mouse example

[ivo1981]

Hi, Can you provide an example on how i can catch data from a connected mouse?
the mouse gets detected, but nothing more then that. if I plug in a keyboard i at least get raw data

[tobozo]

hey @ivo1981 thanks for your feedback

see how it was detected in issue #4, the cable length/quality, mouse type and wiring really matter to get a clear signal

[ivo1981]

but i shouldn't have to make any changes to the code itself to at least get data.
the cable is 1m long the chip it uses is a A2803 (just for future refference)

[tobozo]

I can't really tell since I haven't fully tested a mouse myself, but this library is quite close to the esp-idf example so it should at least work as in the example and report the mouse events after it is detected.

@zbx-sadman did you have to modify the code to get the mouse events ?

[ivo1981]

so i did some testing; i cut the wires, replaced it with a high quality one, soldered it directly to the pins and pcb, length was 6cm. No joy....
also tried an HP laser mouse, instant data.
now to find out what the issue is, as the mouse works fine on any os I tried (ubuntu 20, win10, osx)
and has completely USB HID ver1.1 compatibility. It is compatible with Microsoft 3D IntelliMouse.
mouse was brand new by the way. i don't think it's an issue with the mouse, but how it's handled by the software

[tobozo]

A few changes had to be done in order to get the original library to work with esp-idf 3.3, and it may have created more failure opportunities indeed.

If you're positive it works fine with the original library, then I can look into it, but I don't have a fully working example for myself (most mouses I tried failed too) so I'm not sure where to begin.

[zbx-sadman]

No, @tobozo , I do not maked any changes with code, only enable debug in the sdima's sources.

When i was connect USB data wires to the ESP directly, i saw data.

[zbx-sadman]

So strange.

I tried just now my setup again with yhe same code and Logitech K120 keyboard & B100 mouse and see mouse detection report only (keyboard is OK - report & data present):

Screenshoot from issue #4 maked when this mouse was connected: ASUS Optical USB Black #M-UV55A . It's like that:

I will take back this short-cabled mouse ~ May 11 and re-test connection.

[tobozo]

maybe a noise filter could help ?

[edit] both the mouse and the ESP32 serial cable would need it

[ivo1981]

honestly i dont think a noise is the issue; if it was the mouse also wouldn't work on a pc. at least not with any accuracy as it would be skipping beats where the data is corrupted. also using a 6cm (thats like 2 inches if you're state side) high quality shielded cable would definitly do the trick; especially considering it's clean power going in already.
I'm pretty confident we are overlooking something on the software side.
I'm not saying it's because of the arduino port specificaly as i haven't tested it yet with the toolchain, i went the arduino route so I didn't have to install it.
I will try to free up some more time so i can setup the toolchain and try with that. we should at least get some idea of the origin of the issue that way.
I think it's an issue with catching the data, but we'll see

[tobozo]

some ideas for the software side:

• Maybe playing with the task priority/core can improve things ?

  ESP32-USB-Soft-Host/src/ESP32-USBSoftHost.hpp

  Line 86 in 49098bb

  xTaskCreatePinnedToCore(USB_SOFT_HOST::TimerTask, "USB Soft Host Timer Task", 4096, NULL, 5, NULL, 1);

• Some USB-LS ports on the ESP can give better results depending on the HID device, so it may be a good idea to try them all

• Once the best pins are identified, disabling the unused ports can save some cycles

  #define DP_P0  16  // always enabled
  #define DM_P0  17  // always enabled
  #define DP_P1 18 // -1 to disable
  #define DM_P1  19 // -1 to disable
  #define DP_P2  -1 // -1 to disable
  #define DM_P2  -1 // -1 to disable
  #define DP_P3  -1 // -1 to disable
  #define DM_P3  -1 // -1 to disable

[zbx-sadman]

A few more information...

Just now i has some play with oscilloscope and level shifter (have no luck again). But i catched my Logitech B100 when it being works. Mouse is powered from the ESP32 board with 5V.

I also noticed that on "live connection" mouse's LED is on constantly, and on "dead connection" just blinks on connection and turned off. May be it's power problem? Who knows...

Sketch "config":

#define DP_P0  18  // always enabled
#define DM_P0  19  // always enabled
#define DP_P1  -1 // -1 to disable
#define DM_P1  -1 // -1 to disable
#define DP_P2  -1 // -1 to disable
#define DM_P2  -1 // -1 to disable
#define DP_P3  -1 // -1 to disable
#define DM_P3  -1 // -1 to disable

Below is dumps with uncomments some driver's debug lines.

Case N1, "dead mouse connection":

desc.bcdUSB          = 200
desc.bDeviceClass    = 00
desc.bDeviceSubClass = 00
desc.bDeviceProtocol = 00
desc.bMaxPacketSize0 = 08
desc.idVendor        = 46d
desc.idProduct       = c077
desc.bcdDevice       = 7200
desc.iManufacturer   = 01
desc.iProduct        = 02
desc.iSerialNumber   = 00
desc.bNumConfigurations = 01

