After having acquired my own TCA yoke I decided to investigate the reason for the roll axis issues reported by some of the users. The first thing to check was the roll axis sensor as some of the users were reporting a strange dependency of sensor operation on the external magnetic field (which shouldn't be the case with a properly designed sensor).

So after the disassembly I was very much surprised to see the roll axis sensor to be designed around a single AH49E chip which is a simple "old-style" analog Hall-effect sensor. There is a couple of issues with this solution:

1. AH49E is reacting to a magnetic field strength variation in any direction relative to it's body. When a small round diametral- magnetized magnet is rotated close to it, the field strength change is non-linear in relation to the rotation angle. Which means that the yoke's reaction to roll axis rotation is stronger near the center and becomes weaker close to the rotation extremes. Of course it can be compensated in the sim by choosing a non-linear response curve, but in my opinion having a sensor with a non-linear response and no internal compensation for it is simply not right.

2. The magnet is not too strong, so the output signal swing for a full side-to-side yoke rotation is mush less than a sensor power supply voltage. When the yoke internal calibration is performed, the signal is- well- 'amplified" digitally within a yoke controller to match the controller's reference signal voltage swing (from 0 to full power supply voltage). Unfortunately any "noise" (magnetic or electric) that might be induced to the sensor circuit/ connecting wires (that are quite long) will be amplified as well increasing the roll axis sensitivity to the external interference.

3. As AH49E is sensing any changes in the adjacent magnetic field strength, any external ferromagnet object near the back of the yoke handle will create a disturbance in the roll axis; the same goes for the stability of the internal magnet. In some cases this might require a yoke re-calibration.

My solution to this issue is the replacement of a simple Hall-effect sensor by a different type specifically designed for registering an axis rotation. In my opinion the best DIY option here is AS5600 from AMS. The main features of the chip are:

1. It is reacting to a degree of magnetic field rotation in relation to a chip's upper/ lower cover. The signal is directly proportional to the rotation angle with a very good linearity.

2. AS5600 offers a very simple internal calibration routine which adjusts the sensor's output signal swing (from 0 to power supply voltage) to a full rotational angle required. This means a much better immunity to the external "noise".

3. AS5600 is not sensitive to the magnetic field strength variations so the external ferromagnet objects do not influence the sensor operation. Also there is no need of a periodic yoke re-calibration.

In practical terms the option that I've chosen was to use the ready-made AS5600 board available from Amazon or eBay (also from AliExpress). It has all the external elements needed and the size allows to fit the board into the existing opening in the roll sensor base without the need of any mechanical trimming of either the sensor or the base. The board requires soldering one additional jumper to set the correct power supply voltage. Here is the set of pictures showing the mod. After removing the existing sensor and mounting the new board the new sensor needs to be calibrated internally but it's a very simple procedure not requiring any additional tools.

Having done the mod I'm quite happy with the results. If any of you are interested, I'll be happy to provide the additional details.

 

Thanks for the info. It falls along the original post here. Good to know I can order some parts if the magnet fix I'm using stops working. 

 

53 minutes ago, Wise87 said:

It falls along the original post here.

Yep; actually your post was one of those triggering my investigation- thanks for that!😀.

@SergeyPe

Hi there!

Chanced upon your thread while searching for the possible fixes. Wonder if you could help me on the following questions

 

1) how do u internally calibrate the new sensor?

2) what filament u used for the sensor holder?

Many thanks!

Edited August 22, 20241 yr by Ray Proudfoot
Long quoted post removed.

Hi!

1. The calibration is explained in the AS5600 datasheet (Page 23- Angle Programming Through the OUT Pin). Please note that in addition to the red jumper that you can see on my pictures #7 and #8 you'll also need to connect the two upper leftmost contacts on the PCB to set the output voltage increase with a magnet clockwise rotation (I've just put a drop of solder there). The wire loop that can be seen on the pics is connecting the PGO pin to the ground to put the chip into programming mode; it should be cut off after the calibration.

2. I used PETG, but PLA or ABS will do as well.

Regards,

Sergey

I was actually going to post an update on the other thread regarding this. My first yoke had this issue so I returned it, then after a few months the replacement started doing the exact same thing. I tried taping a magnet to the back of the yoke but it would not calibrate, I tried using zip ties to hold the ribbon cables to the yoke shaft, nothing worked and it started getting worse to the point it was unusable so I reached out to ThrustMaster support. After some initial investigations (they were just going through their paces) they had me send it back to a repair centre and from Canada the one they chose was in France ! I was a little freaked out that it would be gone for weeks possibly longer. I sent it via FedEx on Aug 20 and it was delivered back to me yesterday via UPS (all at TM's cost).

I sent them an email asking what the root cause and fix was and they replied within 15 minutes with the following:

"A simple maintenance procedure was performed on your TCA Yoke Boeing Edition. The firmware has been updated, and the product was reworked, with reinforced components and soldering."

The last part was the key as I had tried firmware and countless calibrations to no avail, I read in another post a different user had to do the same thing. I commend SergeyPe for hi post but that solution would have been a bit too much for my skills lol.

If anyone else is reading this and has this issue contact TM support and create a ticket, for me the whole process was super simple and very quick and I am happy to report my yoke is working perfectly. I just departed EGLL in the 777 and hand flew her all the way up to the cruise FL370
 

Hi guys, I just have registered. I'm so sorry for not posting any welcome things somewhere in the forum.
I will welcome you here :P.
So, to the clue - I have a TCA Yoke - a brand new one. And one thing which is very annoying is a huge deadzone. Of course I know that pitch axis deadzone is not so big problem, because we trim mostly a pitch axis, right.
But roll axis... it's so hard to do precise landings, especially when we have a soft windshear.
I've recorded a video and I've sent it to Thrustmaster with a question how to reduce a deadzone. Still waiting for response.
Now when I finally reached to this topic, I wonder is that maybe eaven more worth than loosing warranty.

So guys... the most important question from me comes down below:
Dear @SergeyPe, did you recognize any differences with a deadzone after changing hall sensor to the AS5600?

I wonder if that deadzone is set on TM board or it depends on specific hall sensor.
Maybe TM did so wide deadzone because of AH49E sensor, which maybe cannot handle precize center positions?

Hi @TomaszBartchie and welcome to the club! 

Honestly speaking, I haven't checked the deadzone difference after changing the sensor, also because the TCA Yoke is not my main one- I am usually flying with a Fulcrum (heavily modified), which in my opinion is well ahead of any other non-feedback yokes in the market.

It might well be that TM used a deadzone to cope with a non-linearity of AH49 which is most sensitive around the center position. One thing that comes into my mind is that the possibility of adjusting the response curves to reduce the deadzone is built into TM Target software (if I am not mistaken, as I didn't use Target for quite a long time). Please give me some time to pick the yoke out of storage and play with Target- I'll be back with the results.

OK, I've checked the roll deadzone on my TCA Yoke. If measured at the outmost of the yoke handle it's about 5 mm in total which IMHO is quite OK. I wonder what's the result on your unit?

BTW- a dumb question: are you getting a big deadzone both in the sim and in the testing program? Typically the deadzones for all the axes are set quite big by default in all the sims.

I've also checked TM Target- the deadzones there are set to 0 by default for TCA Yoke, so it won't help.

Hello Sergey.

If I see right. This yoke has a design flaw in the form of an outdated hall sensor board. So dead zone of roll axis is larger then honeycomb alpha? Wnat do you think? 

Hello alex40.

Although I had an Alpha some time ago, I've never compared the deadzones in these two yokes.