Talk:ET-STM32 Stamp

A/D Noise
Has anyone toyed with the A/D conversion yet? The ET-ARM stamp apparently had some grounding issues that caused the A/D converter to pick up a fair bit of noise. I'd be interested to know if the situation is better on this stamp. Chris Russell 05:37, 8 May 2010 (UTC)


 * I have done some tests, at first look it seems a lot cleaner at 12 bit than the ET-ARM stamp did at 10 bit. I am not sure where the true limit is, it might be better than 12 bit by averaging several samples by using the built in white noise generator and oversmpling (while grounding all nearby pins). At the moment we are making the most advanced microcontroller oscilloscope in the universe http://www.sciencezero.org/index.php?title=STM32_Oscilloscope --Bjørn Bæverfjord 18:34, 14 May 2010 (UTC)

I've been pondering for a while, creating a device that will capture large amounts of data from an antenna for later analysis. Since so many urban environments are heavily saturated with RF noise, those who have an interest in VLF/LF listening often must seek out an RF-quiet environment from which to listen. Unfortunately, this has drawbacks. You can only stay in an RF-quiet environment for so long, because the area will generally be rugged, isolated, and primitive. Computers brought along for the purposes of digital signal processing will introduce RF noise to the environment, and can be too bulky to bring some places. And finally, you can generally only chase down one signal at a time. A computer may be able to record a wider slice of bandwidth for later analysis, but you're still going to be missing quite a bit.

I have an extremely high quality broadband amplifier, from 12kHz-2MHz, already working well with extremely good characteristics. What I'd like to do is sample the antenna directly, at a rate of at least 1MHz (2MHz is a nice option to have, but 24Mbps is faster than the SPI interface can handle, and there's a possibility of phase errors), and dump all that raw data to a storage device (probably SD, via SPI). That way, I can create a small, portable, battery-powered device that can easily be transported to an RF-quiet environment, grab a bunch of data, and then pour over and process the data at a later time to see what signals have been captured. Six hours of data (three at 2MHz) could fit onto a 32GB SD card, and would probably provide enough data for many long nights worth of analysis. Smaller capacity cards could be used to record at regular intervals, such as during grey-line propagation.

Unfortunately, on second thought, it's going to be pretty important to extend coverage to at least 515kHz, as activity in the 600 meter band is increasing as of late. Looks like I may need to use an external ADC after all. Chris Russell 23:03, 17 May 2010 (UTC)


 * This is 10 bit (showing the least significant 8 bit) resolution sampled at 2 MHz, there are some large blips but they seem to be at the same place in the waveform each time they appear so it might be triggered by the code since it is in sync with my trigger algorithm. The rest of the noise depends a lot on the source, high impedance makes a lot of noise. The STM32 is fast enough for realtime lossless compression if the signal is suitable.--Bjørn Bæverfjord 23:48, 18 May 2010 (UTC)
 * The large blip was caused by the debug code switching a pin at maximum slew rate so it is nothing to worry about. The SDM32 has a built in SD memory card interface that supports DMA up to 48 MB/s so is very suitable for advanced data logging applications. One interesting application would be to make it look at the spectrum and store 15 minutes worth of data each time it detects a change. That way it could stay somewhere for weeks and will be able to catch rare signals that is unlikely to be found by grabbing blocks of a few hours at a time. --Bjørn Bæverfjord 06:54, 20 May 2010 (UTC)