Case N2, "live mouse connection":

desc.bcdUSB          = 200
desc.bDeviceClass    = 00
desc.bDeviceSubClass = 00
desc.bDeviceProtocol = 00
desc.bMaxPacketSize0 = 08
desc.idVendor        = 46d
desc.idProduct       = c077
desc.bcdDevice       = 7200
desc.iManufacturer   = 01
desc.iProduct        = 02
desc.iSerialNumber   = 00
desc.bNumConfigurations = 01

cfg.bLength         = 09
cfg.bType           = 02
cfg.wLength         = 22
cfg.bNumIntf        = 01
cfg.bCV             = 01
cfg.bIndex          = 00
cfg.bAttr           = a0
cfg.bMaxPower       = 50

sIntf.bLength      = 09
sIntf.bType        = 04
sIntf.iNum         = 00
sIntf.iAltString   = 00
sIntf.bEndPoints   = 01
sIntf.iClass       = 03
sIntf.iSub         = 01
sIntf.iProto       = 2
sIntf.iIndex       = 0

hid.bLength          = 09
hid.bDescriptorType  = 21
hid.bcdHID           = 111
hid.bCountryCode     = 00
hid.bNumDescriptors  = 01
hid.bReportDescriptorType = 22
hid.wItemLengthH         = 00
hid.wItemLengthL         = 2e

pcurrent->epCount = 1
epd.bEPAdd        = 81
epd.bAttr         = 03
epd.wPayLoad      = 04
epd.bInterval     = 0a

in :00 38 f1 00 
in :00 7f ce 00 
in :00 06 fe 00 
in :00 01 01 00 
in :00 00 01 00 
in :00 e6 00 00 

Case N3, "live keyboard connection":

desc.bcdUSB          = 110
desc.bDeviceClass    = 00
desc.bDeviceSubClass = 00
desc.bDeviceProtocol = 00
desc.bMaxPacketSize0 = 08
desc.idVendor        = 46d
desc.idProduct       = c31c
desc.bcdDevice       = 6400
desc.iManufacturer   = 01
desc.iProduct        = 02
desc.iSerialNumber   = 00
desc.bNumConfigurations = 01

cfg.bLength         = 09
cfg.bType           = 02
cfg.wLength         = 3b
cfg.bNumIntf        = 02
cfg.bCV             = 01
cfg.bIndex          = 03
cfg.bAttr           = a0
cfg.bMaxPower       = 45

sIntf.bLength      = 09
sIntf.bType        = 04
sIntf.iNum         = 00
sIntf.iAltString   = 00
sIntf.bEndPoints   = 01
sIntf.iClass       = 03
sIntf.iSub         = 01
sIntf.iProto       = 1
sIntf.iIndex       = 2

hid.bLength          = 09
hid.bDescriptorType  = 21
hid.bcdHID           = 110
hid.bCountryCode     = 00
hid.bNumDescriptors  = 01
hid.bReportDescriptorType = 22
hid.wItemLengthH         = 00
hid.wItemLengthL         = 41

pcurrent->epCount = 1
epd.bEPAdd        = 81
epd.bAttr         = 03
epd.wPayLoad      = 08
epd.bInterval     = 0a

sIntf.bLength      = 09
sIntf.bType        = 04
sIntf.iNum         = 01
sIntf.iAltString   = 00
sIntf.bEndPoints   = 01
sIntf.iClass       = 03
sIntf.iSub         = 00
sIntf.iProto       = 0
sIntf.iIndex       = 2

hid.bLength          = 09
hid.bDescriptorType  = 21
hid.bcdHID           = 110
hid.bCountryCode     = 00
hid.bNumDescriptors  = 01
hid.bReportDescriptorType = 22
hid.wItemLengthH         = 00
hid.wItemLengthL         = 9f

pcurrent->epCount = 2
epd.bEPAdd        = 82
epd.bAttr         = 03
epd.wPayLoad      = 04
epd.bInterval     = ff


in :00 00 51 00 00 00 00 00 
in :00 00 00 00 00 00 00 00 

Mouse connected:

P.S. I've add big capacitor on mouse power line - no any changes. It detects in 10-25% of all connection tries.

[tobozo]

if the ESP32 cannot provide enough power, you can add another 5v power source to your mouse/keyboard, then disconnect the VCC between between the ESP and the USB HID devices (only keep GND, Din, Dout)

you can eventually use an external battery + regulator, but I prefer picking 5v from another port of the same PC using a DIY cable

[zbx-sadman]

@tobozo i downloaded latest sources, trying fork task on Core 0, increase memory and maximize priority: xTaskCreatePinnedToCore(USB_SOFT_HOST::TimerTask, "USB Soft Host Timer Task", 20480, NULL, configMAX_PRIORITIES, NULL, 0);

No changes.

I've powered ESP32 from USB-port (500mA), mouse must take 100mA, and i connected capacitor 2200uF. I think this must be enough...

May be perepheral device sometime take more time for init and internal process do not take it in account?

P.S. I used connected mouse periodically within 4 hours. HID reports received w/o errors, ESP no hanged. Seems that problem is happens on init stage...

[ivo1981]

i did some research on the different all in one optical mouse packages and went over several data sheets, it can take up to 10ms for the init to complete.
What i find interesting is that @zbx-sadman reports a 10-25% success rate; that makes me think we are either to slow or to quick with handeling the init.
I guess i'll be desoldering my super short cable mouse as I think we can rule out it's cables causing the issue.

[tobozo]

thanks for this research

I seem to have missed some changes from usb_soft_host repository.

So I did some integration+testing, somehow hoping that a better calibration at init would solve the mouse events detection problem, but I can't observe any improvement.

Since it doesn't seem to break anything either, I've pushed those changes on the main branch.

@ivo1981 this kind of change could eventually impact the software side your were mentioning earlier, but my guess is it probably won't.