Wanhao D7 clipping plate when displayed

Kreeper · 2018-04-14 08:21:46

I have been battling with this for most of the day. If anyone could help that would be amazing.

I have the printer working in all other regards except that the displayed image (e.g. mask generator) is slightly too large for the LCD. It overhangs by around 10mm on the "Right" and about half that on the "Top". I even have the calibration dialled in and get 10mm cubes to within 0.1 mm

My machine setup in nano is standard: 1440x2560, 47.5 XY res, 0 display number, no framebuffer value set.

I have played with overscan settings, including overscan_scale for a while and realized since nano on pi does not use the framebuffer this was a waste of effort.
CURRENT: Switched off.

tried changing hdmi_timings but that did nothing helpful. Thought I might be able to do something with front/back porch.
CURRENT: hdmi_timings=1440 0 50 10 35 2560 0 4 2 2 0 0 0 50 0 198000000 0

Other settings from config.txt:

display_rotate=0x10000
hdmi_group=2
hdmi_mode=87
hdmi_pixel_freq_limit=500000000
hvs_priority=0x32ff
hdmi_drive=2
framebuffer_width=1440
framebuffer_height=2560
max_framebuffer_width=1440
max_framebuffer_height=2560
framebuffer_depth=24
framebuffer_ignore_alpha=1
config_hdmi_boost=4
gpu_mem=196
dtparam=audio=on
disable_camera_led=1

Does anyone have any ideas? It would be VERY much appreciated!

Shahin · 2018-04-14 16:38:13

overscan effect nanodlp too, you need to enable overscan disable_overscan=0

Kreeper · 2018-04-14 19:16:55

Shahin wrote:

overscan effect nanodlp too, you need to enable overscan disable_overscan=0

Hi Shahin, thanks for the reply and all the the hard work in creating Nano.

I should have made it clearer but I have tried almost every possible combination of overscan settings including the suggested. I even outlined the location of some of the mask squares in pen so I could see small changes in their location. I could see no measurable effect - as if it was just ignoring my settings.
I have some experience with the pi over the years as I have had every version of it hooked up to my TV and used it to create a midi controller.

While I could make the squares move occasionally by changing he hdmi_timings, the screen would become unstable and the movements did not appear logical or get me closer to my goal. (I tried over a dozen incremental movements on these, noting down the effect each time before realising it was not doing anything helpful)

This is all very strange.

NB: I am testing only the Mask Generator page. I am assumimg there is nothing special about the mask generator that would ignore overscan..

But you have confirmed that these settings are not a waste of time and implied they should be a solution to my problem which is very helpful in of itself.

I will attack the problem again (having had a good night's rest and cleared my head) with this in mind and report back. It is possible I have typed something wrong somewhere so I will triple check.

Thanks again Shahin!

Last edited by Kreeper (2018-04-14 19:28:39)

Kreeper · 2018-04-14 21:39:39

It appears the config.txt file may have been corrupt. It was not making changes as per yesterday and then windows detected corruption and "fixed" the problem by deleting the file. I recreated it and changed are being made. Happy days....

....or not...

disable_overscan=0
overscan_scale=1
overscan_left=0
overscan_right=5
overscan_top=0
overscan_bottom=0

What on earth!?
I get the same if I set:
overscan_right=2
OR if I set
overscan_top=1

Using:

disable_overscan=0
overscan_scale=1
overscan_left=0
overscan_right=0
overscan_top=0
overscan_bottom=0

Gives

So back to where I was to begin with as you can see from the pen marks.

Will keep going...could this be a corrupt SSD installation? If so, it is a very subtle corruption...

NB: camera is taking mirrored pictures...

Last edited by Kreeper (2018-04-14 21:45:58)

Shahin · 2018-04-14 21:54:05

Could you use and share result of boundary image on projector calibration page?

Kreeper · 2018-04-14 22:06:55

Here is it:

Those horizontal lines worry me...are those supposed to be there?? I thought they were part of the mask!

NB: I have NEVER seen a boundary line since I have had the printer. I read somewhere they were typically too small to see so thought nothing of it.

Those horizontal lines are new....

Last edited by Kreeper (2018-04-14 22:08:03)

Kreeper · 2018-04-14 22:10:57

Dynamic Calibration:

This is very washed out and those damned lines again....

disable_overscan=0
overscan_scale=0
overscan_left=0
overscan_right=5
overscan_top=0
overscan_bottom=0

hdmi_timings=1440 0 50 10 35 2560 0 4 2 2 0 0 0 50 0 198000000 0
hdmi_force_hotplug=1
hdmi_pixel_encoding=2
config_hdmi_boost=4
framebuffer_width=1440
framebuffer_height=2560
max_framebuffer_width=1440
max_framebuffer_height=2560
framebuffer_depth=24
framebuffer_ignore_alpha=1
hdmi_pixel_encoding=2

Last edited by Kreeper (2018-04-14 22:13:16)

Kreeper · 2018-04-14 23:34:14

Ok. Changed the cable for another. Now I have the Dyn Calibration and a mostly black boundary window as I have had in the past. (with that same cable BTW...which came with the printer)

However still not boundary lines (as previous).

