Declination of the accelerometer [Archive]

JayRayMee.NL

18th June 2008, 08:41 PM

Hi folks,

I have figured that there are Diamonds out there with a declination in the accelerometer, which can be noticed by playing Teeter.
Some of them do have the declination, some of them don't.
Mine does :(.

When you hold the Diamond completely straight, or lay the device down on a flat surface (table), the ball doesn't stay in place, but rolls down in my case.
Another guy having the Diamond has the same, but in his case the ball rolls to the other side. A few other Diamonds I have tested where working the proper way, when "flat" means really "flat".

Who has the same? And would there be a way to sort of calibrate the accelerometer?

demonizator

18th June 2008, 09:21 PM

Could it be because of the back casing not being completely flat?

:confused:

JayRayMee.NL

18th June 2008, 09:23 PM

Could be, but I have seen Diamonds where the ball just stayed in the middle without movement when lying flat on a table.

rondol1

18th June 2008, 09:25 PM

Could be, but I have seen Diamonds where the ball just stayed in the middle without movement when lying flat on a table.

You're sure the table was levelled? :D:rolleyes::p

someone1234

18th June 2008, 09:26 PM

I'm sure they only calibrate it roughly. There probabaly is a way to do it yourself, but first we'll need to know the API's for the accelerometer.

slicker

18th June 2008, 09:27 PM

And have you measured the table with a water-lever or similar device? Because if not then putting the diamond on a table that you just believe to be flat seems like a really really stupid argument....

demonizator

18th June 2008, 09:27 PM

The only way you can tell is to put them alltogether on the same table. If you can hold the Diamond in your hands in such a possition so the ball doesn't move anymore I would say your table is just not exactly horizontal.

:D

cina

18th June 2008, 09:32 PM

Try to remove back cover...

JayRayMee.NL

18th June 2008, 09:46 PM

demonizator

18th June 2008, 09:58 PM

May be they're not calibrated right then, like someone1234 said. Can you hold the Diamond in your hands in such a possition - even if tilted, so the ball doesn't move anymore?

JayRayMee.NL

18th June 2008, 10:01 PM

May be they're not calibrated right then, like someone1234 said. Can you hold the Diamond in your hands in such a possition - even if tilted, so the ball doesn't move anymore?

Yes I can.
And then it's tilted (not much, but it is tilted).

cina

18th June 2008, 10:22 PM

I found the solution. HTC knows that they unable to calibrate the accelerometer. So they chose to have that fancy looking backplate to cover this shortness of the device. Hmm. Evil me. :D Sorry

R4nd0m

18th June 2008, 10:33 PM

In my diamond, inside teeter, the ball fall down, even if in a flat surface.

tomorrow it will be replaced with a new one, and i hope it will got a well calibrated one.

at the moment its not a big problem, but think about the future, maybe "improtant" application developed starting from the accelerometer, that will get a wrong feedback...that could be a problem...

JayRayMee.NL

18th June 2008, 10:53 PM

They (or a third party) just need to come with a tool that allows the user to calibrate the accelerometer himself, in sort of way the touchscreen calibration goes. Or something.

Developers challenge?

ljames28

18th June 2008, 11:02 PM

Now, i'm not saying you don't have a right to be annoyed and I see exactly why you'd want to get i fixed. However, does it matter that much? I'm just thinking, it is only an issue when you place the Diamond on a flat surface, how many games are you going to play that involve you placing the diamond on a table? What i mean is, the whole point of the sensor is it detects you moving the device, if you place it on a table the sensor becomes useless anyway.

JayRayMee.NL

18th June 2008, 11:23 PM

Like I said: other than the table, when I hold the device in my hands I need to tilt the Diamond in order to have the ball stand still.
This means there is a declination in the accelerometer, and I would like to have that calibrated. And yes, that matters to me.

l3v5y

19th June 2008, 12:02 AM

In the OEM pack for the Sensor (PACKAGE_Diamond_SensorSDK)
REGEDIT4

REGEDIT4
[HKEY_LOCAL_MACHINE\Services\HTCSENSOR]
"Order"=dword:00000008
"Index"=dword:00000000
"Flags"=dword:00000000
"Keep"=dword:00000001
"Context"=dword:00000000
"Dll"="HTCSensorService.dll"
"Prefix"="SEN"
"DisplayName"="HTC Sensor Service"
"Description"="HTC Sensor Service"
[HKEY_LOCAL_MACHINE\Software\HTC\HTCSensor\GSensor]
"RotateAngleThreshold"=dword:0000002d
"CheckingTime"=dword:000001f4
"TilingAngleThreshold"=dword:00000014
"NotifyAppRotationMask"=dword:0000003f
"FaceupAsPortrait"=dword:00000001
"FaceupToLandscape"=dword:00000000
"LandscapeToP0Threshold"=dword:00000028
[HKEY_LOCAL_MACHINE\SOFTWARE\OEM\MASD]
"SensorSDK"="3_3_18153516_0"
Does editing the "RotateAngleThreshold" value do anything? I think an increase should reduce the sensitivity.

