2
$\begingroup$

I am looking for a simple schematic for building an effective RF detector which can trigger a GPIO. The goal is to (passively?) detect a 10 mW RF signal (10% duty cycle) and have that raise voltage on a pin to somewhere north of 0.5V within 2 seconds of the signal going live.

I have some Avago/Broadband/Agilent/HP HSMS-2855 chips coming my way, which I understand are wonderfully good zero-bias RF detectors. However, it's not clear to me whether I can slap a random antenna on and go have some fun, or if I still need to pay careful attention to impedance matching, filtering, etc...

I have read An Ultra-Low Power Wake-Up Receiver for Real-time constrained Wireless Sensor Networks, but the pictured board shows a complex PCB with a high part count and large footprint.

Radio-Triggered Wake-Up for Wireless Sensor Networks is wonderfully well written and gives great hand-holding in case the designer wants to add range. However, Figure 3, where it gives a simple RF detector schematic, is confusing to the layman because there's no ground drawn to the antenna [Update: see comment section below]. Here's Figure 3 from that paper:

Figure 3 from the second reference, showing a very simple radio detector schematic

Design Optimization and Implementation for RF Energy Harvesting Circuits (related spinoff papers 1 and 2) is another very well written resource, diving into fascinating detail on practical optimality choices, but it does not present the final tested schematic.

I'm hoping this is the RF detector equivalent of a paperweight, that is to say there's no need for refinement, it doesn't matter if it's ugly, and there's almost no way to do it wrong.

Appendix

Typically, one would want an OOK stage in order to disambiguate between other signals which might trigger a false positive, but because in my application the nearest possible source of EM radiation is hundreds of meters away, and because the consequences of a false positive-- even if it's tens of times a day-- are negligible there's much less need for complexity. KISS, price, and compactness are higher priorities.

Improve this question
$\endgroup$
4
  • 1
    $\begingroup$ Figure 3 in your second article looks fine to me, and it clearly has a ground. Very little current will flow, assuming that $V_{out}$ is being measured by a circuit with high input impedance, but a voltage will be developed and that's what you care about. $\endgroup$ – rclocher3 Sep 23 '20 at 15:39
  • $\begingroup$ @rclocher3 thanks for the comment. Could you help me understand how is that ground connected to the antenna? In the drawing, those seem like two different circuits, which are electrically distinct. I can easily imagine the antenna ground plane being grounded to the capacitor's ground, but that's not explicit and I don't know that guessing will get me very far. $\endgroup$ – Kenn Sebesta Sep 23 '20 at 18:33
  • 1
    $\begingroup$ You make an excellent point. Not showing a connection to the antenna's ground plane is a common radio schematic convention since the earliest days of radio; it's just assumed that the operator knows to include one. Many old receivers have a single-conductor plug for a wire antenna, just like the schematic shows, and a separate ground lug on the chassis for the connection to the antenna's ground plane, which was often a wire with the other end wrapped around a metal cold-water pipe. Sorry, I should have explained that convention earlier, and I completely understand your confusion now. $\endgroup$ – rclocher3 Sep 23 '20 at 18:49
  • 2
    $\begingroup$ I took the liberty of editing your question Kenn, to insert a copy of the schematic of Figure 3. I hope you don't mind! If you do mind, please of course feel free to revert my edit, or make further edits. $\endgroup$ – rclocher3 Sep 23 '20 at 19:03
2
$\begingroup$

Well, your GPIO will need a defined voltage level and a couple of µA of drive strength, so completely "passive" is probably not really possible.

Now, you'll need the following:

  1. Antenna
    1. potentially, amplification here
  2. Filtering
    1. potentially, amplification here
  3. Rectification
  4. Thresholding
  5. GPIO activation

For antenna choices, see Texas Instrument's antenna selection guide.

  1. Will probably simply be a stumpy coil antenna for 433 MHz, and might be PCB antenna or even a chip antenna for 900 MHz.
  2. Antenna -> capacitor -> LNA (example, ca 10 ct) -> Filter (example, ca 40ct)
  3. RF Schottky Diode (example) -> RC low-pass filter with cutoff above 1/(minimum pulse length)
  4. opamp in comparator configuration (DON'T use any opamp before 1984): connecting the non-inverting input to a virtual ground defined by a trimmer between VCC and ground
  5. (optional) buffering of opamp output with MOSFET or BJT

Important: learn to use SPICE so that you can simulate the elements of your detector. It's time well-invested, as "KISS" means "don't try to measure something for hours that simply doesn't work, which you could have seen in simulation".

Your RF detectors can take jobs 3 to 5, but you'll still have to make the things frequency selective, so there's little way around 2.

