"pitch accelerometer" is a nice word for a crude mechanism. It's a spring with a weight and it bounces from one side of the cage to the other making electrical contact. Opening the S-TEC "computer" and seeing how it is implemented was one of the biggest disappointments...
You are joking!!!!
Honeywell use the ADXL05. It has been obsolete for some time which may be one reason why Honeywell don't want to pick up the KFC225 project again. They probably do not have any electronics design expertise at Olathe nowadays. The replacement is the ADXL321 which is also going obsolete...
Avidyne's approach of certifying various glass cockpits as roll/pitch source for the DFC90 is very smart and so is the use of existing S-TEC servos (although S-TEC are trying to sabotage that but upgrading existing S-TEC installations, it's a good solution).
I am sure Avidyne used later technology...
What incidentally does amaze me is that nobody is using brushless motors. It's such an obvious thing. Anybody here into model planes? The STEC servos are 100% sure to pack up, and when they do, replacement is quite complicated because the bridle cable needs to be removed, replaced, and retensioned.
Avidyne were recently heard talking about dropping the STEC servos (probably partly because as you say STEC have sabotaged that option commercially) and are going to develop King-compatible servos.
But all of these use brush motors, with the King servos using brush motors of particularly low quality (from Globe, USA).
I would bet the same goes for the GFC700 but nobody I know actually knows. I haven't got my hands on the MM for those yet 
In fact MMs are almost impossible to find for any Garmin gear. The GNS430 MM has only just escaped into the wild, 14 years later (although the site I found it on has been rapidly taken down). A load of IMs are out, however.
whereas STEC use the TC (together with a pitch accelerometer, I believe)
"pitch accelerometer" is a nice word for a crude mechanism. It's a spring with a weight and it bounces from one side of the cage to the other making electrical contact. Opening the S-TEC "computer" and seeing how it is implemented was one of the biggest disappointments...
Avidyne's approach of certifying various glass cockpits as roll/pitch source for the DFC90 is very smart and so is the use of existing S-TEC servos (although S-TEC are trying to sabotage that but upgrading existing S-TEC installations, it's a good solution).
Reading the Avidyne sites, it looks like they need to install and do fairly extensive ground and flight testing on each new type before the FAA will grant the STC. An expensive and time consuming process. I would assume this could be complicated with the DFC-90 due to the s&l button and the envelope protection features.
On a point of detail 
the G1000 can be retrofitted but only into some high-end types.
Socata charge $400k for the TBM700 install.
There must be something about the S-TEC that makes STC a lot simpler than other autopilots. Maybe because it is an old design for which other rules apply?
That's why I am asking the question.
Either they did it using some old rules, or they used some "grandfather route" (certify one type and base later "similar" certs on that; it's often done in e.g. EMC compliance in electronics), or they did, ahem, something else
The King APs use a KI256 as the pitch/roll source (together with a pitch accelerometer), whereas STEC use the TC (together with a pitch accelerometer, I believe). The King system produces far better control, and there are no known marginal stability issues, but I can't see it fundamentally harder to certify, other than the possible factory that if you lose the KI256 or the vac pump, you lose the AP also.
GFC700 do work very well (I flew a TBM850 with the 700, in 2010) but they are not STCd for many aircraft
The GFC700 doesn't have a single STC (supplemental type certificate), it only comes with the G1000 which is not available for retrofit. A shame, the GFC700 would be very popular.
The effort to get an STC must be considerable. The Extra 500 which is built around the Avidyne Entegra R9 still does not have an STC for the DFC90 and ships with a ridiculous S-TEC 55X. There must be something about the S-TEC that makes STC a lot simpler than other autopilots. Maybe because it is an old design for which other rules apply?
What I would be particularly interested in is what data has to be collected to get an autopilot STC.
AIUI, a TSO for the equipment is independent of the STC which allows it to be installed in a particular aircraft type.
The DFC90 and GFC700 do work very well (I flew a TBM850 with the 700, in 2010) but they are not STCd for many aircraft.
I don't know the answer to your question, nor do I have a copy, but the current TSO requires compliance with RTCA DO-325, Minimum Operational Performance Standards (MOPS) for Automatic Flight Guidance and Control Systems and Equipment. It can be purchased from RTCA for $187.50. Copies may not be shared so you are unlikely to find it on the internet in PDF form. It is available in PDF form, but is PW protected and I believe you have to sign a non disclosure agreement.
I don't know what issue you have with Avidyne but for me, they are the only way out of the Garmin monopoly and in the case of autopilots of the S-Tec option.
I have no personal issue with Avidyne (have just spent £12k on the TAS605 for example, in preference to the Garmin product) and agree re the Garmin monopoly being a very bad thing.
And if you read the US forums there is a very strong feeling that anything should be done to block the Garmin monopoly getting any worse.
However I also think that in this game "we" are shamelessly used as both beta testers and - those asked to put down deposits - working capital providers, and it pays to be a bit "selfish" and go for stuff that is more proven, etc. Also some of the "headline" products are quite unreliable if you speak privately to the owners. And Garmin products are basically pretty good.
For example, you know what I think of Honeywell and the shameless way they have treated KFC225 owners, but I am still keeping my KLN94/KMD550 until I absolutely have to rip them out (basically if enroute PRNAV becomes operationally significant in Europe and is enforced so tightly that there is no way around it, which I hope is a very long time) because they are good solid debugged boxes which do all that is actually required for European IFR.
Does anybody know what is involved in determining the stability margins on an autopilot?
I have a DFC-90 and it is a fantastic autopilot.
Peter,
I don't know what issue you have with Avidyne but for me, they are the only way out of the Garmin monopoly and in the case of autopilots of the S-Tec option.
I am considering what to do at the moment as my plane as it stands is not IFR certifiable under EASA, so I'll need to upgrade significantly. Now the question is, do I go the Garmin road with an S-Tec ap or do I wait for Avidyne?
Their 440/540 direct replacements for the Garmin 430/530 series look like a genuine breath of fresh air at competitive pricing and minimal installation cost. Knowing that Garmin will eventually cut support for my 430 1st generation (they have already done so on the 28V version) I am genuinely interested in swapping, possibly for a 440/540 combo to sort out the 8.33 requirement for my com2 as well.
AP, I am not very happy with the recent pricing development with S-Tec, nor with the reports I read. Generally, I'd need a 55X together with an Aspen PFD to get my plane IFR, as I will need both a 2nd Altimeter and HSI and lack the panel space. So why not opt for the DFC90 once it becomes available? Avidyne seems to heavily work with Aspen for the retrofit market.
Achim, the way it looks is your wait may well be at an end: DFC90 EASA certified for C182