®ichard

19th June 2008, 09:24 AM

I have the same problem.. I was about to start a new thread when I saw this one.

I just sent a mail to HTC with de problem description. Maybe they can tell me what to do about it. If not.. I will contact the company where I did order the device. So they can ship me a new one.

rwijnhov

19th June 2008, 10:24 AM

same problem here

Riel

19th June 2008, 10:25 AM

MasK

19th June 2008, 11:11 AM

Isn't that thing meant for measuring DIFFERENCES (change in G-force) instead of continuously measure currend position of your Diamond?

The second option is pointless to me, and so rough calibration should do exactly what it is supposed to do: know when you change your Diamonds position and to where it changed...

Yes.. thats exactly what I was thinking while reading tru this thread.
I believe differences from 1-3 degrees shouldn't matter much. If you have to significantly tilt the device to get the ball steady, then it's a problem.

Bear in mind that as the levels go higher in teeter, there are different surfaces with various gradients, so you WILL have to tilt it to an angle to get the ball steady.

So far.. Ive done 30+ levels in teeter.. too engrossed in the game to notice any deviations. Importantly, if your Opera browser and photo album react well to the accelerometer, then it's fine.

®ichard

19th June 2008, 04:09 PM

I really have to tilt the device a lot before the ball stays where it is.

If - as some people suggested - I put it on a table and tilt the table to be - what those same people said - flat as far as the device knows.. It's sliding of the table before it is steady..

So it is pretty out of line here.

JayRayMee.NL

19th June 2008, 04:10 PM

If you have to significantly tilt the device to get the ball steady, then it's a problem.

That's the case here.

JayRayMee.NL

19th June 2008, 04:10 PM

KukurikU

19th June 2008, 04:41 PM

Any suggestions of what I have to put there?
i suggest you'll change all 4's with 7's and all 0's with 3's.....:D
just jocking....lol

l3v5y

19th June 2008, 04:43 PM

Any suggestions of what I have to put there?

These two values:
"RotateAngleThreshold"=dword:0000002d
"TilingAngleThreshold"=dword:00000014
back them up, and then increase them to decrease sensitivity...

GanjaBlood

19th June 2008, 04:45 PM

Has anyone tried to reset the device whilst on a flat surface?

A shot in the dark but it's what came to mind whilst reading this thread...

l3v5y

19th June 2008, 04:55 PM

Has anyone tried to reset the device whilst on a flat surface?

A shot in the dark but it's what came to mind whilst reading this thread...
I would try it, but I don't have a device (and I doubt I shall get one).

Sounds like it might work, but if it's the hardware rather than the OS, it may not...

JayRayMee.NL

19th June 2008, 05:04 PM

These two values:
"RotateAngleThreshold"=dword:0000002d
"TilingAngleThreshold"=dword:00000014
back them up, and then increase them to decrease sensitivity...

Done so, soft reset, doesn't change anything.

l3v5y

19th June 2008, 05:06 PM

Done so, soft reset, doesn't change anything.
Check if the registry settings are the values you changed them too.

Riel

19th June 2008, 05:54 PM

Ofcourse soft resetting won't change, when I would soft reset I usually have it somewhere as tilted as can be to reach the red dot :p

JayRayMee.NL

19th June 2008, 07:04 PM

Ofcourse soft resetting won't change, when I would soft reset I usually have it somewhere as tilted as can be to reach the red dot :p

You can, by keeping the on/off button pressed. Confirm to the question that follows.

But, not that it matters, since that doesn't calibrate or anything.

JayRayMee.NL

19th June 2008, 07:04 PM

Check if the registry settings are the values you changed them too.

They have.

l3v5y

19th June 2008, 07:12 PM

They have.Then it may be hardware...

Out of interest, does that registry setting change the responsiveness of the tilting?

JayRayMee.NL

19th June 2008, 07:38 PM

Then it may be hardware...

Out of interest, does that registry setting change the responsiveness of the tilting?

Not that I have noticed.

R4nd0m

19th June 2008, 08:07 PM

got mine replaced with a new one; accelerometer perfectly setted

xwaiz

22nd June 2008, 06:13 AM

same problem here. Put my friend's diamond and mine on the same table in the same room. The ball (Teeter) goes to the bottom on my friend's diamond; and goes to the bottom and left on mine.

