Flex 5000 - Its ability to produce positive peaks cleanly

[Steve - K4HX]

I would think it's just up-conversion. It wouldn't matter the mixer type, the modulation should not be affected.

[W1AEX]

Bruce - KF1Z   Very interesting results regarding the I/Q imbalance causing asymmetry, and your ability to restore symmetry by adjustments in both the software and the hardware. I have not run the calibration procedures for my rig, but it might be interesting to see what turns up in that area.

For better or worse, I made another short home movie of the Flex running a 10 watt carrier, as it was subjected to a triangle wave generated by the internal SG in Power SDR 2.0.16. It's 1.5 minutes long and best viewed at 480p full-screen:

http://www.youtube.com/watch?v=KemFreHZLJ0

I extracted several still shots from the video using VLC:

Shot 1 - The point where the signal just reaches 100% positive
Shot 2 - The point where the signal just reaches 100% negative
Shot 3 - The point where the signal reaches 200% positive with diamond wavelets

As in the sine wave version, audio compression and limiting was disabled, and the EQ was flat.

Having no training with interpretation of a triangle wave test, I will leave that to anyone with the skill to explain whether it's good or bad!

Rob

[K1JJ]

Rob,


The waveform looks impeckerable, of course. What audio freq did you use for the triangle? Though wid a Flex, it's doesn't really matter since it will pass anything DC to light.

Higher freqs, above say, 200 hz are easy to pass with good looking slopes for average AM transmitters. But as the freq drops down into the 100 hz and lower zone, it is VERY difficult for transformer-laden rigs. (phase shift and other non-linearities)

You probably remember the triangle tests I ran on my 4X1 rig a year ago? The waveform started to get sloped below 100hz due to the mod xfmr, even though the MOSFET driver was good.

However, a direct coupled class E rig, or any of the balanced modulator AM generated rigs (like the Flex or riceboxes) shud have no problem passing low triangle freqs below 50hz, depending upon design.

That really is a good test to separate the rigs which can pass lows well and those that can't. Try it on a stock Apache and you will see what I mean..

T

[KF1Z]

Try the test over again, with the triangle wave frequency above 7 khz.

The way the software hadles the carrier AND the tone, is after the carrier freq ( ~6khz) the phase of one of the audio channels from the soundcard advances 180 degrees.

Oddly, as the tone increases from 0 to ~6khz the ACTUALL tone being sent to the rig DECREASES, then after the turn around point ( ~6khz carrier freq) , starts to increase again.

This scheme with the tone handling, along with the 180 phase shift of one of the channels is how the software handles the AM carrier and modulation.

So I wouldn't be supprised if your modulation peak relationship (pos to neg ) levels change after ~6khz.

I didn't look at this on mine yet.....
It SHOULDN"T change, unless there is an imbalance through the mixer....

[W1AEX]

*IGNORE this since the DSP TX bandwidth was set too narrow.*

Tom and Bruce, thanks for the analysis. I used a 1000 cps tone for that test. Using the internal Power SDR signal generator I swept the Flex from 50 cps to 8000 cps and recorded it. Unfortunately, it may not be terribly useful since the scope display is not always clear, and I had to intervene and adjust the scope's sweep frequency a few times so it could lock. For greater entertainment value and so that it wouldn't be another silent movie experience, I hung a pair of headphones on the camera tripod so you can hear the tone as it incrementally slides up.

http://www.youtube.com/watch?v=VaoweQfWiTQ

The compilation of still shots below might be more revealing. It looks like things begin to fall apart at around 4k. I didn't see anything unusual happen between 6k and 8k Bruce, however, the amplitude seemed to fall off slightly at the upper end of the sweep.

Rob

[w1vtp]

Rob

Those shots look exactly as expected.  What BW did you have the transmit audio set to?  I'm guessing around 5 KC.  Without the higher frequencies (harmonics) all you get is the fundamental hence the sine wave appearence.

This is a whole different experience with AM  At this point we need to see displays in the frequency domain rather than the time domain.  I'm betting you do not have any issues with that (the classic "splatter" problem) at these levels but we won't know that until we see a spectrum analysis display.  You should be able to do that if you have the second receiver board I think that's doable. I've not done that so far.

Al

[W1AEX]

Those shots look exactly as expected.  What BW did you have the transmit audio set to?  I'm guessing around 5 KC.  Without the higher frequencies (harmonics) all you get is the fundamental hence the sine wave appearence.

Al,

Thanks for pointing that out! I thought I had all the operator variables removed but never thought to check the TX bandwidth. Unfortunately, as you guessed, it was set for 5kc (10kc AM profile) during the sweep. I'll open it up to 20kc (40 kc AM profile) to keep the DSP out of the picture and run it again. That should give a much cleaner picture of what's going on.

Rob

[W1AEX]

As Al suggested, increasing the TX bandwidth was necessary to see the square waves through the complete range. The sweep was set to run for 10 cps through 8000 cps. To completely remove the DSP action as a variable, the filter bandwidth was set for 20 kc, which would produce a 40 kc AM signal. Obviously, it was necessary to adjust the sweep of the oscilloscope as the generator moved up in frequency, but it catches synch a number of times through the range. The updated video is posted here:

http://www.youtube.com/watch?v=UpoCipk-U08

The compilation picture below shows various still shots from 10 cps through 9000 cps. The edges of the generated patterns start to look soft at around 6000 cps to my eyes.