RGB to HDMI using a Pi Zero and a small CPLD

discuss both original and modern hardware for the bbc micro/electron
User avatar
hoglet
Posts: 9287
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by hoglet » Mon Jun 22, 2020 10:07 am

*STEVE wrote:
Mon Jun 22, 2020 2:12 am
So, I've built one up using parts I had available, the XC9572XL dev board in the original jumper-wired version and proto-board, but to the latest spec schematic (rgb-to-hdmi.shc and using the front-end from the three-bit-buffer (buffer.sch). I've programmed the CPLD sucessfully (to the newer 6-bit version since i've built wired this up to the 6-bit schematic) but am seeing two things when it boots. One, it's saying 'no sync detected' although there is on the SYNC line (buffered through the 74LS08) and the interface is stuck at '6 BIT RGB', although I think this should reflect 3-BIT TTL.

I suspect this is just a wiring error and the PI thinks I'm using the 6 bit analogue interface when I want to use the 3-Bit TTL.. Q1: how is the interface type detected?
I'm sure Ian will correct me if I get anything wrong here (I've not tried one of his buffer boards myself yet).

I think all the detection is working correctly, and an interface type of 6_BIT_RGB is what I would expect.

The software needs to detect the presence of the analog interface, because it contains DACs that need programming over an SPI interface. It does this by looking at the level on the DETECT signal (CPLD pin 19). This is pulled down on the CPLD board by a weak (10K) pulldown, and pulled up on the analog board by a stronger (1K) pullup. So when the analog interface is fitted, DETECT is read back as high.

However, I your case you have effectively built one of the buffer boards, which is entirely transparent to the software.

In this case, the "Interface" value shown by the software simply reflects the CPLD design programmed, and as you have programmed the 6-bit RGB CPLD, it's showing up as "6 BIT RGB". This is exactly what I would expect, so I think the problem lies elsewhere.

I also should say the the 6-bit RGB CPLD is capable of handling both 3-bit and 6-bit inputs, so it's the right version to have programmed.
*STEVE wrote:
Mon Jun 22, 2020 2:12 am
I my 3-bit buffered RGB inputs go to RED, GREEN, BLUE and I've left BRED BGREEN, BBLUE not-connected.
This sounds right to me.

The wiring details for various systems are detailed here:
https://github.com/hoglet67/RGBtoHDMI/wiki/Cables

What type of computer system are you trying to connect?

What profile have you selected?

I think the most likely scenario is you have a wiring error somewhere relating to SYNC.

The composite sync signal should go to pin 23 of the CPLD (marked SYNC). You should also then see CSYNC come back out of the CPLD on pin 20, which should connect to GPIO23 on the Pi.

Can you try to follow this through all the way to the Pi?

Can you post a couple of photos of your construction?
*STEVE wrote:
Mon Jun 22, 2020 2:12 am
Q2: Are the 4K7 pull-down's necessary?
Not really, they just provide a small amount of termination.

In fact, if it's a Beeb that is driving this interface, that's quite a low impedence source, so you might benefit from much lower value resistors (e.g. 220R). But all the original versions of the design didn't have any termimation at all, and worked fine.

Dave

retrorgb
Posts: 2
Joined: Mon Jun 22, 2020 3:29 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by retrorgb » Mon Jun 22, 2020 3:51 pm

I'm just now stumbling across this project and it looks absolutely amazing! I tried reading through as much of the github as I could, so please excuse me if some of these questions have already been answered:

1) Would this work for game consoles too? It seems like it would work with any RGBs source, but I wanted to confirm it'll work with consoles like SNES, Genesis/Mega Drive, etc.

2) Is it possible that it would work with a Raspberry Pi 4? In most cases, 1080p scaling is fine, however I think scanline filters can start do truly mimic the full mask (horizontal and vertical lines) of CRT's in 4K resolution. It would be pretty incredible to have 9x & 10x integer scaling options with scanlines.

I'd be happy to build a board and test if you think it'll work. Thanks again for this awesome project.

User avatar
hoglet
Posts: 9287
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by hoglet » Mon Jun 22, 2020 4:12 pm

retrorgb wrote:
Mon Jun 22, 2020 3:51 pm
1) Would this work for game consoles too? It seems like it would work with any RGBs source, but I wanted to confirm it'll work with consoles like SNES, Genesis/Mega Drive, etc.
No, as these platforms have far too many discrete colours.

For more context, please my answer to the same question a couple of posts up.
retrorgb wrote:
Mon Jun 22, 2020 3:51 pm
2) Is it possible that it would work with a Raspberry Pi 4? In most cases, 1080p scaling is fine, however I think scanline filters can start do truly mimic the full mask (horizontal and vertical lines) of CRT's in 4K resolution. It would be pretty incredible to have 9x & 10x integer scaling options with scanlines.
The only hardware we officially support is a Pi Zero.

