teensy 3.1, Radar proximity/speed sensor readout

Status
Not open for further replies.

eduardo

Active member
Hi.
Does anybody have experience connecting a radar proximity sensor to teensy 3.1?

I am thinking about the Hygrosens RSM-3650. (www.voelkner.de/products/143210/Radar-Bewegungsmelder-Rsm-3650.html)
The sensor is based on the doppler principle and produces an AC signal from tens to hundreds of Hz.
The interesting part is that it actually gives information on the speed/movement of an object and not on the distance. However the amplitude of the signal varies greatly with distance and size of the object.

My idea is using the internal 16 bit ADC and the audio lib with an FFT.
The signal would be AC coupled (High pass ~20Hz). Eventually the internal PGA could be dynamically adapted to the varying signal amplitude. Output is an LCD and several LEDs.

Any input regarding the use of the audio lib is apreciated.
The project is "just for fun/demo" at the moment and I will not start before July.
Regards.
Eduardo
 
Last edited:
That sounds like a fun project. I actually did something very similar (a doppler radar) for one of my projects while I was getting my degree.

I don't have much experience with the audio library, so I can't really help you there.

My only advice is to find a good low-noise amplifier that can give you the kind of gain you'll need without compromising the low output signal levels of that device. The datasheet on the sensor you linked to mentions that the maximum output signal level is only 300uV, so you'll be looking for a gain of up to 11,000 V/V (~80dB) to get a full scale input on the ADC using a 3.3V reference, or a gain of 4000 V/V (72dB) if you use the 1.2V reference.

Also make sure that you band-limit the amplifier circuit (use a low pass filter type configuration) to minimize the noise that gets amplified.

Good luck!
 
You are right about the gain. Good point.:)
However noise is only a problem in combination with bandwidth. BW is <1KHz.;)
An OpAmp with 3nV/SQRT(HZ) gives 90nVolts equivalent at the input. That should be ok. Care must be taken with impedances keeping them at the lower end.
I suggest using 1.2Volts as ADC reference, differential input, some gain inside the µC-PGA and 30..40dB with a cheap low noise Opamp from Microchip or so.
 
Sounds like you have everything covered :)

Those sensors you linked to have a pretty impressive level of integration. Might have to look into using them in my own projects :)
 
Hi Whollender.
I made a quick measurement. Output waveform moving one hand away ~20cm distance. Signal is 300mV (Millivolts) maximum moving my hand very close to the sensor.
Newfile1.png
Detecting objects at 2..10 meters will give < 1 mVolts probably.
For Opamps I found some better types at Analog Devices. The MCP parts have strong noise at very low freq.
Is there any info on your degree project with doppler?
 
Last edited:
Really interesting;)! I would consider some opamp from Linear Technology together with the excellent from Analog Devices, I've got a couple of samples with extremely low DC offset and ultra stable amp transfer in relation power supply, I've used to drive a 1.2v very high speed adc for video purposes.
But again thanks to bring here project like this!
 
Hey Eduardo,

I went through some of my backups last night, but I wasn't able to find any good overall documentation. I did find the schematic, but it's in PADS format and I had to download the free viewer for it to get a look at it.

Here's a quick overview of the stages of the IF section (the section after the mixer): The first stage was a Sallen-Key type two pole active low pass filter. The second stage used an SA604 FM IF chip for it's high gain limiting amplifier stages (92dB of gain, with output limited to a few hundred millivolts). The third stage amplified the output of the SA604 from a few hundred millivolts into a 1-2 volt signal that was output to a 3.5mm audio jack.

The second stage is really the central part of the circuit. The SA604 (http://www.nxp.com/products/rf/fm_if_system_ics/SA604AD.html) is a great chip for this kind of application because of it's high gain, and the limiting gives you a nice square wave output with a consistent amplitude for most input amplitudes. You just have to be careful, because it has very high gain and bandwidth, so a good layout is key to keep it from oscillating (dead-bugging it would also probably work well, but normal prototyping techniques probably won't work at all).

If I remember correctly, the first stage active low pass filter actually added more noise than it was worth, so I would suggest using a simple RC lowpass in front of an SA604, then to an ompamp to get a FS signal for the ADC. I've attached the SA604 part of the schematic for reference.

SA604.jpg

If you do go with the SA604, there are a couple things you'll want to watch out for:
-> The SA604 is spec'd with a 6V power supply. Based on the datasheet, it'll work down to 4.5V, but you won't want to take the 5V directly off of USB power as that will be really noisy and probably screw up your results. A simple RC lowpass filter off the 5V rail would probably work (high C with low R to prevent too much voltage drop).
-> I had some issues with the SA604 oscillating under certain circumstances. All the electrolytic decoupling caps (C32,33,34,36, and 37) should either be replaced by low ESR chip caps (0.1-1uF) or you could put the chip caps in parallel with the electrolytics for very wideband decoupling.

With the kind of output signal that you'll get (mostly a FS square wave) you may be able to use one of the frequency counting libraries instead of using FFTs. If you go that route, you can change the opamp stage after the SA604 to a 3.3V comparator to get good logic level outputs.
 
Hi whollender.
My late reply. I posponed the project for many reasons.Here is my update.
I found several examples for simple doppler speed measuring with Arduino. A high gain amp and a bandpass is used most of the times. SA604 is probably a very good choice because of the inherent limiting/gain control. It is fairly expensive though. Frequency/pulse measuring gives the speed for a single dominant object. But no information on direction: approaching or moving away.

Last week a guy asked me for a redesign of his ball speed metering system. This is when I learned about radar modules with IQ output. Phase := info on direction. For good filtering a complex fft is necessary followed by a couple of filters to find the wanted object, a soccer ball in the middle of noise caused by the players.
I am gathering technical info at the moment.
I need a two channel synchronous sampling ADC and a complex FFT to start with. Not sure if ADC and Audio libraries offer this. Specially the two sync channels of the internal ADC.
If anybody wants links to examples using simple freq/pulse approach please ask me.
 
Last edited:
Status
Not open for further replies.
Back
Top