Hi David 1) Put a 10 MHz tone into it and measure the SNR. Heavily undersampling means you usually lose alot of bits- maybe 12 bits sub-nyquist turns into 8 bits 5x super-nyquist . But with lots of oversampling, you iwll get some oversmapling advantage. However, a loss in SNR in the front end translates to more gain required in everything before the ADC, which impacts your potential radio performance.
10 MHz tone in at -1dBfs, get the data out and put that into octave..... get as many samples as you can hold in the RAM of the STM32. Then, build yourself up a two tone test generator- a couple of 10 MHz xtals 25-100kHz apart .. or something. whatever.... pad heavily to isolate each oscillator/source .... measure two tone IMD.... I'd suggest a 2nd conversion to 455kHz and undersample that.... 2) The IF and filters- this is always a thing that alot of time is spent on. and if you dont get the matching to the filter correct, the response will be awful. multiple filter sets are usually required- say 6 to 8 poles.....of xtal filter over the entire IF chain are usually used. each 2 pole set might get you 40dB at 1 MHz. Filters also have multiple modes and responses pop up causing issues, so overall LC roofing is usualy required, or a coupel of different filter types. consider that you are trying to hear a signal -120dBm, and there is a nearby signal say 10km away that is -30dBm.... - there is a 90dB requirement.... The second IF at 455 k solves a few problems. Certainly makes it easier to meet off channel, and allows for more gain. The second IF using a NE602 is reasonable. glen On 26/02/2015 11:51 AM, David Rowe wrote: > Hi Glen, > > 1/ What characteristics should I look for to determine if the S/H is > good (or bad)? Is there test I should try, like perhaps two signals at > once in the ADC passband? > > The S/H RC time constant has a 3dB point of 7MHz. > > I'm not expecting sparkling performance, but would like to experiment a > bit more to see if it's a usable as a platform for our VHF FreeDV > experiments. > > 2/ Thanks you for the comments on the blog. yes I enjoy explaining > things, and the blog is an outlet for the teacher in me :-) Learning > and helping others understand is something about open source that I like > and appreciate. > > 3/ Thanks for the IF filter spec. Yep, was going to push for 60dB on > the next pass. It's those +/- 1MHz points that are critical, as they > alias on top of my wanted signals. A few kHz either side and the DSP > will nail it. So its likely to have better +/- 5kHz than +/- 1 MHz > performance. But I guess every rx has its weak spots where images or > some equivalent pop up. > > Is it possible to get an xtal filter that has 100kHz 3dB BW (or whatever > we need for 2 channel diversity) and is > 60dB down at +/- 1 MHz? > > The other option is a few more coupled LC sections, they seem to offer > about 20dB extra attn per section. Not sure if there is a more elegant way. > ------------------------------------------------------------------------------ Dive into the World of Parallel Programming The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ _______________________________________________ Freetel-codec2 mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/freetel-codec2
