 1N5711 is 70V. I think the diode is good at RF. What about you? Do you think it is good?If you can , please tell me your readings(as above) of your 1N5711s.

Thank You.

Bert,

You can test the diodes pretty much the same way as you would test a normal PN junction diode. They will conduct significant current only in one direction and their forward drop should be generally less than 500 mV. Your multimeter probably has a diode test range, you can try that.

Otherwise put a 1k resistor in series with them across a voltage source and measure their drop. You can confirm their forward drop is as the datasheet suggests and their reverse leakage is within spec at room temperature.

These particular diodes aren't ideal for RF detection, probably similar in performance to a 1N4148, but give them a go. Easiest way to test them as a detector is to use them in a peak-reading power meter circuit and apply some RF from a signal generator. Comparison with other diodes will let you determine their relative performance. With suitable construction you can provide a bias current too and fully characterise their performance.

The red-ish colour you see looking through the glass is probably the copper/glass seal of the encapsulation. Glass bonds well to copper so it is often used as a surface plating for glass/metal seals. Its expansion coefficient is not suitable in bulk form however, so it is normally plated onto a nickel/iron alloy with its coefficient adjusted to match that of the glass. Almost all leaded components use copper-clad ferrous alloy wire which is then tinned. Exactly how they assemble the diode without frying it is somewhat of a mystery to me, but even early point-contact Ge diodes were encapsulated in glass.

Regards,

Alan

Finally, got some 1N6263 schottky diodes. But it seems like it is rejected product as I found the diodes in very large quantities(about kgs) in a big plastic bag, and almost half of them are not marked, just can see the red copper oxide inside the diode. So, can you suggest me some ways to test schottky diodes?

Thanks.

Thanks for your advice.

Bert,

Yeah the MPSA18 is a very high gain device (at low collector currents). A BC550C is fairly similar. Really any old transistor can be used with some loss of sensitivity, or use a Darlington pair if you prefer. The frequency response doesn't matter, only the DC beta. The meter movement needs to be sensitive too, 250 uA FSD or better ideally.

You could use an op-amp and just about any mA meter, but the simplicity and low current drain is lost. The MPSA18 lets you forget about using a power switch, which is kinda useful.

Replace the 1M2 resistor with a 2M pot if you want to experiment with different transistors. They will likely need more bias current.

Regards,

Alan

At the first sight, I thought that the MPSA18 is a RF transistor, so I used a BF199 and I found it not working. Then, I found that MPSA18 is only a general purpose transistor...So, this means that I need a high gain transistor, right?

As for this circuit, is it that the most important/critical part of this circuit is the biasing?

Thank you.

Bert,

Yes you could use a 1N4148 instead of the 1N5711, but it will be less sensitive, especially at higher frequencies.

You might like to try other schottky small signal devices, like the BAT72 or similar BAT range devices (if you can get them)... Otherwise you could try a diode-connected JFET or BJT transistor.

Ideally the biasing dropper should be the same kind of device as the detector and biased to the same forward current, so their temperature coefficients are matched and will tend to compensate for thermal drift. In my experience it wasn't too critical, so I just used two 1N4148s. You could probably use a resistor network if you really wanted.

Nothing is very critical in the circuit, feel free to experiment. You can find 1N5711s on eBay, and worst case I can send you some. I got some from Kits and Parts and another lot from the eBay seller dpi4parts. The eBay seller also generally carries MPSA18s and a host of other goodies at reasonable prices.

Regards,

Alan

Can I use a 1N4148 in the place of 1N5711,and substitute one of the biasing 1N4148 with B-E junction of a transistor(to preserve the biasing voltage of the transistor)?1N5711 and schottky diodes are not available at places around me .

I would like to hear your opinions on this modification of the circuit. Thank you.

Kevin,

I've seen RF power meters that just amplify the input with some feedback amplifiers or MMICs before detection with a diode. That basically just shifts their range down the dBm scale by the gain of the amplifier.

Sounds like a reasonable approach, especially for a qualitative device (like a sniffer) rather than a quantitative one. That said it could be calibrated. Log detector chips are obviously a better solution, offering more than 70 dB of dynamic range over their bandwidth. The most common AD one tops out at 500 MHz, but there are devices available into the GHz I believe?

The "infinite impedance" detector is good for amplitude demodulation. Not sure about using it for level detection. Same with "zero-bias" MOSFETs and similar exotic "detectors". Haven't played with them really.

Another option might be a gain-control loop with either a voltage-controlled-gain amplifier or PIN attenuators trying to keep a constant detected signal level. The level control voltage becomes the output. Kinda like AGC line in a receiver driving the S-meter.

BTW, it isn't too hard to build your own log amp. The hard part is making them flat and precise. For a fixed frequency, say a 10 MHz IF for a SA or receiver you could build a bank of cascaded differential amplifiers out of discrete transistors. Lots of stages are required for a good dynamic range, but each is identical and doing it yourself does give you a bit more control over its noise bandwidth which could perhaps make it more appropriate for an SA log detector.

I once had a vague idea in mind to have each stage as it saturated turn on a LED, forming a bargraph. Lots of fiddly construction involved, but I think the idea in principle is valid. Probably easier to use a commercial log amp and linear bargraph driver like the LM3914.

Regards,

Alan

A another nice little oscillator, I wish my boards could look so tidy. In the search for ever more sensitive detectors I was wondering what are on you comments on using a rf transistor in place of the diode - I think they used to call them infinite impedance detector. Use the same technique to allmost forward bias the the transistor, and the high ft to get gain.