search

LaserWeb / lw.comm-server

Unified communications server for LaserWeb4 (and other frontends)
GNU General Public License v3.0
38 stars 47 forks source link

Win10/USB3 bug OverLapped Results #39

Open ghost opened 7 years ago

ghost commented 7 years ago

1) Connect to Smoothie 2) Start a Job 3) Click the Feedrate Overides 3.1) Get pissed at 1% being too small an increment (make it +10% please - practically speaking i cant see why anyone would do 1% increments during a job. When shit starts chattering you want to adjust the feed FAST! you dont have time for hundreds of clicks. ) ) 3.2) Lose connection to your machine because of the overlapped write, break endmill, ruin material. Realise its last material you had on hand. fuck

cprezzi commented 7 years ago

You can CTRL-Click to get 10% stepps (help showed on button mouseover). 1% stepps are very usefull to find the sweet spot, especially with laser power.

cprezzi commented 7 years ago

What do you mean with "create overlapped results"?

ghost commented 7 years ago

That works for me. (: (ctrl+click) hover tooltips and touchscreens - thats another challenge though (PiPo, Nexbook, Chuwi Netbooks)

On Jun 27, 2017 6:53 PM, "Claudio Prezzi" notifications@github.com wrote:

You can CTRL-Click to get 10% stepps (help showed on button mousover). 1% stepps are very usefull to find the sweet spot, especially with laser power.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/LaserWeb/lw.comm-server/issues/39#issuecomment-311418917, or mute the thread https://github.com/notifications/unsubscribe-auth/AHVr2xtMOGnt3xa0kh92yKPgQk4LKpipks5sITONgaJpZM4OGleM .

ghost commented 7 years ago

"What do you mean with "create overlapped results"?": Total Guesswork: I think it happens when node-serialport is trying to write two streams to the port (queue and realtime)

https://github.com/EmergingTechnologyAdvisors/node-serialport/blob/master/src/serialport_win.cpp#L277

ghost commented 7 years ago

PS: on my faster desktop it doesnt happen often. but on my cheap, slow, https://www.walmart.com/ip/Nextbook-Flexx-10.1-2-in-1-Tablet-32GB-Intel-Atom-Z3735F-Quad-Core-Processor-Windows-10/46664255 windows 10 netbook, it happens almost every time i try using override, but always after a couple clicks, 10+

ghost commented 7 years ago

Or it could be that win7 works better, and win10 doesnt... dont have another win10 to test against

cprezzi commented 7 years ago

I work with Win10 but never got this error related to overrides or other "real time" commands. It just happend sometimes mid job with smoothieware, but not anymore with grbl-LPC. Probably has something to do with different USB protocolls.

ghost commented 7 years ago

Okay, still happens on the new Asus laptop, so ruling out slower CPU being the main culprit. Did however notice it happening easier during short segment moves. Mentally i hold off feedrate adjustments till its about to do longer sections.

Error code is 121

ghost commented 7 years ago

We have just had a few reports from users that they are getting the overlapped results error as well. twoxlje0exs6lnr4rm0baa

Users are running 'Windows 10 home version 1607' Is there any further development on this? Is there any information that would be useful fur us to provide to help resolve this?

ghost commented 7 years ago

@domenic-d , also happened when using feed/power overrides?

ghost commented 7 years ago

I checked with them after your feedback and no users reported making any changes to feed/power overrides.

ghost commented 7 years ago

Find out EXACTLY when / why they got the error. It shouldnt happen during normal operations (as during normal operation we run just the single file queue (out, ok, out, ok). Its during overrides that an extra line of commands are dumped onto the port...

ghost commented 7 years ago

(Maybe they dont know the up/down buttons on the DRO are called "feedrate override" and "S override" - so they report they didnt because they dont know what they clicked>?

cprezzi commented 7 years ago

According to the screenshot, it's not a firmware-cnc version. We need firmware-cnc! And what are those two Ellapsed lines? Is this a new special smoothie firmware?

ghost commented 7 years ago

@cprezzi This is the current version of Smoothiware we use. There is nothing special about it. What classifies it as smoothiware-cnc on your system?

cprezzi commented 7 years ago

Smoothieware should send a second line CNC Build 3 axis after the first line, but probably this is only visible at the server console. I will download and test the latest smoothieware later today.

cprezzi commented 7 years ago

This is how it should look like when using the right firmware-cnc.bin:

grafik

cprezzi commented 7 years ago

Please advise they should use this firmware: https://github.com/Smoothieware/Smoothieware/blob/edge/FirmwareBin/firmware-cnc.bin

cprezzi commented 7 years ago

@domenic-d Can you provide a file and the settings which caused the OverlapedResult error?

cprezzi commented 7 years ago

I was able to replicate the OverlappedResult error when using a cheap 1.8m USB cable, while it worked flawlessly with a good quality 1.8m cable or a standard blue 0.5m cable (without ferrites)!

All this was tested with the same front panel USB port on a desktop PC with Windows 10 and my C3D Remix.

Then I measured the USB voltage on the board side:

Changed to a rear USB port (onboard) I got 4.16V with the cheap 1.8m cable and the communication was stable.

It seems that not all USB ports provide the correct 5V and bad cables decrease that even more. If the voltage on the board dops below ~ 4.2V, the serial communication (and whatever else) seems to be unstable.

ghost commented 7 years ago

Thats why my boards, and Domenic's custom boards, neither rely on the USB bus voltage. We have 5v regulators onboard (;

On Jul 11, 2017 12:52 PM, "Claudio Prezzi" notifications@github.com wrote:

I was able to replicate the OverlappedResult error when using a cheap 1.8m USB cable, while it worked flawlessly with a good quality 1.8m cable or a standard blue 0.5m cable (without ferrites)!

All this was tested with the same front panel USB port on a desktop PC with Windows 10 and my C3D Remix.

Then I measured the USB voltage on the board side:

  • Standard blue 0.5m cable: 4.3V
  • Good 1.8m cable: 4.3V
  • Cheap 1.8m cable: 4.0V !

Changed to a rear USB port (onboard) I got 4.16V with the cheap 1.8m cable and the communication was stable.

It seems that not all USB ports provide the correct 5V and bad cables decrease that even more. If the voltage on the board dops below ~ 4.2V, the serial communication (and whatever else) seems to be unstable.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/LaserWeb/lw.comm-server/issues/39#issuecomment-314407829, or mute the thread https://github.com/notifications/unsubscribe-auth/AHVr293Y0M2IgAtAZXbnfVtBd7DNGLFgks5sM1PhgaJpZM4OGleM .

cprezzi commented 7 years ago

Unfortunately most users with Smoothie or C3D boards do not have a 5V supply.

ghost commented 7 years ago

@cprezzi @openhardwarecoza I am asking for more information. I'll pass it on as soon as I have it.

ghost commented 7 years ago

Indeed... But in the case of me and Domenic, the issue is not supply voltage

On Jul 11, 2017 12:56 PM, "Claudio Prezzi" notifications@github.com wrote:

Unfortunately most users with Smoothie or C3D boards do not have a 5V supply.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/LaserWeb/lw.comm-server/issues/39#issuecomment-314408604, or mute the thread https://github.com/notifications/unsubscribe-auth/AHVr2xKuT01F7ZBqXQsc0e2U7Bz71Pwbks5sM1TFgaJpZM4OGleM .

ghost commented 7 years ago

C3D should also have proper supply. Its based off my Smoothiebrainz where i already added the psu. Ray just swopped the LM2596 for a smaller XL1509. I am using the XL1509 on my newer boards too.

On Jul 11, 2017 1:57 PM, "Peter van der Walt (Gmail)" < peter.plaaswerf@gmail.com> wrote:

Indeed... But in the case of me and Domenic, the issue is not supply voltage

On Jul 11, 2017 12:56 PM, "Claudio Prezzi" notifications@github.com wrote:

Unfortunately most users with Smoothie or C3D boards do not have a 5V supply.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/LaserWeb/lw.comm-server/issues/39#issuecomment-314408604, or mute the thread https://github.com/notifications/unsubscribe-auth/AHVr2xKuT01F7ZBqXQsc0e2U7Bz71Pwbks5sM1TFgaJpZM4OGleM .

cprezzi commented 7 years ago

Ok. I was testing with the Cohesion Remix but without external power connected. If I connect the external 12V supply, the board voltage increases to 4.8V and it works also with the cheap 1.8m cable.

ghost commented 7 years ago

We have been able to replicate 'our' specific problem on our machines in the lab. It is caused by plugging the smoothieboard into a USB3 or SSUSB port. After a short period of time, the firmware crashes and these errors are reported. It is not related to Laserweb.

We will take this up with the Smoothie team.

ghost commented 7 years ago

Hmmm, interesting observation. I'll check if the old Asus i stole from my wife (which she got new like 5 years ago) even has USB3 (; - i think its too old (;

On Wed, Jul 12, 2017 at 3:34 AM, Domenic Di Giorgio < notifications@github.com> wrote:

We have been able to replicate 'our' specific problem on our machines in the lab. It is caused by plugging the smoothieboard into a USB3 or SSUSB port. After a short period of time, the firmware crashes and these errors are reported. It is not related to Laserweb.

We will take this up with the Smoothie team.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/LaserWeb/lw.comm-server/issues/39#issuecomment-314619049, or mute the thread https://github.com/notifications/unsubscribe-auth/AHVr27qiMqNfxVmVPc4zQ6FwSYr7-kv2ks5sNCKXgaJpZM4OGleM .

cprezzi commented 7 years ago

My PC also hasn't USB3, but one of the ports is combined USB/eSATA, and this port doesn't work with the Remix board.

ghost commented 7 years ago

It is an unusual problem which we have been testing all day. We have tested the latest smoothiware firmware (both cnc and reprap). Both behave the same, with the firmware crash. This all became apparent with the new laptop we purchased for our QA workstation. None of our other older computers can produce this problem.

ghost commented 7 years ago

I am outside in the workshop now. Checked and low and behold this laptop does have usb3. I am not sure if thats where i was plugged into... Let me try deliberately using a usb2 port this afternoon (got some lasering todo for the morning and thats on a Grbl based board. Will hook up the mill (which runs smoothie) this afternoon and compare. Looks like you are maybe onto something Dom!

On Jul 12, 2017 9:39 AM, "Domenic Di Giorgio" notifications@github.com wrote:

It is an unusual problem which we have been testing all day. We have tested the latest smoothiware firmware (both cnc and reprap). Both behave the same, with the firmware crash. This all became apparent with the new laptop we purchased for our QA workstation. None of our other older computers can produce this problem.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/LaserWeb/lw.comm-server/issues/39#issuecomment-314682135, or mute the thread https://github.com/notifications/unsubscribe-auth/AHVr24XZisHCqdszaDYAIrTu7rjtbqyiks5sNHg5gaJpZM4OGleM .

ghost commented 7 years ago

Okay. So running on the usb2 port for the last couple days have indeed proven to be more stable. But not eliminated the issue. Now, i still get getOverlapped-121 errors, but, they happen when i have feedrate override over 180% and, then it happens randomly mid job (not only on fro clicks). Looking more and more like usb driver. Might try grbl-lpc to see