2017-11-18 02:12:26

by Giulio Benetti

[permalink] [raw]
Subject: Re: edt-ft5x06 question

Hi Luca,

Il 17/11/2017 18:16, Luca Ceresoli ha scritto:
> Hi Giulio,
>
> On 15/11/2017 17:30, Giulio Benetti wrote:
>> Hi Luca,
>>
>> I had problems with e-mail filter, so I rebuilt yours by hand.
>> Forgive me.
>>
>> See below.
>>
>> Thanks
>> Kind regards
>>
>>> Hi Giulio,
>>>
>>> On 14/11/2017 22:42, Giulio Benetti wrote:
>>>> Hello everybody,
>>>>
>>>> I'm using ft5206 with edt-ft5x06.c driver,
>>>> but what I see is that registers with M09 or M06
>>>> seem different from focaltech ft5206
>>>> https://www.buydisplay.com/download/ic/FT5206.pdf
>>>>
>>>> For example, except address 0x80 for threshold register(thgroup),
>>>> all the others don't appear:
>>>> GAIN 0x30 or 0x92 should be 0x82(thcal maybe?)
>>>>
>>>> Can someone clarify this?
>>>
>>> Are you having any problems, or just investigating?
>>
>> I need to add some other IC register to calibrate sensibility with
>> different glasses, but starting from ft5x06 or other Focaltech standard
>> IC FW.
>
> If I understood correctly you want to implement auto-calibration, and in
> order to do that you need the four extra registers, not mentioned in the
> Focaltech datasheet. Right?

Yes, I need them inside the driver.
And some of them could be useful also as attributes.

>
>>>> Maybe M06 or M09 are a custom firmware for focaltech ft5x,
>>>> then it has different registers?
>>>
>>> All the FT5x06 are 8051-based chips designed to implement touchscreen
>>> controllers. The FT5x26 are similar but with a more powerful core
>>> (Cortex-M IIRC).
>>>
>>> As far as I could understand and guess, their register interface is
>>> fully implemented in firmware, so it could change completely from one
>>> panel to another. However all panels that I have seen seem to use
>>> firmwares with either the "M06" or the "M09" protocol, which are
>>> implemented in the Linux driver.
>>
>> In my case driver recognizes M09, but only address 0x80
>> register(thgroup) is correct.
>> The others:
>> GAIN: 0x92
>> GAIN: 0x93
>> GAIN: 0x94
>> GAIN: 0x95
>>
>> are not described in that datasheet.
>
> What about all the other registers? The datasheet lists a few dozen, not
> only 0x80. Are they implemented in the panel you are working on?

Yes they are, but as Datasheet is signed, this makes me believe that
that is the default firmware focaltech load in factory.
So yes, they're implemented in 4 different panels I have.

>
>> The point is that datasheet seems to be official by Focaltech,
>> like if they deliver that IC with a standard FW inside.
>> I can't find a way to safely probe if it's a standard FW or from EDT M09
>> or M06.
>
> Since everything is based on the programmed firmware, and there's no
> "standard" way to detect firmware version, every firmware detection is
> bound to be a hack. But there's no better way I'm afraid. Thus device
> tree seems the only reliable way...

I think the same.

>
> However, did you check registers FIRMID, FT5201ID (CTPM Vendor ID) and
> ID_G_LIB_VERSION_H/L? According to their name they might hold
> interesting info.

Still not, I will.

>
> And did you consider read the "additional registers" tentatively during
> probe to detect whether they are present? What effect would this have on
> chips without those registers?

This doesn't sound so good to me.
It gives me an'idea of unstability.

So how can I go on?
Do I try to write a specific focaltech-ft5x.c driver?

>
> Bye,
>


--
Giulio Benetti
R&D Manager &
Advanced Research

MICRONOVA SRL
Sede: Via A. Niedda 3 - 35010 Vigonza (PD)
Tel. 049/8931563 - Fax 049/8931346
Cod.Fiscale - P.IVA 02663420285
Capitale Sociale � 26.000 i.v.
Iscritta al Reg. Imprese di Padova N. 02663420285
Numero R.E.A. 258642

From 1584356367443559859@xxx Fri Nov 17 23:06:15 +0000 2017
X-GM-THRID: 1584085398794430736
X-Gmail-Labels: Inbox,Category Forums,HistoricalUnread


2017-11-17 08:51:39

by Giulio Benetti

[permalink] [raw]
Subject: Re: Re: edt-ft5x06 question

Hi Simon,

sorry for the mess with mailing list.
My mail filter gave me problems and removed e-mails from linux-input
mailing list.

> Hi Giulio
>
> On Tue, 2017-11-14 at 22:42 +0100, Giulio Benetti wrote:
> > I'm using ft5206 with edt-ft5x06.c driver,
> > but what I see is that registers with M09 or M06
> > seem different from focaltech ft5206
> > https://www.buydisplay.com/download/ic/FT5206.pdf
> >
> > For example, except address 0x80 for threshold register(thgroup),
> > all the others don't appear:
> > GAIN 0x30 or 0x92 should be 0x82(thcal maybe?)
> >
> > Can someone clarify this?
>
> The Registers depend heavily on the actual firmware flashed into the
> ft5x06. M06 and M09 refers to the resp. touch glasses from EDT (hence
> the name of the driver), who have a custom firmware installed. Despite
> being based on the same focaltec chips, their i2c interface is totally
> different...
>
> Most non-EDT ft5x06 touches do use a communication protocol similiar to
> the M09 series. Unfortunately I've witnessed a lot that it is not
> possible to rely on the registers that are supposed to provide
> vendor/model information (according to the Focaltec datasheet). It
> always takes a little bit of tweaking to figure out what resolution
> they actually provide (which is possible with the custom EDT-M09
> firmware).
>
> There is a patch series I've sent to this mailinglist (and need to
> resend, because the first part of the patch series is bogus) that might
> improve the model detection a little bit, but it by no means is
> bulletproof.
>
> (see https://patchwork.kernel.org/patch/9987467/ and
> https://patchwork.kernel.org/patch/9987471/)
>
> For non-EDT touches that are bonded together with a display it is a
> reasonable assumption that the touch resolution matches the screen
> resolution (Which I consider unfortunate: Why throw away more
> resolution?).
>
> Anyway if you have suggestions for improving this driver I am
> interested in feedback.

I only have an idea,
since we can't know for sure what's the Vendor,
and we can't know for sure max_x and max_y,
I would like to add a field in dt to select which vendor to use.
For example, if you choose "Edt" as vendor, than driver behaves like it
does now, otherwise there could be "Focaltech" as vendor, and in that
case I could find a way to get max_x and max_y.
If I can't find a way, we put the need to be inserted in dts both max_x
and max_y.

Also I would like to export more registers as attributes and dt.

And last. In my company, on previous projects,
we experienced problems with temperature and mechanical drifts.
So the only solution was to send an autocalibrate every n seconds,
when ts is untouched.
I did this adding a timed tasklet inside driver,
so I would like to do the same.

Now what comes in my mind is,
do we need to keep everything in the same driver edt-ft5x.c?
Or does it make sense to write new one(focaltech-ft5x.c) based on
edt-ft5x.c?

Best regards

>
> Bye,
> Simon
>
> --
> kernel concepts GmbH Simon Budig
> Sieghuetter Hauptweg 48 simon.budig@xxxxxxxxxxxxxxxxx
> D-57072 Siegen +49-271-771091-17
> http://www.kernelconcepts.de/
> HR Siegen, HR B 9613; Gesch�ftsf�hrer: Ole Reinhardt
--
Giulio Benetti
R&D Manager &
Advanced Research

MICRONOVA SRL
Sede: Via A. Niedda 3 - 35010 Vigonza (PD)
Tel. 049/8931563 - Fax 049/8931346
Cod.Fiscale - P.IVA 02663420285
Capitale Sociale � 26.000 i.v.
Iscritta al Reg. Imprese di Padova N. 02663420285
Numero R.E.A. 258642

