I have a small radio transmitter and I’m successfully generating tones for it to send data via PWM. The issue is that it is scattering the transmission on multiple channels and this is mentioned as a result of an impedance mismatch between the driver for the transmitter and the transmitter. The manual includes a sample circuit for matching up the impedance if the output impedance of the driving device is 2K - does anyone happen to know the output impedance for a Panda from the PWM pin? Does it differ for a Panda 2? I haven’t seen any info on this in any of the support downloads for the Panda.
The pins on panda are connected directly to the processor so maybe take a look at the processor datasheet? LPC2387
What kind of modulation are you using ? I.E. USB or FM ?
The problem might as well come because you are injecting square waves into a transmitter, which generates a lot of harmonics. You should consider interfacing the panda and the transmitter with an audio low pass filter, even a passive one 
Your input level might be too strong as well, staturating the first pre-amplifier stage of the modulator. Are you injecting your PWM into the high impedance microphone jack, or in a dedicated input jack with low impedance ?
Here’s the transmitter and user manual: [url]Argent Data Systems
It does also suggest a low pass filter, I was going to do that in addition to the impedance matching but figured it would be good to know the output impedance before I started in on that. The input is a dedicated modulation line, not a microphone line.
I looked at the LPC manual and didn’t see anything giving actual impedance values.
The impedance should match your dedicated modulation line.
Just add a passive low pass filter (serial res, cap to ground, serial res, cap to ground, serial caps high value for decoupling, then a variable res to adjust your level).
-R–R--C-
…|…|…|
…C…C…Pot----- Input TX Line
…|…|…|
--------------- GND
It’s important to adjust the level at the minimum acceptable value. Start from 0 (no modulation) and then increase slowly untill you get an acceptable modulation, but don’t exceed that value).
Good luck 
!
Ok, I’ll try that out and if it works, measure the resistance in the variable and replace it with a fixed resistor. I’ll also try out what the manual recommends in the off chance the output impedance is comparable to other devices that use this radio.
Just calculate the R/C filter values for a low pass of 2500Hz and you should be fine
For the last serial C before the Pot just use a higher value like 1uF
So I looked into the low pass filter stuff a bit and tried out just modulation in to single 100 Ohm resistor to modulation out to single 0.47uf capacitor to ground and that seems to confine the transmission a bit more. What would be the benefit of using your more complicated example and how could I measure if it had a benefit and by how much? The way I am “debugging” this issue is looking at the output in fldigi and seeing how close the transmitted signal is to the two tones in the waterfall display. Also not sure what RC values to use in your example since some are in series and some are in parallel. With the one I did I calculated it as 1/(2piR*C) which is around 3K Hz.