I can make it level/balanced by slightly tilting the diamond. So, not a big issue, but, a calibration software to make it "perfectly" level/balanced would be appreciated.

tretre

25th June 2008, 06:13 PM

Same problem here. Btw to see if you have a problem you can lay it on a flat surface, see where the ball goes, and then turn it 180 degrees and retest. If it goes in the same direction (relative to you) then it's ok.

ChiefmasterB

30th June 2008, 11:28 AM

Any new hints on this problem?

Have the same prob than many others...when lying my diamond (without backcover) on a flat (real flat!) table the teeter ball doesn´t stay on his place....

samtaha

30th June 2008, 06:54 PM

Guys,

Trust me, I know what I'm saying.
The thing is that my and a few other Diamonds have the Teeter ball rolling towards one side when held exactly the same way (or lying on the same table) while other Diamonds have the ball lying still in the middle of the screen.

Well, if it makes you feel better, mine is quite a bit off too. The ball will roll quite fast to one side even if the unit is pretty flat. Would be nice to see an application that can calibrate it..

®ichard

1st July 2008, 12:49 PM

After waiting for about 2 weeks for a reply on the mail I sent HTC I decided to make a call.

The first guy (who didn't sound talk like he was having any fun in his job) told me I had to do a hard reset and then whilst starting it again laying it on a flat surface.

Well I did.. and what do you think.. Ofcourse the problem wasn't solved.

So today I made a second phonecall (and got a dutch girl this time which communicated a lot easyer).. After a lot of asking on her colleagues she decided that it was best if I contacted the webshop where I orderd the phone first.

so .. nothing further yet..
now I'll have to search their number first.

I still have a little bit of hope that someone finds a repair manual or something from where a solution might come.

laporte

1st July 2008, 09:44 PM

i think it would be really great and needed to solve this problem.

after i read this thread i had an idea related with the "flip phone to mute" problem many users reporting, just like i have the problem, too, that the flip function does not work.

so i tested right now:

i have to slope my diamond in a special position not to let the teeter-ball roll. (i knew that before ;-))

so, if i hold my diamond in excactely this position and call it, then flip it, the phone becomes mute!
another point to mention: the angle is about 25° face up, 165° face down, so NOT 180° for one flip needed..., but also not the same angle mirrored.

i hope you guys understand what i wanted to describe? :-)

any new thoughts based on my experiences?

drack

1st July 2008, 10:42 PM

I noticed that the phone will mute only if it is layed down on a flat surface, then turning it
when ringing mute the call. If I have the phone on my pocket and then it rings, I can turn it
any position, but it doesn't mute anymore....

laporte

1st July 2008, 10:47 PM

yeh, that's right of course. the phone has to lay face up, the be flipped face down.

but in my and the case of others here, the phone does not know that it lays face up because the sensor has gone wrong and "thinks" it is sloped.

the question is: how, if possible, to change whith values in the registry to recalibrate the sensor manually :-/

laporte

3rd July 2008, 03:19 PM

*push* anyone has ideas or additional input?

®ichard

3rd July 2008, 03:25 PM

*push* anyone has ideas or additional input?

I think that we don't have to look for an answer in WindowsMobile. In that case you would only be fooling WM and learn it what bad input should be right.

With some HTC phone's you can enter a special system menu (sometimes also used to unbrick devices I believe) afaik.

I just don't know which key combi this would be.

laporte

3rd July 2008, 03:29 PM

fooling WM and learn it what bad input should be right. would be ok for me as long it then works properly! ;-)

With some HTC phone's you can enter a special system menu (sometimes also used to unbrick devices I believe) afaik.

never heard about that, anyone knows more about that? i wonder, if there were such a menu, than we would find some post in forum or in tweak-thread?

happyguy82

3rd July 2008, 03:56 PM

Ingvarr

3rd July 2008, 04:24 PM

A real accelerometer will be able to tell you the unit's acceleration in (e.g. meter per second per second).
Obviously.
And when unit is stationary near the surface of Earth, you always get roughly 9.81m/s^2 upwards acceleration vector, thats how accelerometer can be used to determine vertical orientation :)

Condor8

3rd July 2008, 04:36 PM

I have this issue as well, but only noticed it after dropping my unit to the ground. But I'm not sure if it's related.

Btw, it's not an accelerometer... it's probably just a center of gravity detector like an mercury level or something. A real accelerometer will be able to tell you the unit's acceleration in (e.g. meter per second per second).