From 1584186291369816083@xxx Thu Nov 16 02:02:58 +0000 2017
X-GM-THRID: 1584085398794430736
X-Gmail-Labels: Inbox,Category Forums,HistoricalUnread

2017-11-15 17:21:30

by Giulio Benetti

[permalink] [raw]
Subject: Re: Re: edt-ft5x06 question

Hi Luca,

I had problems with e-mail filter, so I rebuilt yours by hand.
Forgive me.

See below.

Thanks
Kind regards

> Hi Giulio,
>
> On 14/11/2017 22:42, Giulio Benetti wrote:
> > Hello everybody,
> >
> > I'm using ft5206 with edt-ft5x06.c driver,
> > but what I see is that registers with M09 or M06
> > seem different from focaltech ft5206
> > https://www.buydisplay.com/download/ic/FT5206.pdf
> >
> > For example, except address 0x80 for threshold register(thgroup),
> > all the others don't appear:
> > GAIN 0x30 or 0x92 should be 0x82(thcal maybe?)
> >
> > Can someone clarify this?
>
> Are you having any problems, or just investigating?

I need to add some other IC register to calibrate sensibility with
different glasses, but starting from ft5x06 or other Focaltech standard
IC FW.

>
> > Maybe M06 or M09 are a custom firmware for focaltech ft5x,
> > then it has different registers?
>
> All the FT5x06 are 8051-based chips designed to implement touchscreen
> controllers. The FT5x26 are similar but with a more powerful core
> (Cortex-M IIRC).
>
> As far as I could understand and guess, their register interface is
> fully implemented in firmware, so it could change completely from one
> panel to another. However all panels that I have seen seem to use
> firmwares with either the "M06" or the "M09" protocol, which are
> implemented in the Linux driver.

In my case driver recognizes M09, but only address 0x80
register(thgroup) is correct.
The others:
GAIN: 0x92
GAIN: 0x93
GAIN: 0x94
GAIN: 0x95

are not described in that datasheet.
The point is that datasheet seems to be official by Focaltech,
like if they deliver that IC with a standard FW inside.
I can't find a way to safely probe if it's a standard FW or from EDT M09
or M06.
I wouldn't write a different driver ft5x.c because it's a repeatition,
but EDT seems a special version(FW) of ft5x.

Maybe on dt-bindings I could force to be EDT or standard.
On legacy sunxi-3.4 the driver handled this as datasheet I mentioned.
What do you think?

>
> The driver also performs some checks to understand which of the two
> protocols is implemented by the controller. These checks are quite
> horrible, but they worked for me.

I have FW with registers as described at
https://www.buydisplay.com/download/ic/FT5206.pdf
and I would like to add them as attributes more and dt-bindings to
improve trimming.
I would call those registers the same as manual:
THGROUP
THCAL
etc.

And I would also like to add a tasklet that recalibrate every
minute(specified by attr or dt-binding).
I want to do this(already done succesfully on 3.4) because I've
experienced problems on temperature and mechanical drifts.

>
> The best thing is that you ask a datasheet to your supplier, in order to
> make sure about the protocol and avoid guessing.
>
> Regards,
> --
> Luca


--
Giulio Benetti
R&D Manager &
Advanced Research

MICRONOVA SRL
Sede: Via A. Niedda 3 - 35010 Vigonza (PD)
Tel. 049/8931563 - Fax 049/8931346
Cod.Fiscale - P.IVA 02663420285
Capitale Sociale � 26.000 i.v.
Iscritta al Reg. Imprese di Padova N. 02663420285
Numero R.E.A. 258642

From 1584125581413642826@xxx Wed Nov 15 09:58:00 +0000 2017
X-GM-THRID: 1584085398794430736
X-Gmail-Labels: Inbox,Category Forums,HistoricalUnread