But there is a change in the mask...have seen this before when changing settings. No idea why this would be the case...

NB: overscan off for this one.

Still...one issue down....

EDIT:

Spoke too soon.

I changed the overscan_scale from 0 to 1 and I am back to the original problem. It appears the overscan scale is causing the issue at 1, even when overscan is switched off.
I added this after finding it (not explained well as per usual) on forums. I will comment this out completely from now on.

Last edited by Kreeper (2018-04-14 23:43:51)

Kreeper · 2018-04-14 23:50:41

Ok. Once overscan is removed it defaults, as expected, back to 1 which results in original image with light banding.

ok. Time to nuke the system.

Saving machine profiles and reinstalling from scratch.

EDIT:

Nope. Back to the same issue with a fresh reinstall and old machine/config installed.

So not the cable. Not a corrupt install. Must be config.

Last edited by Kreeper (2018-04-15 00:29:10)

Kreeper · 2018-04-15 03:23:35

Ok. Now this has got even stranger.

But I think I have zeroed in on the general area to look at. Not sure if you can help Shahin but is SEEMS that it is coming from NanoDLP rather than RPi or D7.

There are two issues I am trying to fix: one is the overscan but the one I am working on now is that when something is displayed on the screen there is a strange bleed that creates a "fuzzy" horizontal line from one end of the screen to the other where ever there is a pixel anywhere on the screen activated.
On the mask page it looks like bands since the mask squares create bands on the horizontal axis as they "bleed" across the screen.
On the Dynamic Calibration Screen it is a fuzzy haze across the entire screen as there are pixels engaged across the entire horizontal axis. (see images above)

The weird part is that the initial image displayed when it FIRST appears is exactly as it should be, then over the next ~ seconds the haze slowly appears and comes to full strength. Almost as if a naughty elf turns up the exposure/brightness somehow.

Things I have tried and the results and conclusions I have come to:
1) Basic HDMI from another OS
Hook up D7 as a basic external monitor.
Drag a window from the desktop to the D7 so I can see it.
RESULT:
Works fine. No haze or bands on screen. Background is black.
CONCLUSION:
The LCD and hardware are not faulty.

2) Basic HDMI Test from Pi
Make the RPi console appear on the D7.
RESULT:
Works fine, no bleed. Background is black.
CONCLUSION:
The LCD is not faulty and the basic HDMI connection to the PI is not the issue.

3) Another OS with Nano
Installed and configured Nano for windows.
RESULT:
Everything works as expected but the bands and fuzz are present.
CONCLUSION:
RPi is not at fault. The Nano software itself is doing something to the D7 LCD?

4) Another OS and Software Platform
Use CW hooked up from windows.
RESULT:
Due to different firmware to the one CW wants, I have to turn shutter on manually but otherwise the slices are displayed as expected. Background is black, no bands, no fuzz.
CONCLUSION:
It is most certainly something in the way nano is transferring images to the D7.
I have checked the setup/etc page and I simply cannot find a setting that would cause this?
Unless I am going crazy.

NOTES:
I started on the Beta version of Nano and am now using the latest stable release for these latest tests.
Also, I tested it before and after doing a apt-get upgrade.
The RPi firmware has been updated to latest also.

As always. Any help would be VERY much appreciated! Even some pointers as to where to look.

Last edited by Kreeper (2018-04-15 03:28:08)

Kreeper · 2018-04-15 04:26:26

Ok....I think I have it.

This is what I get if I have a plate in the middle of the screen:

But if I move it to the side so it overlaps....

Notice the band is only in a certain spot? No idea why this is...

So I am completely lost now.

It appears that it is not nano, but when a screen overlaps.

I can verify this via windows by dragging windows to either side of the LCD

So I think I need to work on the overscan/overlap issue. Perhaps if I get the LCD locked into position the bleed will vanish.

Took a day to work out, but I would still call that a win! (assuming the overscanning actually works this time....)

I am going to play with overscan and attempt to get the borders locked in. Perhaps this is some sort of bizarre edge bleed.

Kreeper · 2018-04-15 05:20:51

Ok, I am out of ideas.

I have checked all LCD cables and they are tightly and evenly attached. I tried bumping them gently while the screen was on and they were fine.

Tried creating a test plate that overlapped edges but it was fine. It appears to only happen when overlapping windows or using one of the calibration screens. Very annoying but I guess I can work around it.

I only need the mask screen to display one square at a time so I can read it with the photo resister/LDR. this was what I was about to attempt 2 days ago but this issue sort of destroyed that idea! Note that the bands do not appear when there is only one square at a time, even when it is on the edge.

One good point is I now know why my attempts to make a mask using a full white plate and a webcam failed!

I will try to fix the over scan issue now.

EDIT:

It DOES seem to appear with only one square, just much lighter than the others.

