Sunday, March 07, 2010

Calibration update for Windows 7 Viliv S5 HID Driver

Last week I posted about the release of Viliv's Human Interface Device (HID) Driver for their S5 Mobile Internet Device (MID). Many of you reported mixed results (most came via Engadgets story) but sadly a common theme amongst the feedback was a change in sensitivity of touch, while others reported total loss of calibration and even dead spots. Ultimately I want to thank you for participating in the process and commend all of you for being part of the solution! Part the reason I write this blog is to act as a bridge between manufacturers and end-users and with your contributions mere conversations become so much more!

I spent the week working with Viliv Senior Technicians to resolve the HID Driver issues and reporting back each and everyone of your comments. Something that has become clear is that HID users need to realise that the presence of a HID Driver will in fact alter the touch experience; there are many technical reasons for this but in plain English the difference is based on the new layer of "thinking" being conducted by the driver to help Windows understanding exactly why you are touching and what your intended action is. You must expect a change in sensitivity but by no means should you expect dead spots!

With that in mind...here is the solution from the factory:

Please use the instructions below to enhance the sensitivity and accuracy for S5 
  1. Check the attached “S5 reset.jpg” file
  2. Reset the Calibration from Tablet PC setting
  3. Execute S5cal.bat and calibrate 25point
As always I will make myself available in the coming days to answer your queries and pass on your comments once again.

Cheers! Enjoy!

23 comments:

quenthal said...

First comments: sadly no changes. There is still this "drifting" issue with lefts side and the top of the screen.

Outside those drifting areas the accuracy is good.

The drifting area begins about 25 pixels before the top or the left side of the screen. For example when you are moving the pen at constant, smooth and slow speed from right to left, the cursor moves accurately under the pen.

When you move over that invisible 25 pixel border, the cursor begins to move faster and quickly reaches the border of the screen ahead of the pen. Same happens with top of the screen.

So to summary: With 16 or even with that 25 point calibration instructed here you can get rid of dead spots, but there is totally no accuracy under those drifting areas. If at later time I make simply 4 point calibration, those drifting zones become completely dead.

Hugo Ortega said...

Once again Quenthal, well done on the feedback. I have not heard from anyone else so it is either OK for everyone else, or no one has tried it yet. :-)

I will report this to Viliv and see what they say! Cheers.

PS. Let me know your EC info (found in Viliv Manager) and OS and perhaps any other info that might help.

Scott Sanders said...

How do you activate the reset button? mine is grayed out.

airbillion said...

Thanks for the effort, but no luck. When I run the file the 16 point calibration tool runs, not a 25 point tool. I did the calibration and when it finishes and I press OK, I receive an error message: Linearity calibration already present. Please use the reset button in the Tablet PC Settings control panel applet to remove the data. What could be the problem? I have reset the calibration data and preformed the calibration a few times without any luck. Please help!

jANjALANi said...

Yeah I agree with Quenthal, It didn't fix the drifting problems. the 25 point calibration didn't fix it. most probably its a driver problem. because if you go back to the old Atom Touch Driver there's no drifting... but there's no tablet mode too.

Ryan said...

Hugo, first of all thanks for the added instructions. But quenthal is right about the top and left part of the screen still "drifting". I can say that the 25 point calibration has made my pen point more accurate, but the top and left few pixels still have the previous problem. If you currently have a Viliv S5 with you, try using a stylus on the Win7 desktop screen. From the bottom of the screen, tap and drag your stylus towards the top. You'll notice the point/cursor corresponds to the position of your stylus point all the way (very accurate), until it reaches this invisible boundary (as quenthal has penned it) a few pixels before the top where the point/cursor suddenly moves faster to the top than your actual stylus point. Same goes with tapping and dragging the stylus from right towards the left. It makes me think the HID driver thought the top or left end of the screen ends at that invisible border, and it can't really handle inputs outside of this border accurately.

My EC version is 054, BIOS version is 108, Model S5 3G.

Hugo Ortega said...

@Scott: Have you installed the driver from the first link?
@AirBillion: Great feedback mate! I am going to bundle it with other feedback I got and go into bat for you all.
@JanJalani: Thanks mate, perfect. This is exactly what I need to know. Will pass your comment on also.
@Ryan: The EC info will help, and so does your detailed respsone. Cheers!

@All: I'm working on this between meetings today so stay patient. Your feedback is great Team. Well done!

Ryan said...

@airbillion: When you execute the "s5cal.bat", just click on the reset button (and if it is grayed out, no need to do anything), and press okay. The 25 point calibration comes AFTER you press the okay button. If you did a calibration before pressing okay, then the error message you've mentioned displays on screen, because calibration data was already populated at that point. Let the 25 point calibration populate it instead.

airbillion said...

@Hugo & @Ryan
Thanks for the help! I got the 25 point calibration to work without any errors, however I have the same drift problem as everyone else. It must be either a driver or EC issue, but I still would not go back to the old drivers, the windows 7 touch features are too important to me! Hopefully viliv will sort this out soon. Thanks for your effort Hugo, keep up the good work!

quenthal said...

Again, thanks go to you Hugo for your efforts on this issue.

About my current configuration:

I'm using is Windows 7 Ultimate with latest updates and latest (beta) driver for Intel GMA.

EC is 054 and BIOS 108.

Not sure if it has anything to do with this issue, but when I've cleared all the calibration information or I've done clean reinstall, I can see that the cursor is slightly top left. Could this default "calibration" offered by the driver have something to do with this top/left drifting issue?

Secondly, has anyone else noticed that using Viliv's joystick in mousemode acts funny when you go over the left or top border of the screen? Or is this normal?

Also (I can't check my device right now) has anyone else noticed anything strange regarding touch driver in their Event Log?

sysmgr2 said...

I've tested two of my S5's, and I have to say things are much better with the new calibration tool and the use of a Cross stylus pen. I see no evidence of 'drift', and the calibration now works correctly to all edges of the screen. I'm running bios 1.08 and EC 022.

Hugo Ortega said...

@quenthal & @sysmgr2: Thanks heaps for this elvel of detail. Interesting that Sysmgr2 got the result we wanted. I know Viliv i working on this so I'll keeo you posted!

quenthal said...

@Hugo & sysmgr2:
One thing to be noted though is that there might be some accuracy gained elsewhere on the screen, by doing full 25 or 16 point calibration. Also (sysmgr2 can correct me) I think that sysmgr2 didn't suffer from this "drifting" issue even before this update, only that the second s5 had slight offset.

Hugo Ortega said...

@quenthal: that's why I am not ruling out that it is a hardware idiosyncracy on certain builds.

quenthal said...

@Hugo:

Do you think we should try to collect serial numbers? I think it is possible to check out the s/n without removing the battery by using Viliv Manager.

Hugo Ortega said...

Not just yet mate! Viliv is telling me there is "no" issue (as a general statement) but is working on a solution based on the "software" issue you are experiencing, i.e. they think "if" there is an issue it is in fact software related. If I need the numbers I will ask for them! Cheers!

sysmgr2 said...

For the record, let me be clear about my experiences:

S5#1: No issue, but more accurate after 25-pt calibration.

S5#2: Without calibration, cursor offset almost 3/16 of an inch vertical and left. No observed speed change or accuracy problem in left/upper part of the screen. 25-pt calibration resolves offset issue.

Discussion: #1 is and S5 Premium-EN, while #2 is a 'first batch' H-LOG machine. Both are running bios 1.08 and EC 022.

Hope this helps...

airbillion said...

Why hasn't viliv officially released the driver on their site yet? They have the old touch driver, but not the new on yet. Maybe the driver we are using is a beta one, and that is why we are having drifting issues.

Hugo Ortega said...

@sysmgr2: Thanks!

@airbillion: It was developed specifically for a project in Australia. Due to my close connections in the Viliv office I have permission to hand this out and therefore help test the driver and report back bugs. You guys are a close connection of mine hence the handout! :-)

While it is not officially in BETA, or not officially released, it is a finished product. Ironically for the project it was developed, it works well. By all accounts there should not be any issues so they have halted the offical release thanks to your feedback and would like to get a solution before the official release.

Shows you just how effective blogs can be! ;-)

quenthal said...

@Hugo

Especially your blog, as it helps Viliv users all around the world:

There has been some discussion on this same matter at Pocketables forum (Viliv S5 subforum), but as you can see, the most active thread about touch drivers over there (Link) is based on Link.

As a side note, I noticed that used resolution does not have any effect on this drifting issue. When running other resolutions than native 1024x600 in full screen (and stretched if aspect ratio is not native for Viliv), there is the same drifting going on on left and top border of the screen. Naturally this means that the 25 pixel border I mentioned earlier applies only to 1024x600 resolution and the invisible border is more "physical".

Also, I fiddled around the .bat file you provided, and added even more calibration points, but no use.

Hugo Ortega said...

@quenthal: Love your work mate. I also added more calibration points but found nothing a few days back. I like what you did with resolution though and that might well release some new train of thought. Am passing this on now...

David said...
This comment has been removed by the author.
David said...

ps: i also have the problem that the few pixels on the top of the screen can not be addressed just as it would think the screen end there.