No, it is an accelerometer! If you check the registry, it says 'Kionix KXSD9'. Check here for the details:http://www.kionix.com/accelerometers/accelerometer-KXSD9.html

And here's how an accelerometer is used for tilt-sensing:
http://www.kionix.com/App-Notes/AN005%20Tilt%20Sensing.pdf

l3v5y

3rd July 2008, 04:43 PM

No, it is an accelerometer! If you check the registry, it says 'Kionix KXSD9'. Check here for the details:http://www.kionix.com/accelerometers/accelerometer-KXSD9.html

And here's how an accelerometer is used for tilt-sensing:
http://www.kionix.com/App-Notes/AN005%20Tilt%20Sensing.pdf
Someone has a little too much time! :p

But, thanks to you, it may be possible to work out how the sensor is being applied in the Diamond, and it may even be possible to build apps using the sensor! :D

laporte

3rd July 2008, 06:41 PM

interesting news, can s.o. get new information out of this?
would be really great if there will be any solution anytime! :-)

l3v5y

3rd July 2008, 06:55 PM

interesting news, can s.o. get new information out of this?
would be really great if there will be any solution anytime! :-)
Pester HTC first, as they'll be most able to help!

I'm going to send them an e-mail regarding using the Sensor anyway 'cause I think it would be useful!! :D And, I'll also add something about alignment...

JayRayMee.NL

3rd July 2008, 07:32 PM

Perhaps if we all send them (lots of) emails they come up with some calibration tool.

Something else: I wonder when the first games using the accelerometer will apear. The iPhone has so many.... why not a few for the Touch?

l3v5y

3rd July 2008, 08:46 PM

laporte

5th July 2008, 07:31 PM

there are some tools available on kionox website.
complete demonstration software including a calibration tool for usb demo boards.
additional a developers source code of this demonstration software.

i don't think that it'll work instantly, can't test it anyway because i don't not have my usb-cable with me, will try it later, but here are real geniuses on forum, is there anybody who could get this software connecting to our diamonds??

i and other users i think, too, would also pay for a solution!

http://www.kionix.com/sensors/downloads.html

fuorionda

8th July 2008, 02:31 PM

up!

i've the same problem :(

dejant_9

10th July 2008, 03:44 PM

same problem here.
http://forum.xda-developers.com/showthread.php?p=2389529&posted=1#post2389529
this is a link for spiritlevel, a tool who can define the position of the ppc. a little bit imprecise. Maby it is not imprecise but the problem come from the Diamond...
i've try with a "real" inclinometer (used for my job) and there is a difference . 3.44° in horizontal axis and 13° in vertical axis between the spiritlevel results and the reality.

(sorry for my very bad english)

JayRayMee.NL

11th July 2008, 12:28 PM

Perhaps this could be of some use to developers/tweakers here....

http://msmobiles.com/news.php/7477.html

JayRayMee.NL

14th July 2008, 12:29 AM

Well, the SDK for the tilt sensor API already gave us application GSEN (http://www.winmo.nl/?p=91) and game lunarTilt (http://www.winmo.nl/?p=96), so I guess a calibration tool shouldn't be too far away....?

MrDerrickC

14th July 2008, 01:57 AM

My Teeter game tilts to the right always, no matter if you rotate the phone in any direction.

Not keen on sending it back to the online store I got it from as I know it will take time.
Will calibration solve this problem (when it is available)? Or does it sound like a hardware issue?

Da9L

14th July 2008, 05:22 PM

I have the same problem.. Is there any news on this?

sysinit

15th July 2008, 10:23 AM

Hi Folks,
Yes, but unfortunately with bad news. :(

I logged a ticket with HTC Support, and they suggested for me to go to their service center. I got in touch with the service center through phone, and was told they don't know how to resolve this problem either.

I had asked for HTC Support to send this problem to their Technique Department. But I guess this case will close once the feedback of the problem is sent. :(

Cheers.

sincere8

15th July 2008, 03:00 PM

I guess patching HTCsensor.dll (or whatever it called) practically can make the sensor work the way it should.

Or somebody could write a resident prog that will intercept the HTCsensor.dll calls and get a corrected value.

Like, you know: originally there are X,Y,Z

So the prog should giveaway:
X_new=X-20
Y_new=Y+10
Z_new=Z

Of course, the numbers will be different for each device.

SamirSan

21st July 2008, 12:49 PM

Anyone with calibration problems here has never flashed his Diamond?

Because I started to have the same calibration problem when i flashed to bepe's 0.54 rom.
The odd thing is that everyone has EXACTLY the same calibration problem (ball goes to the left when playing teeter on a flat table)
So i was asking myself if it's a flash related problem.
I will try tonight to flash back to the original HTC rom to see if that fixes it, and will come back for feedback.
if not i will try to change my device in the orange store...