During development we have occasionally tested on other Pi models, but there are issues, and fixing these is not a priority. If you want to dig further into the kind of issue, see this thread for example:
https://github.com/hoglet67/RGBtoHDMI/pull/135

Dave

retrorgb
Posts: 2
Joined: Mon Jun 22, 2020 3:29 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by retrorgb » Mon Jun 22, 2020 4:25 pm

Thanks very much for the reply. I'll check into all of that.

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Mon Jun 22, 2020 9:32 pm

Hi Dave,

Some success...
The composite sync signal should go to pin 23 of the CPLD (marked SYNC). You should also then see CSYNC come back out of the CPLD on pin 20, which should connect to GPIO23 on the Pi.
...I'd made an error at GPIO23 and connected to the pin above (GND). Fortunately the CPLD has survived the short to ground, and this is now working except the blue input seems to be stuck high. Just working on that now, but thanks for the help so far :) Will post some images soon.

Steve.
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Tue Jun 23, 2020 2:16 am

More success...

Turns out my issue with the blue line being stuck high was a fine slither of copper along a score line on the crappy eBay proto-board I was using. The minor score was on the board as delivered but I never anticipated it shorting across 4 pads.
20200622_132553.jpg
Blue stuck high
20200622_151856.jpg
Shorted pads
The two rows of header down the centre of the image is for the PI. The unconnected pad is at 3V3 and was linked with a fine (invisible to my eyes but continuity when buzzed) slither to pin 1 of the 74LS08 sticking the BLUE input to 3V3 (input wire to pins 1 & 2 desoldered here).

With that fixed things are now working (mostly). If you take a look at the video in the link to my google photos share you can see noise (speckles) in the black but only adjacent to white areas on the image. https://photos.app.goo.gl/r927MDkpbLhVPPSn8

I'm wondering if this is a combination of impedance in the power supply lines possibly with inadequate decoupling (aside from decoupling on the Xilinx and Pi boards I added only 100nF around the 74LS08 and nothing else). Right now I'm powering everything from an external supply but will switch to the Beeb power once things are stable. You can see my prototype in the images below, I also imagine there's a lot of stray capacitance in the wiring, but feel it's likely a power/GND/decoupling issue since only present when R, G and B are high... at least that's my theory.
20200622_163917.jpg
Proto wiring side
20200622_163854.jpg
Proto topside
If anyone has seen this speckling before let me know. Once again, thanks to @Hoglet for your confirming that I'd got my design right.

Steve.

EDIT: attachments are working - the problem was the user #-o
Last edited by *STEVE on Tue Jun 23, 2020 3:47 am, edited 1 time in total.
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

User avatar
richardtoohey
Posts: 3933
Joined: Thu Dec 29, 2011 5:13 am
Location: Tauranga, New Zealand
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by richardtoohey » Tue Jun 23, 2020 2:27 am

*STEVE wrote:
Tue Jun 23, 2020 2:16 am
I can't seem to add attachments - do I need access granting on a per forum / per post basis?
Not that I know of. What sort of attachments are you trying to use (JPG, PNG, ZIP, whatever?) and what is stopping you? i.e. does it just not work or are you getting some sort of error message (if so, what?)

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Tue Jun 23, 2020 3:50 am

richardtoohey wrote:
Tue Jun 23, 2020 2:27 am
*STEVE wrote:
Tue Jun 23, 2020 2:16 am
I can't seem to add attachments - do I need access granting on a per forum / per post basis?
Not that I know of. What sort of attachments are you trying to use (JPG, PNG, ZIP, whatever?) and what is stopping you? i.e. does it just not work or are you getting some sort of error message (if so, what?)
Apologies Richard, it's all fine... just user trouble #-o

Steve
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

User avatar
hoglet
Posts: 9287
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by hoglet » Tue Jun 23, 2020 9:52 am

*STEVE wrote:
Tue Jun 23, 2020 2:16 am
If anyone has seen this speckling before let me know. Once again, thanks to @Hoglet for your confirming that I'd got my design right.
These are almost certainly down to the longish wires, and lack of a decent ground plane. It looks like you are only getting them on white->black, which suggests to me is a poor ground on the 74LS08.

I assume you've tried the auto calibration button (press and hold button 3 for a second), and this doesn't help?

You could try adding 220R termination resistors to the R, G and B signals (to ground) where they enter the board.

You could also experiment with replacing the 74LS08 with a difference type, possibly a 74HC08 would have a slightly higher threshold.

