1. Home
  2. Electronics Design
  3. Posting for a friend, WAV file question

Posting for a friend, WAV file question

Ok, a friend of mine recorded some 1970s FM telemetry data onto a digital tape and imported it as wave.

Its 9.0 to 19.6 Khz sorta a hybrid between a triangle wave and a sine, It started out as a triangle, but the old half inch tape deck did some rounding. Chip that made the waveforms and regenerated them is now known as TC9400 by microchip, which is a current source charging a cap..

So about 2-3 cycles on the tape represents one point on the final data waveform, which was roughly dc to 3 khz bandwidth.... He has a hardware decoder,but would like to do it in software... From what I've seen on the scope and a preliminary piece of code that plots the wave file on screen, he has little noise, but a lot of amplitude variation. There is a bit of noise around the zero crossings, so he is weary of looking at crossings.

So what is the standard way in software of picking a for measurement on a not so clean waveform?

I would just square it and clip it, but he is not so happy with that idea, he is worried about loosing resolution in the vector. analog, image (actually calligraphic writing of cad figures) recorded on the tape, or jitter from image to image. Oh, and he's got 40 hours of images he wants to digitize.. recorded as 44.1 khz wave..

Steve

Reply to
osr
Loading thread data ...

Cycle-by-cycle best fit to a sine?

Low-pass filter, seriously upsample, square and clip may work well.

If the lines are fairly straight going through zero, then detect the general area of each zero crossing, then do a least-squares fit for offset and slope, then solve for the zero crossing. You'll find that if you choose a constant span for your fit the coefficients will be static as a consequence of the even sample rate. This will give you plenty of resolution, and will have an inherent low-pass filter effect.

Or do some sort of AGC, followed by a discriminator. Squaring the thing up without interpolation and upsampling will lose you resolution, but AGC won't.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com
Reply to
Tim Wescott

You may also consider posting this to comp.dsp -- you'll find lots of folks there used to processing digital data.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com
Reply to
Tim Wescott

If I understood the problem correctly, the task is demodulate FM signal offline, to maximum accuracy? Since it was digitized at 44.1kHz, it was brickwall lowpassed at ~20kHz, so the transients would be distorted. I'd use some sort of discriminator rather then estimating zero crossings as the discriminator looks at entire waveform.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

formatting link

Reply to
Vladimir Vassilevsky

=46rom what i remember a lot of those old systems were analog FM. The = final=20 frame rates may be as low as 5 minutes per so that the signal would fit = in=20 a reasonable audio channel.

Reply to
JosephKK

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.