Please tell me if you never flashed your rom and have tilt g-sensor calibration problems, and what side of the phone your ball goes in teeter...

laporte

21st July 2008, 12:55 PM

i think problem and probably solution aren't that easy as you think/hope.

i had a decalibrated sensor before flashing to bepe, is still have decalibration und will have decalibration until there is a way to recalibrate manually.

btw. my teeterball rolls extremely to the right side and a little bit downside. it's highly unlikely that all sensors are decalibrated the same way and all decalibrations caused by a romupdate! i do not think even one decalibration could be rom-related?!

SamirSan

21st July 2008, 01:10 PM

ok

Tell me if i'm thinking right
IF my ball in teeter goes to the left on a flat table, then i turn my phone 180 degree (still on flat table) and now it goes to the right, that means this is NOT a hardware problem...right?

if it was one, the ball should go the same way on every side i turn my phone because of the accelerometer?

a little bit more explanation : when my phone is on the table on the right side (mini usb in front of me) the ball stays on bottom left of the screen.
Now i turn my phone backwards 180°, (the on/off button in front of me), the ball didnt move, so it's now at the upper right corner of the screen.

If someone can understand my thinking please answer lol.

So maybe it's just a teeter problem (software) ?
can u unsertand me laporte?

laporte

21st July 2008, 01:16 PM

SamirSan

21st July 2008, 01:26 PM

?? your explanation sound wierd!

the ball always rolls to the same side, that has nothing to do with turning it 180° or s.t.h.! the teetergame doesn't recognize turned screen so it's screen is not rotated like the phone, so the ball now rolls to the other side..

i knew i'll have to explain more lol
lets suppose the problem in teeter is that the ball rolls to the right on a flat table.
you lanch teeter and let the ball go to the right side of the screen
if you turn your phone lateraly 180° (as you can see the screen upside down) the ball is now your left side (which is normal because you turned your phone and what was right side is now left side)
So now IF it was an hardware problem the ball should start to roll the the right side of the screen but it doesnt, it stays at the left side of the screen.
-> if the phone is in normal position, the ball goes to the right, if the phone is upside down, to ball goes to the left!

I hope you understand me better its hard to explain for a french guy...

Maybe it's only my phone, but is my logic right?

Sorry if i gave you a headache...

sysinit

22nd July 2008, 09:53 AM

Hi SamirSan,
Your logic is correct. For me, my ball will roll to the right (to the left when upside down).

Do remember on 2 issues:
1. The table that you're doing testing on should be flat.
2. The back cover of Diamond is not flat, so remember to take it off for the test. (I did the test on a glass table, so that I can place the screen of the Diamond faced down. ;))

Cheers.

pweesng

22nd July 2008, 11:08 AM

the back cover of the diamond "at rest" on the table is flat.

i have a pretty well calibrated accelerometer.... and when i use the bubble level program, at rest, the inclination / declination is almost 0 degrees.

when i take out the cover and try, same effect.

So i belieb despite the very "interesting" design of the diamond, if you rest it, it is level.

Kiwi_hknz

22nd July 2008, 12:55 PM

Yes, my diamond is I think also out of alignment, when I start the twee. game and put the diamond on a flat surface the ball moves down to the left side, maybe there is some program to calibrate it.

mugglesquop

22nd July 2008, 01:13 PM

mine seems to be ok, it doesnt sit quite level on the back, so appears to slowely move to the right, but when held level it is ok.

pathologo

23rd August 2008, 01:28 PM

cina, allow me one comment: You are Not Informed Enough.

JayRayMee.NL

23rd August 2008, 07:20 PM

The issue has been solved by using the calibration tool as provided with the Touch Pro.
It is available as CAB download somewhere here. Use the search function, as I don't have the link anymore.

It worked for my Diamond as I have a perfectly calibrated tilt sensor now. That was different before!!

ricarp

10th October 2008, 03:48 AM

Hello, I calibrate my g-sensor with this:

http://dl.htc.com/downloadfiles/Driver/%5CHTC%20Touch%20Diamond%5CDiamond_GSensorCalibrat or.cab

and works OK!!!

amaro001

10th October 2008, 07:36 PM

Actually,

when you study about sensors you'll verify that a sensor must always be settled down to work properly.

So, just emsembling all sensors at the diamond production doesn't means it will work properly, because sensors are so small that any micro-diference between them can cause distortions, but it is a normal thing.

Utilize programs to set the sensors and it will work properly, and don't worry because this can be very normal with any machine.