Measure signal strength

I'd like to measure signal strength across a wide frequency range, from LF to 3-4GHz or so. Is there an ultra wideband tuner IC, with RSSI output, that I can use to do the job? I could combine a few if required, to cover the desired frequency range. If that's not available, what kind of circuitry would I need, to tune a standard RF Power Detector (i.e., something like a AD8313) to the frequency that I'd like to measure, before reading its RSSI output? Either way, an MCU would step the tuner across the desired range, and sample the RSSI output with its internal ADCs. Thanks

Your specification is incomplete. You also have to think about dynamic range, selectivity respectively bandwidth and scan speed.

Generally speaking, an industry standard spectrum analyzer does what you want.

I don't need to visually inspect the entire spectrum at once, I just need to log it in 2 columns: "frequency" / "strength". Then I will just average out values at every pass. This is part of research on effects of radiation on human health by the way, volunteers will carry the logger with them for a few weeks, 24/7. It has to be really portable and powered by a button-sized battery, 3.7v 300mAh. I'm working with a colleague who is a Ph.D. in antenna design, and we'll have access to a calibration facility to account for any compensations required in the MCU software. However we are struggling with the circuitry at the moment as discussed above.

Dynamic Range: we are just trying to walk around cities and towns around the globe, and measure the level of exposure that people and animals are subjected to. How would you define that range?

Selectivity: low precision, we are not trying to "tune in" to any particular frequency for the purpose of using the transmitted information. We mostly care to be able to identify the RF source by inference to known frequencies, for example, cellphone, microwave oven, radio station, medical equipment, electricity transmission lines, etc.

Speed: nothing fancy, especially given our current draw restrictions. If we could sample the entire range within 10-15sec or so, that would be good enough.

If you had the experience to also put figures in the descriptive specification above, that would also help. We don't have very much experience in RF electronics in the team, so we are a little lost.

Many thanks

You need a mutiple frequency range tuners with a relatively narrowband IF frequency.TV and TV-SAT Tuners can cover up to 2200MHz but IF bandwidths are very variable from 6MHz to 30MHz.
Therefore using these tuners wouldn't be very healty and they cannot be used below 30MHz ( You have mentioned LF and I understand few hunded kHz )
So, you have to design your own tuner that will meet these specifications.It would be very complicated...

How about using a PLL with a mixer, to tune in? Would that work? How would I go about designing it?

Yep, it's possible.
A Superheterodyne structure will work but the operating bandwidth is very large.That's why a simple Mix+Osc configuration wil not be sufficient.
-First, a very wideband ( nearly DC to 4GHz) LNA should be used ( design is very tough but possible ! There are also some products from Minicircuits )
-A verywideband Mixer can be used ( err.. There are some diode or passive FET solutions, it's OK let say mostly possible )
-Local oscillator can not cover this bandwidth.That's the problem...
And all of them won't be low power design.
Another solution for Local Osciallator is to use a Fractional PLL+VCO from Hittite or Maxim
http://www.maximintegrated.com/en/pr...f/MAX2871.html
http://www.analog.com/en/rfif-compon...s/product.html
As you see, Min. synthesizable frequency is starting from 25-30MHz, that's not proper for you...



Another solution is to use a programmable divider with a fixed oscillator circuit in order to desired mixing signal.
http://www.centellax.com/sites/centellax.com/files/UXN14M32K.pdf
For instance, you design a 4GHz fixed ( by another PLL) oscillator and then you divide this signal to mix-down the input signal.Programmation ratio will be 1 and 4,294,967,295 so min. frequency
1Hz and max. frequency will be 4GHz.It seems pretty brillant.
So, theoritically it's possible to down/up convert the signals between 1Hz to 4Ghz to a fixed IF frequency.This IF frequency must be lower than the min. input freqeuncy.
For instance, if the min. input signal is 50kHz, the IF frequency may be 20kHz ( NZIF ).Then this IF signal can be processed by a seperate RSSI circuit to obtain signal strength in dBm.There will also be very carefully designed calibration routines...
It's a typical frequency synthesized signal generator design...

Look at spectrum analyzer architectures, because that is what you are trying to build: a simple spectrum analyzer.

What sort of aerial(s) are you going to use? You will not get an aerial to cover an octave without a significant drop of at its band edges. You could get away with some sort of auto tune/ auto compensator, to keep the outputs from the aerials relatively flat as you sweep the frequency. The problem is switching from aerial to aerial as your sweep progresses. There is also the problem that for frequencies less then 100 MHZ, the field strength is measured in millivolts/meter, while above this is mW/cm^2.
Frank

Post #3 gives a rough idea of the intended range and bandwidth. I think it would be reasonable to answer the question (at least preliminarily) before starting to design a new or chose an existing hardware.

There's a large number of spectrum analyzers and remote controllable wide band receivers that can be used for the project, commercial instruments as well as DIY designs. There's however a major problem about power supply.

The power consumption imagined in post #3 (about 0.5 mA for 4 weeks battery lifetime) is simply unrealistic for continuous receiver operation.

I think the goal is to create something like this, just with a wider frequency range and more advanced data logging:
http://www.fieldsense.com/media/EA-FS-061-D_D.pdf

And a frequency scanning feature (frequency selective field strength monitoring) as far as I understand.

My bad, I forgot to mention that the battery will be Li-ion and that the user will be able to recharge overnight. Sorry about that guys. The target operating duration is 18-24h. The operation will not be continuous, log cycles can be a few minutes apart. We are also planning other optimizations to save power. For example, during the first scan, the entire range is sampled. Let's assume that at the location of the user, the logger found signal only at frequencies X, Y, Z. After a few minutes, another scan cycle begins, but this time only the frequencies identified before are sampled. After a few scan cycles, a full range scan will occur again, to identify potential new sources, or in case the location of the user has now changed significantly. We anticipate that this strategy will result in material power savings?

I'm looking at Apple's iWatch, which allegedly is power by a 400mAh cell. It's got a fast CPU that runs a proper OS, has GPU, power hungry display, tons of sensors, I/O, and general functionality, and it lasts 24h between charges. How?



I should also mention that if it turns out that covering the frequency range originally specified proves impossible to implement in practice, whether due to antenna design limitations, resulting size of antenna or circuitry, or power draw, then we could consider raising the frequency floor to 25-30Mhz or so.