Improve this answer
$\endgroup$
4
  • $\begingroup$ "DON'T use any opamp before 1984" Lol, that's so specific! 1) What happened to op-amps in 1984 and 2) is it a frequent concern to be on the look out for 40 year old components? I agree that the system I'll need a few µA to drive circuits. Having read through the linked paper, though, it doesn't seem unreasonable to be able to harvest enough energy from the EMF to drive 0.5V at 3m. $\endgroup$ – Kenn Sebesta Sep 23 '20 at 17:25
  • $\begingroup$ That TI antenna board is awesome! Regarding SPICE, correct me if I'm wrong, but I feel like filtering is useful for reducing the number of false positive events and won't improve the detection of true positives. In my specific case, I believe that false positives will have negligible impact (I love that balun's price, though!). I'm hoping that KISS gets me to a simple circuit I can do on a breadboard without meticulous attention to trace and ground plane layouts. $\endgroup$ – Kenn Sebesta Sep 23 '20 at 17:36
  • 2
    $\begingroup$ @KennSebesta the number one "my opamp circuit doesn't work" reason on electronics.stackexchange.com is "yeah, you're using a µA741, that's the worst, most obsolete 1968 opamp you can get, and you're using it as if it was an ideal opamp". 1984 just because that means that someone actually picked that specific opamp instead of copying a design from a website that copied it from a website that copied it from a magazine that copied it from a magazine that copied it from a magazine that copied it from a 1970's article, where someone only had maybe 6 opamps to choose from. $\endgroup$ – Marcus Müller Sep 23 '20 at 18:37
  • 2
    $\begingroup$ Re filtering: you're not right – if you don't restrict what you're sensitive to, you won't have "a couple false positives"; you become sensitive to basically everything, including low-power electronics and TV stations, which might simply have more power at your receiver than your intended 10 mW transmitter. $\endgroup$ – Marcus Müller Sep 23 '20 at 18:39
2
$\begingroup$

I'm going to disagree with Marcus and say that you can probably get this to work, for some relaxed definition of "work".

The detector you've drawn has poor selectivity, meaning it will detect quite a lot of stuff besides the intended signal. The inductor and capacitor provide some amount of filtering, but not nearly as good as you'd get in any basic receiver.

Aside: the better that filter is, the more sensitive to temperature variation it will be. Also, at 900 MHz you'll have to give some thought to the circuit layout, since at this frequency parasitics can be pretty significant.

On the other hand, path loss at 900 MHz and 5 meters is only 45.5 dB. Assuming 0 dBi of gain on both antennas that means you're looking at a received power around -36 dBm. That's pretty strong, even feasibly the strongest thing around if there are no other electronics nearby. This is good news because it means poor selectivity of the detector is not much of a problem. If you're near any WiFi radios, cell towers, broadcast stations, or even very near any digital electronics, you'll probably have trouble.

The one area you'll probably need to improve is amplification of Vout. Although -35.5 dBm is fairly strong for a received RF signal, it's probably not going to be enough to reach the GPIO's threshold. Addressing that is a simple matter of adding a comparator, op-amp, or even a simple transistor.

If you want a step up in sophistication, there are integrated RF transmitters and receivers that are really cheap. Besides helping with a lot or all of the amplification and filtering, they also provide digital modulation and demodulation.

Or if you want the satisfaction of building something yourself, it wouldn't be too complex to make a mixer, which would convert the 433 or 900 MHz down to a frequency low enough that it could be sampled by the ADC on a microcontroller. Then you can do some very simple demodulation to confirm the signal is from your device and not noise. A good receiver will include some filtering to reject image frequencies and maximize dynamic range, etc., but if your performance requirements are low then you can skip all that stuff and simplify the circuit substantially.

Improve this answer
$\endgroup$
5
  • $\begingroup$ Oh, good point with the mixer and SDR in the end; you don't need great selectivity if you're just out there waiting for the right sequence to wash through your correlator! +1. $\endgroup$ – Marcus Müller Sep 24 '20 at 13:01
  • $\begingroup$ Thanks, this is great info. So if I understand correctly, I can make an LC tank tuned for my frequency (probably 433 to avoid layout sensitivity), and then give it a go. Since this is for an outdoor receiver, which will see temps from -10 all the way up to 40 degrees, temperature sensitivity would cause hard-to-debug issues. Out of curiosity, what ultra-low-power TX and RX exist for this? If I can find a COTS solution, that would be a million times better than working off a schematic. $\endgroup$ – Kenn Sebesta Sep 25 '20 at 15:27
  • $\begingroup$ @KennSebesta Well I don't know what your application is, but there are super cheap RF modules for under $7. Not sure that particular one qualifies as ultra low power. Bluetooth Low Energy does though. Zigbee is another thing to investigate. $\endgroup$ – Phil Frost - W8II Sep 25 '20 at 16:07
  • $\begingroup$ @KennSebesta If you really want to optimize for cost, you probably want to go the route of a microcontroller with an embedded RF stack. For example the ESP8266 is available in single quantities under $8 and has a whole TCP and WiFi stack in it, and comes with the antenna and everything. Those are popular with hobbyists: if you want to make a thing in quantity it can be much cheaper but you'll have to do some RF engineering and assembly. $\endgroup$ – Phil Frost - W8II Sep 25 '20 at 16:12
  • $\begingroup$ @PhilFrost-W8II, BLE and Zigbee are both, unfortunately, about 2-3 orders of magnitude higher energy consumption for the asynchronous system I'd like to build. I've got bunches of ESP8266, they're the best! Alas, their lowest sleep state with an active RX stack is in the 5-10mW range. This will be generally true for all complex digital comms. The only exception I know of is OOK. $\endgroup$ – Kenn Sebesta Sep 28 '20 at 14:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.