OT: CRT timing question

109 views
Skip to first unread message

jb-electronics

unread,
Feb 20, 2022, 8:26:05 PM2/20/22
to neoni...@googlegroups.com

Dear all,

Apologies for the off topic question, but I know there are some CRT experts in our group and I have been banging my head on this one for a while.

I picked up a cute 6" monochrome CRT module which accepts TTL video. This is the timing chart:


My goal is to run this thing using the DPI module of the Raspberry Pi. But for all of that to work I have to specify certain parameters and pass them along to the DPI module, and here is what I read off the sheet:

  • h sync: 4.615 us
  • h front porch: 2.637 us
  • h back porch: 2.637 us
  • v sync: 0.208 ms
  • v front porch: 0.209 ms
  • v back porch: 2.917 ms
  • active lines: 256

I have verified that at least the vertical signal timing is implemented correctly, my scope is a tad too slow to resolve the horizontal timing. But there appears only gibberish on the screen, highly distorted text. Here is how it looks like: http://jb-electronics.de/tmp/screen01.jpg

And this is what it should be (composite video output on my Apple IIc monitor) for the same "scene": http://jb-electronics.de/tmp/screen02.jpg

Can somebody tell me if I am on the right track here, or if there is a fundamental mistake somewhere in the numbers I read off? Your help is very much appreciated.

Best wishes
Jens

David Forbes

unread,
Feb 20, 2022, 8:58:19 PM2/20/22
to NeoNixie
Jens,
The horizontal line timing is incorrect. This is why the image is diagonal. 
The numbers that you provided do not specify the horizontal line total time or frequency. 
The monitor requires 52 microseconds or 19.2 kHz.
You need to find a way to specify this, perhaps it is the character count or pixel count and the pixel frequency.




--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/5a549b7b-127f-42c3-3dde-4973892d9780%40jb-electronics.de.

Nick Sargeant

unread,
Feb 20, 2022, 10:52:54 PM2/20/22
to neonixie-l
Comparing the timings in the picture and going for common denominators, you can see that the front porch and back porch are  (4 units of 660 ns) long, hsync is (7 units of 660ns) long, video itself is (64 units of 660ns) long. So I would offer that there is a 64 character display, where each character is made up of ?6 pixels. So, if you use a pixel rate of 110ns, then use those numbers for the parameters of dpi timings, and you should get there. Front porch, back porch are 24 pixels each, hsync is 42 pixels, active pixels is 384, ...

jb-electronics

unread,
Feb 21, 2022, 5:44:37 PM2/21/22
to neoni...@googlegroups.com
David,

Thanks for your pointers! In the Raspberry Pi's DPI configuration I set the amount of horizontal pixels to 844px, which at a pixel clock of 20MHz should result in 19.2kHz if I am not mistaken. Sorry, I forgot to mention this part.

Regardless of how I think I configured it, you are of course right that it looks like there is something wrong with the horizontal timing. But the absoltue value of the pixel clock should not play any role here, should it, assuming that I calculated the correct horizontal timing?

Best wishes
Jens

jb-electronics

unread,
Feb 21, 2022, 5:47:34 PM2/21/22
to neoni...@googlegroups.com
Nick, nice decoding of the numbers there. I can set the pixel clock to 110ns per pixel (9.09 MHz), but since this is a free parameter in the Raspberry Pi it should also work with the 20MHz that I originally specified, right?

Best wishes
Jens

David Forbes

unread,
Feb 21, 2022, 8:00:03 PM2/21/22
to NeoNixie
Jens,
Connect your oscilloscope to the H Drive signal and verify that it is 19.2 kHz. If your oscilloscope cannot do this job, then do yourself a favor and buy a better one. It will be worth it.
I say this because I did a lot of video work in the 1980s, and a Tektronix 465 scope or its equivalent was essential for this work.
You should also be able to see the video on the oscilloscope. It may not be displayed in full resolution, but the text portion should be a higher voltage than the black portions at the sides of the screen. 
You can change the settings in the Pi and you should be able to see the signal timing change to follow the settings. If you cannot do this, it will never work correctly.
Good luck!



jb-electronics

unread,
Feb 25, 2022, 11:34:44 PM2/25/22
to neoni...@googlegroups.com
Hey David, I had a few minutes spare time tonight and I verified that both the hsync and vsync signals appear at exactly the right frequency (19.2kHz and 60Hz, respectively) and have the correct duration. Yet, the image is still garbage. How can that be, any ideas?

Jens

jb-electronics

unread,
Feb 25, 2022, 11:59:45 PM2/25/22
to neoni...@googlegroups.com
PS: Also the video data's timing is spot on. Silly question: The Pi runs at 3.3V, the display expects TTL (5V). Could this lead to these distortions? I have not yet added a proper level shifter.

David Forbes

unread,
Feb 26, 2022, 1:10:42 AM2/26/22
to NeoNixie
Jens,
The CRT monitor may not be adjusted properly. The horizontal oscillator probably has a frequency control. This what we called Horizontal Hold on TV sets. 
Do you have any documentation such as a schematic, service manual or similar for it?

Alex

unread,
Feb 26, 2022, 5:21:29 AM2/26/22
to neonixie-l
I get the feeling I have a very similar crt in my collection, but I also think I mentioned this a while back on here, have you posted already about this? I think mine is either Toshiba or Panasonic and was a unit often used in T&M kit, Tektronix protocol analysers come to mind? I had mine running from a pic and iirc it a) needed an inverted sync pulse to convention and likely also needed 5v p-p signals?

Hope you get it working, it's a nice little CRT!

jb-electronics

unread,
Feb 26, 2022, 9:38:38 AM2/26/22
to neoni...@googlegroups.com

To add to the mystery, I should say this: I managed to get the display to work with a composite signal and a GS1881 sync separator to extract the hsync and vsync pulses. I know that the timing is a bit off, but by tweaking the v-hold and h-hold potentiometers I goit a stable image, but it was seriously underscanned (i.e. did only cover a small part of the screen). So the monitor works well, I believe.

Jens
--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

jb-electronics

unread,
Feb 26, 2022, 9:40:39 AM2/26/22
to neoni...@googlegroups.com
Yeah, I posted about this a while ago, but just now discovered the Raspberry Pi's DPI module. The timing chart [ http://jb-electronics.de/tmp/timing.jpg ] does not mention inverted pulses, though. Do you have any intuition what could go wrong if you use 3.3V p-p signals instead? Could this explain the weird image?

Jens
--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages