T16 Transponder
Posted: Tue Jan 30, 2024 11:09 am
It's time for an update - probably my longest development yet.
The first prototype was done in 2016 - pretty much a traditional receiver and using a TX transistor made for the purpose (rather expensive little bugger that). Did not have much in the way of suitable RF test equipment at the time but the basics where OK.
I then did a second prototype ditching the old fashioned superhet receiver and experimenting with a satellite receiver chip - that was interesting from a mode-s perspective as decoding the data uplink from secondary radar would be much more robust (I/Q receiver).
To reduce cost the entire pulse decoding and receiver (including mode-s phase decoder) was moved into the processor. This handled the TX as well. No FPGA in sight. It's quite a challenge - you have to receive data at 4MBits/s on the Mode-S link and measure pulse amplitudes and times to high levels of accuracy. Usually this is done with dedicated hardware - but that costs money and needs more parts. Anyway, got that working (interestingly managing to do 8 million arctan calculations per second with precise timing). Transmitter starting to produce a little power- but not enough. My idea to further save costs - the whole transponder, like the V16 radio must be on a single PCB and further to that no exotic PCB materials are to be used (you normally need impedance controlled ceramic substrates for high frequencies - costly stuff).
The third prototype cleaned things up a little and introduced a small amount of additional hardware to assist with pulse edge detection as the previous version could not get down to the required level of resolution (nano-seconds). Transmitter still struggled to get enough power out but it was just at the level needed for a Class 1 transponder. I wanted nearly double that.
So I relented and did a forth prototype which moved the final stage of the transmitter to its own PCB - but still without the need for expensive materials.
Then Covid hit.
The transponder was put on hold, some of the components we already bought for a first production batch where used up to keep other product lines going - component shortages and supply issues forced a redesign of nearly all our products to use alternative parts. It was a nightmare. Then I needed to spend around a year to do the G4 CPU which will be phased into future EFIS systems - in addition to a rather involved time spent on porting the Linux operating system to it - and porting the EFIS so it would run in Linux.
Now that is done and I have dug out the transponder again - switched the forth prototype on for the first time and fiddled with the transmitter. This now works great - I'm getting about 230W of power and everything is properly within limits (spectral masks, harmonics etc). Nice.
It works really well as Mode-A/C at the moment and the test results are quite a bit better compared to an expensive transponder I am using as reference for comparison. The Mode-S side of things is implemented and I'm going through that at the moment. It all looks good - should be done by end of the week.
The official classification of this transponder is a Class 1, level 2el to the latest DO-181F standard (equivalent to Class 1, level 2els in previous versions - the "s" now being mandatory and no longer part of the label).
So far so good - but of course there is a road block - there always is. Now for the red tape. Sadly our local CAA has lost the plot and is no longer able to certify anything and has not renewed bilateral agreements with the FAA. This means a manufacturer in our country can no longer produce a certified part for an aircraft. It's not the end of the World - but I am now looking for options on how to get this thing to you.
The first prototype was done in 2016 - pretty much a traditional receiver and using a TX transistor made for the purpose (rather expensive little bugger that). Did not have much in the way of suitable RF test equipment at the time but the basics where OK.
I then did a second prototype ditching the old fashioned superhet receiver and experimenting with a satellite receiver chip - that was interesting from a mode-s perspective as decoding the data uplink from secondary radar would be much more robust (I/Q receiver).
To reduce cost the entire pulse decoding and receiver (including mode-s phase decoder) was moved into the processor. This handled the TX as well. No FPGA in sight. It's quite a challenge - you have to receive data at 4MBits/s on the Mode-S link and measure pulse amplitudes and times to high levels of accuracy. Usually this is done with dedicated hardware - but that costs money and needs more parts. Anyway, got that working (interestingly managing to do 8 million arctan calculations per second with precise timing). Transmitter starting to produce a little power- but not enough. My idea to further save costs - the whole transponder, like the V16 radio must be on a single PCB and further to that no exotic PCB materials are to be used (you normally need impedance controlled ceramic substrates for high frequencies - costly stuff).
The third prototype cleaned things up a little and introduced a small amount of additional hardware to assist with pulse edge detection as the previous version could not get down to the required level of resolution (nano-seconds). Transmitter still struggled to get enough power out but it was just at the level needed for a Class 1 transponder. I wanted nearly double that.
So I relented and did a forth prototype which moved the final stage of the transmitter to its own PCB - but still without the need for expensive materials.
Then Covid hit.
The transponder was put on hold, some of the components we already bought for a first production batch where used up to keep other product lines going - component shortages and supply issues forced a redesign of nearly all our products to use alternative parts. It was a nightmare. Then I needed to spend around a year to do the G4 CPU which will be phased into future EFIS systems - in addition to a rather involved time spent on porting the Linux operating system to it - and porting the EFIS so it would run in Linux.
Now that is done and I have dug out the transponder again - switched the forth prototype on for the first time and fiddled with the transmitter. This now works great - I'm getting about 230W of power and everything is properly within limits (spectral masks, harmonics etc). Nice.
It works really well as Mode-A/C at the moment and the test results are quite a bit better compared to an expensive transponder I am using as reference for comparison. The Mode-S side of things is implemented and I'm going through that at the moment. It all looks good - should be done by end of the week.
The official classification of this transponder is a Class 1, level 2el to the latest DO-181F standard (equivalent to Class 1, level 2els in previous versions - the "s" now being mandatory and no longer part of the label).
So far so good - but of course there is a road block - there always is. Now for the red tape. Sadly our local CAA has lost the plot and is no longer able to certify anything and has not renewed bilateral agreements with the FAA. This means a manufacturer in our country can no longer produce a certified part for an aircraft. It's not the end of the World - but I am now looking for options on how to get this thing to you.