Last edited by Kreeper (2018-04-15 06:21:17)

Kreeper · 2018-04-15 21:43:22

Ok. I was trying to get the overscan correct but that is not happening.

Either nothing happens for -ve and some positive values or the screen goes crazy (e.g. full white bars across screen as in photo). No middle ground.
I have no idea and many hours of google searching turns up nothing.

The other issue I am having has been solved.

On the third day and after much research into how LCDs and their control boards fail I have worked out that it is a failed LCD.

Turns out that while LCDs are a digitally controlled device, they can fail in analogue ways that resemble trippy LSD visions.

I plugged in a replacement from Wanhao (love those guys) and the image is not only not ghosting, it is a much crisper image.

Overscan issue is still present though...

Kreeper · 2018-04-16 01:54:51

Ok. Finally zeroing in on the solution. Very slow and frustrating process (and masking still to come) but I can now make changes that show results.

I have tried so many combos of things it is hard to keep track. Have a spreadsheet now so I can be a little more scientific about it. The results are sometimes not so much.

I have noticed that the screen is skewed on one corner, so I think overscan might help here in combination with the porches....assuming it works. This does mean I need to take the measurements from the same square each time.

Nevertheless, slowly getting there...

I know I am talking to no one here so this has become my scientific journal I guess.

Last edited by Kreeper (2018-04-16 02:20:49)

Kreeper · 2018-04-16 07:42:58

While I can get close, no cigar.

Really at a loss what to try next. I contact Wanhao for screen specs and even they are scratching their head and sound like they will not be able to help. They directed me to the forums....
The info on this topic is patchy at best and many people do not really seem to know what they are talking about when it comes to this sort of thing. (a lot of copy and pasting and code/config fragments without information as to why)
After 3 days I think it is time to call it quits and go another route. Hate to admit defeat on anything but at some point you have to admit that you are just wasting time...

Shahin · 2018-04-20 04:32:57

To make sure it is not OS/Application level issue, only use boundary image on the projector page. I am not sure if you have rerun all overscan tries with boundary.

Also SSH into rpi run command below to see real resolution that being used maybe it is different than what you have in the setup.
fbset -s

Kreeper · 2018-04-20 05:10:51

Hi Shahin,

Thanks for the help. I had run those commands. I have also tested this on windows with Nano and CW and the issue is still there. Also it is not the LCD as the old, broken one shows the same issue.
In my testing with the borders I noticed that the border is present on on short end and about 1/4 of one long end of the screen. So the screen is skewed also. I think the dimensions are all wrong also.

I am still in discussion with Wanhao about this for the last few days but so far they have not come up with a solution. I have also emailed a seller of the control boards on AliExpress as they said they have the full tech specs available by email.

My best guess at this point is a bad or misconfigured HDMI controller board. It is even possible this is what caused the first LCD to be destroyed. Cannot say for sure.

Hopefully it is something that simple and they will just send out a new one and it will be fixed. My luck has not been too good in this regard lately...

PS: So sorry for wasting your time!

Last edited by Kreeper (2018-04-20 05:11:55)

sinned6915 · 2019-10-11 18:39:13

Kreeper-

I know it is now 1.5 years after the fact... but how did you make out?

I am wrestling with similar issues.

I appreciate your scientific journaling. I wish more people would post their experiences, good and bad.

sinneD

nanoDLP

#1 2018-04-14 08:21:46

Wanhao D7 clipping plate when displayed

#2 2018-04-14 16:38:13

Re: Wanhao D7 clipping plate when displayed

#3 2018-04-14 19:16:55

Re: Wanhao D7 clipping plate when displayed

#4 2018-04-14 21:39:39

Re: Wanhao D7 clipping plate when displayed

#5 2018-04-14 21:54:05

Re: Wanhao D7 clipping plate when displayed

#6 2018-04-14 22:06:55

Re: Wanhao D7 clipping plate when displayed

#7 2018-04-14 22:10:57

Re: Wanhao D7 clipping plate when displayed

#8 2018-04-14 23:34:14

Re: Wanhao D7 clipping plate when displayed

#9 2018-04-14 23:50:41

Re: Wanhao D7 clipping plate when displayed

#10 2018-04-15 03:23:35

Re: Wanhao D7 clipping plate when displayed

#11 2018-04-15 04:26:26

Re: Wanhao D7 clipping plate when displayed

#12 2018-04-15 05:20:51

Re: Wanhao D7 clipping plate when displayed

#13 2018-04-15 21:43:22

Re: Wanhao D7 clipping plate when displayed

#14 2018-04-16 01:54:51

Re: Wanhao D7 clipping plate when displayed

#15 2018-04-16 07:42:58

Re: Wanhao D7 clipping plate when displayed

#16 2018-04-20 04:32:57

Re: Wanhao D7 clipping plate when displayed

#17 2018-04-20 05:10:51

Re: Wanhao D7 clipping plate when displayed

#18 2019-10-11 18:39:13

Re: Wanhao D7 clipping plate when displayed

Board footer