Goetz Romahn's Homepage


Last updated 2002-12-22

I tried two differnt decoders just to compare their performance. The first one is the Rob Alblas design the other one is from the NOAA95 group in Italy.

Alblas decoder

The decoder I use in regular service is the Rob Alblas design. Meanwhile I constructed two of them. Both of them working identical. I made one change to my boards though for reasons to be described below. Additionally there seems to be some strange behaviour of the PLL that could not be resolved in discussion with Rob until now.

Top view of my version of the Alblas decoder, still with the original oscillator circuit (top left).

Alblas decoder

The bad lines with Alblas decoder story

Although the Alblas decoder is very easy to construct and to align ( many thanks for that Rob ) one problem remained with both specimen of my decoders: there were some ten to twenty distorted lines during each satellite pass in the captured images even when the signal from the satellite was very strong. To find the reason for those bad lines I tried nearly everything one could imagine.
Main candidate was the PLL. So I exchanged every component of that, tried different component values and even complete different oscillator designs, VCXOs included.
--- Wait for some satellite passes to check for success -- No improvement!
I tried different receivers. Yes, as you know, I own two very different designs.
--- Wait for some satellite passes to check for success -- No improvement!
Tried different computers, tried different software
--- Wait for some satellite passes to check for success -- No improvement!
Swapped all cables, checked antenna, swapped preamps
--- Wait for some satellite passes to check for success -- No improvement!
Considered even constructing another decoder design ( see below ).
Took me in total about three months of frustrating research.
--- Wait for some satellite passes to check for success -- No improvement!
Finally I tried the nearly unthinkable: I fiddled around the clock oscillator of the "elastic memory" and .... something happened !! I tried various combinations of Xtals and associated Capcitors with no reliable results.
--- Wait for some satellite passes to check for success -- sometimes improvement, sometimes not!
So I decided to take the radical cure, totally removed the original oscillator cicuit and replaced it by an integrated oscillator module.
--- Wait for some satellite passes to check for success -- and voilá, no bad lines anymore !

So if you ever suffered from distorted lines with Alblas decoder although receiving strong signals, consider replacing the oscillator circuit of the elastic memory by some readymade oscillator module connected to pin 53 of the elastic memory FPGA. I tried two different modules. One with frequency of 5.12 MHz the other one with 10 (!!) MHz. Both are working!

See my simple modification of the decoder below. FPGA IC4 is used for "elastic memory" in Rob's design

Modification of Alblas-Decoder

Decoder alternatives. The NOAA95 decoder.

To my knowledge there are only two decoder designs published for the DIYer namely the Alblas decoder and the Noaa95 decoder. Both designs could be called digital in contrast to analog decoders. Just to compare the performance of both ( and as a last resort if all my bad-lines-experiments had failed, see above ) I also constructed one specimen of the NOAA95 decoder. Construction was done with the help of an ISA bus prototype board and wiring was done by threading enamelled copper wire, since PCBs from the NOAA95 group are sold out. Only the decoder part was built actually, not the IF- and demodulator part.
To make the story short: performance ( or better say noise margin ) equalled that of the Alblas decoder up to single lines.

Analog decoders are said to give better ( up to 3dB ) signal to noise margin than digital decoders because of their integrate-and-dump function for bit recovery. Up to now I have not found any published benchmark that compares different designs of decoders to confirm this advantage of analog decoders. Remember, 3 dB advantage translates to double the surface area of a dish antenna or double the length of a beam antenna, so this could have serious impact on total system performance. Please give me a note if you have more information on that aspect.