Friday, March 20, 2009


My friend K3RR inquired how is it that I can speak so confidently when I talk about measuring the behavior of my radio and its filters and MDS etc. Since I bought this radio I have grown used to thinking in dBm and not S units. I still write about S units because that's how most hams think.

The reason I can speak so confidently is because the radio calibrates itself across the spectrum.

There was a recent article that looked at the speech processor in the K3. I marveled at all the equipment needed to analyze that processor. Here is a flow diagram

Sheesh that tens of thousands of bux worth of hardware!!! To analyze your radio much less tune it up!! This was how I always did it before. I would put the rig on the bench and spend half the night trying to null the carrier out of my SSB radio, or trying to calibrate my S meter to 50mv.

Here is what I use today:

and this

and this:

This is Power SDR in scope mode and the recording console. I simply speak into the mic and make a recording as I adjust the various controls.

Here are some OLA's with no processing"

and here are some OLA's with compander at 1

and here are some OLA's with DX at 3

I can make a single recording as I OLA through the various settings, and know immediately what sounds good and what sounds like crap errr contest audio.

3x3 OLA's

Before you even click the audio you can tell DX is going to sound wayyyy processed. This is far more useful than the 50 grand of equipment. If I want to adjust my EQ I use this method, put a nail in it and go work some DX. I may spend a few minutes fooling around with this but no half the night.

But how does that answer Joe's question?

It turns out the F5K has a full time virtually dedicated signal strength generator built right in. The DDS (direct digital synthesizer) can give up one of its channels during calibration. The DDS output is constant across the entire fleet of F5K radios (+- 0.1 dB) so you have a virtually constant well known and well characterized super clean signal source in every F5K (and 3K).

What this means is you can change the software for an entire fleet of radios and NOT have to recalibrate all the time. That my friend is sexy beyond belief if you think about it. Once you have a known signal, the rest is all math. 10dB = ten times increase in signal etc. In addition the signal can be picked off anywhere along the signal chain for analysis, like before the filter or after the filter. The S meter reading is picked off after the filter. so you can see why I can so blithely make comments about how my MDS changes by 2 dBm or how 5H3RK who is presently on 10.108 is coming in at -115dBm while my band noise without the preamp is -121 dBm on 30M (-124 dBm with the preamp on) I can also know if I am talking to another F5K owner that if I compare my band noise with his band noise we are talking apples to apples, not just some relative gobbledygook. This is also why I can use this radio as the basis of a lab quality field strength meter, and make real measurements across antennas that actually make sense.

The radio also calibrates itself if you want or need to. Here are some of the calibration screens. Recall this radio is a phasing radio not a crystal filter kind of radio. Phasing radios have many superior features to crystal radios but one thing you have to do is calibrate them. Once calibrated however you can achieve levels of spectral purity beyond what you can achieve with a crystal radio. In the past using passive components obtain the necessary I/Q, non linearity and distortion caused the result to be less than stellar. However in a sampled software radio it can be brought very close to perfect. It is not just another way of doing things it is now a better way of doing things.

Here is a shot of the radio performing a TX image calibration

And here is the the test nearly completed

You can see the signal is almost nulled out

The calibration scheme runs several general tests like a PLL test and a noise test, and then band sepcific tests both on RX and TX. It is these tests that things like RX filters are tested and the RX S-meter is calibrated for each band, and the TX has the filters tested, image nulled and the carrier nulled for each band. As you can see you can choose any or all tests. The result of each test is then included in the central list and when a series is done the results are stored to the EEPROM for later retrivial for use in PowerSDR.

Here is a shot that includes a filter failure on 40M

Don't freak out!! I don't have to box the sucker up and sent it to Austin. The program automatically cycles through and retests any failures, and voila'

I'm ready to pick off that 5H3. As you can see my radio calibrated itself in 11 minutes.

Here are screen shots of the data

If you look closely you will see image rejection numbers on the order of -90 dBm and carrier null numbers in the -120 dBm range. Also you can see that 40M failure in the 3rd screen, and then the retest at the end

Finally you can view your waveform on the scope simultaneously with your transmitted spectrum.

and see how changing something like compander changes look on the spectrum analyzer.

I'm sick of calibration (actually I think this feature is cool beyond belief. This IS how radio should be done.) Time to work some DX!!!!