RF monitor for remote CW analysis

K9SO

I wanted to share a simple little gadget I came up with so I could quantitatively evaluate my Flex remote CW at the actual RF output level. My remote is 800 miles away so evaluating the quality of my actual CW RF output would give me the ability to evaluate the quality of my signal. Listening to my signal wasn't enough.

1. Any missing dits in a long string (50 dits)?
2. Any merged characters?
3. Any clipped or elongated dits due to Internet connection jitter?
4. Is the power output constant?
5. Do I have "first dit distortion" before the amplifier comes up?
6. Can I get an oscilloscope-like view of my actual signal?
7. Any power slump or RF power overshoot?

The hardware to do this couldn't be simpler. Everything is available on Amazon and totals only about $35

Per spec, the AD8310 will work from .1 to 400MHz and can handle up to +17 dBm of input power. This requires an RF sampler or attenuator for high power, but for my initial tests I used the FlexRadio's transverter output set to 0 dBm. While not sampling my actual RF signal yet, it is enough to see if I had any distortions due to Internet jitter, etc.

I needed WIFI connectivity so I could view results over my network and an internal web server that could display results w/o loading any apps on my PC or phone. So I chose the Arduino Nano ESP32 as the controller.

My initial results are encouraging. In fact, it clearly demonstrates the incredibly perfect RF waveform of my 6600's CW waveform. Perfectly shaped rise and fall times, no first pulse distortion, and no truncation when using full break-in.

This is the screen capture of the website's summary page. This shows the basic results from sending a number '5' at 20 WPM. Importantly it shows the maximum difference between the pulse lengths in the string. It also shows the difference between the first pulse width and the average of the rest. In a series of 50 dits, this would immediately flag any issues.

In the "advanced" page, I've captured the actual RF waveform of the 4th dit in an oscilloscope view. In the table, it displays a more detailed look at each dit:

I took the timing measurements at the 1/2 power points of the waveform. This is a logarithmic voltage display so it doesn't immediately look like that. The ripple in the 'space' is wideband noise over -100dB down.

That waveform was with a 20WPM setting and a 1mS full break-in TX delay. I've tested this from 5 WPM to 60 WPM and from 1MHz to 60 MHz.

Tests with a function generator show these results are very close to reality. This is a 10MHz signal amplitude modulated with a square wave that had 1mS rise/fall times. Modulation simulates 20 wpm. The display is of the third pulse in a string of 27 sent. Each pulse is evaluated in the table.

To show the level of detail the SW is capable of displaying, I set the TX delay in the radio to 10mS and again used full break-in:

A small ~10mS "shelf" appears at the beginning of each dit. This looks terrible, but it's just the transmitter's wideband noise when it operates PTT in anticipation of the actual dit. Again, the scale is logarithmic so this noise is actually -40 dB down, wideband, and completely harmless. The detector is wideband so it picks up all the transmitter background noise.

For high power sampling, I used a BIRD power attenuator (-30dB) and an additional 20 dB attenuator at the output for a total of 50 dB attenuation. That gives me a 0 dBm signal at 100w and +12 dBm at full power. A calibrated attenuator can turn this into a pulse-measuring wattmeter. An in-line sampler can be used as well if precautions are taken to stay within the detector's power limits. I've tested the sensitivity from -20 dBm to +10 dBm.

Call for testers:

I'd appreciate any comments and if you'd like to help out in the SW evaluation and Beta testing, send an email to my QRZ address.

Thanks for listening,

Fred

K9SO

K9SO

I added a linear view of the pulse to the program. I think this is more intuitive. It doesn't change the data, but it displays it in a way that is more expected (linear vertical with power). It also makes the 50% power measuring point more obvious. It loses the details the log view gives at the very low levels (the noise shelf goes away) so I left in the option to view it either way.

Linear CW pulse view of a FlexRadio:

Logarithmic pulse view:

No one has posted any comments here so I conclude there's not much interest amongst Flex users. I can understand why: I can now see that the Flex RF waveforms are nearly perfect anyway so what I'm showing here is only what we already knew. But at least this will give me a quantitative way to test my end to end paddle CW over a 1000km path. With lesser radios it will display first dit distortion, RF overshoot and RF rise/fall pulse shaping too. It can tell me if my amplifier sequencing is clipping my first dit or syllable. For FT-8 it can measure linearity and power slumping.

In any event, it can be a very inexpensive station signal monitor.

I'll be posting the source code on my QRZ page once it stabilizes.

73, Fred

K9SO.

Mike-VA3MW

Nice work Fred! The cool geek stuff takes a few more 'overs' before it gets traction it seems. :)

N5NHJ

Great work, Fred!

I can't wait to test your system.