Monday, April 6, 2009

AGC-T and hearing weak ones

This morning I was listening to VK9GMW out on Mellish Reef on 160. Initially I didn't hear him, but by and by I got things adjusted. The key was narrowing down the filters to 25hz and . I flipped on 50 and I could hear his signal. I flipped on 25 and raised up the AGC-T and I could copy him sending out reports he was weak to be sure but I could copy everything.

Over the weekend I have been busy with family and such. I happened to look at a EHAM article stating if your band noise on 80 or 160 is S-1 then you have problems with your antenna. This is such a ridiculous statement.

Here is my setup on VK9GWM. This shot was taken just after he stopped transmitting. This antenna is an inv-L, nothing fancy

Here is my band noise on 80 this morning

under S-1 My MDS is down around -138 with this set up which is why being able to adjust the AGC-T is so important. Even at -124 dBm I'm still 14 dB stronger than MDS, so its important to be able to adjust AGC-T. The AGC-T is the threshold is measured in - dBm and is were the AGC starts to reduce the gain of an incoming signal. My noise is 14dB stronger than MDS. Below is an example of a RX that starts its AGC at 13 db

This is the Racal RA6790/GM

and this is the RA6790's AGC curve run by Clifton laboratories on this RX

Near Textbook perfect is a circa 1970 receiver. Here are a set of curves from the K-2

Notice how the addition of pre-amplification changes the leftward part of the curves. You often hear that adjusting the RF gain is "just like" adjusting the AGC-T. Clearly these graphs tell a different story. In the cases where the RF amp is on it is the left end of the curve that is adjusted the most. The right end where the AGC is doing its thing is relatively constant. AGC-T allows you to adjust the knee of the curve. It allows you to adjust the point where AGC starts to happen to your radio. In the RA6790 example you can see that there is about 13 dB of head room between MDS and where the AGC starts to operate on the signal. Once the AGC starts to operate there is near perfect dB for dB reduction in gain for over 110 dB variation is signal strength. I don;t know the exact transfer curve of PowerSDR but in my experience it is probably similar to the RA6790, in terms of its linear aspect, but probably has a bit of an upward slope.

Note how important being able to adjust the AGC-T directly instead of by proxy. If you look at the way my RX is set up for VK9GWM, this is where AGC-T comes into play. As you recall my MDS is -138 dBm with the preamp off. My band noise is -119 to -120 dBm. If my AGC-T was set to 13dB like the Racal, my noise would be already forcing my AGC to work. With a signal only a couple dB out of the noise my ability to amplify that few dB is being squelched. To my AGC the VK9 is not much different than the noise, and just when I need the amplification my AGC is turning down the volume. With AGC-T, I can set the knee of the AGC high enough that the VK9 is still being amplified at full tilt, and often for weak signals this is the difference between copy and no copy. The audio sensation is often described as if the station "pops" out of the noise. My experience is that the station goes from non readable to readable.

It is a superior method to AGC adjustment compared to turning down the RF gain or adding attenuation to" tame" the AGC. The way it "sounds" is like the band becomes very quiet in terms of background hash. The fatigue of listening to a noisy band is much much reduced.

Here is a pic of the K-3 AGC

As you can see it is very complicated, with hardware parts and software parts. The various parts are

1 Weak signal
2 Below threshold
3 Software AGC active
4 Hardware AGC active

I've never used a K-3 but the Orion had a similar AGC-T setup and while it was better than my FT-1000D it was still never as good as the PowerSDR AGC loop. I've found this idea that hardware is somehow superior to software derived behavior to not be the case. In fact I have found this to be exactly opposite of the case.

I was in a discussion with a ham who was opining on "baseband" DSP and how they are all the same. It was clear he considered the DSP to be the tail of the dog. He made some kind of big distinction between SDR's that run RF strait into the A/D, and SDR's that convert to baseband. I found his understanding bizarre. And after looking at the AGC of the K-3 it becomes clearer. He still looks at the DSP as some kind of appendage to the "real" radio, instead of the other way around. I guess when you no longer have anything between your antenna and the A/D converter then somehow in his mind that is a different model. The distinction really lays in how much is between you antenna and your A/D. In other words if you have only a really tiny dog and a really huge tail, then perhaps it is the tail that becomes truly the thing of interest. Once the dog is taken to vanishingly small (as in the case of no hardware between ant and A/D), you are still left with the really big tail, and the tail becomes all there is, but it is in that tail where all the work gets done. In fact even if you have a very tiny dog virtually all the work gets done in the tail.

In a legacy radio like the K-3 there is a whole bunch of crap still between the antenna and the DSP The AGC curve tells the story. In the K-3, you see the hardware does subsume everything below it. In my own experience with DSP I can understand where this notion comes from. The DSP was basically an after thought added to the "radio" so it could be claimed as a "feature". In fact there were outboard DSP processors like the Timewave 599 processors that you could add onto any "radio". In the case of PowerSDR, there is only a single direct conversion between RF and A/D. Basically there is nothing that gets in the way of directly processing the signal. In fact PowerSDR could easily be rewritten to run one of these direct from antenna to A/D schemes. It would simply view the direct A/D as A/D data, and process it accordingly. In this kind of scheme PowerSDR is agnostic as to where it gets its data on which to operate. What gets into PowerSDR is something close to audio, but it is not audio. Because you hang the name "baseband" on it does not change its character. It is simply a down converted analogue of the RF. It is a signal 192kh wide. This is nothing like a Timewave 599 or some afterthought on something like the Omni-6. If you plug your headphones into this signal path you won't hear anything distinguishable, because the "radio" which resides in software has yet to operate on this signal and make it into something intelligible. In the case of something like the Timewave, the "radio" had already added all its distprtion etc before DSP. Its interesting how old judgments die hard.

I'm glad I've kissed the majority of the hardware goodbye.