Dave

User avatar
IanB
Posts: 582
Joined: Sun Sep 04, 2011 8:28 pm
Location: South Wales
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by IanB » Tue Jun 23, 2020 1:26 pm

*STEVE wrote:
Tue Jun 23, 2020 2:16 am
If anyone has seen this speckling before let me know. Once again, thanks to @Hoglet for your confirming that I'd got my design right.
hoglet wrote:
Tue Jun 23, 2020 9:52 am
These are almost certainly down to the longish wires, and lack of a decent ground plane. It looks like you are only getting them on white->black, which suggests to me is a poor ground on the 74LS08.
You could try adding 220R termination resistors to the R, G and B signals (to ground) where they enter the board.
I agree with Dave, you could also try adding pull downs at the RGB inputs to the CPLD board.
*STEVE wrote:
Mon Jun 22, 2020 2:12 am
I appreciate this is a bit of a franken-version but trying to use this to check performance before investing in getting PCBs manufactured.
I still have blank pcbs available if you're interested.

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Wed Jun 24, 2020 2:41 am

Thanks for all the suggestions...

First of all, yes I did the auto calibration - before this there was noise at the edges of characters which is a different appearance to these speckles.
hoglet wrote:
Tue Jun 23, 2020 9:52 am
These are almost certainly down to the longish wires, and lack of a decent ground plane. It looks like you are only getting them on white->black, which suggests to me is a poor ground on the 74LS08.
You could try adding 220R termination resistors to the R, G and B signals (to ground) where they enter the board.
So first of all I've improved the ground-plane with more hefty and direct grounds (and power) between the input, PI, CPLD and 74LS08. I've also added another 100nF ceramic cap directly across the power pins of the 74LS08, no improvement really from any of these.

I've also tried 220R resistors to GND on each of the ARED, AGREEN and ABLUE inputs from the Beeb, again no improvement.

Oh, and for what it's worth I'm now powering this from the Beeb's 5V and as expected it's the same.
IanB wrote:
Tue Jun 23, 2020 1:26 pm
I agree with Dave, you could also try adding pull downs at the RGB inputs to the CPLD board.
Also tried moving these 220R resistors to the CPLD RED, GREEN and BLUE inputs (pins 32, 31, 30 again all to GND) with no improvement.

One thought... the BRED, BGREEN, BBLUE (pin 34, 36, 37 of the CPLD) are floating (I didn't fit the 4K7 pull down's here)... could that be a cause?
That's about the only though I have right now (other than to go probing around with the scope to see if something looks amiss or trying to reduce the length of some of the CPLD to PI wiring).

I might have to live with this for now - it's not too off-putting and serves the immediate purpose (getting displays for my B+ and Master at the same time so I can play with Econet).
IanB wrote:
Tue Jun 23, 2020 1:26 pm
I still have blank pcbs available if you're interested.
Very much so. In fact I'd like to build up two units, one with the logic buffer front-end like this (for my B+/Master) and one with the analogue front-end to experiment with other machines (Spectrum in particular). However, I'm US, California based so was concerned at the postage and wondering if just getting them manufactured from China might be more cost-effective. If you could check on shipping cost to California (speed is not necessary) that would be super appreciated.

Thanks,
Steve.
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

riscyrobot
Posts: 6
Joined: Wed Jun 24, 2020 1:26 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by riscyrobot » Wed Jun 24, 2020 2:18 pm

IanB wrote:
Tue Jun 23, 2020 1:26 pm
I still have blank pcbs available if you're interested.
Hi Ian.
Do you still have any pcbs? I could use a couple if you have them - for my beeb and master, to use in some robotics projects.

-riscy.

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Wed Jun 24, 2020 8:38 pm

... additional info on the 'sparkly' artefacts on white->black transitions... if on the Sampling menu I set Sample Mode to 6 Bits Per Pixel this removes the issue (except in MODE 7 where it remains) . Any further thoughts?

Steve
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

User avatar
IanB
Posts: 582
Joined: Sun Sep 04, 2011 8:28 pm
Location: South Wales
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by IanB » Wed Jun 24, 2020 11:27 pm

riscyrobot wrote:
Wed Jun 24, 2020 2:18 pm
Do you still have any pcbs? I could use a couple if you have them - for my beeb and master, to use in some robotics projects.
Yes, I've just sent you a PM with some details.

User avatar
IanB
Posts: 582
Joined: Sun Sep 04, 2011 8:28 pm
Location: South Wales
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by IanB » Wed Jun 24, 2020 11:41 pm

*STEVE wrote:
Wed Jun 24, 2020 8:38 pm
... additional info on the 'sparkly' artefacts on white->black transitions... if on the Sampling menu I set Sample Mode to 6 Bits Per Pixel this removes the issue (except in MODE 7 where it remains) . Any further thoughts?
Mode 7 will only work in 3 BPP mode as there isn't enough time to capture at 6 BPP and deinterlace it.
Have you tried adding 4.7K pull downs to the unused secondary RGB inputs and Vsync input? Floating inputs are not a good thing especially in a noisy environment. (If you actually built the latest 6/8 bit design, there are two further inputs X1 and X2 that should be pulled down as well)

If that doesn't work, try programming in the 3 bit version of the CPLD. It is slightly different but will run on the six bit design and might behave differently under noisy conditions.

Have you wired the CPLD up for self programming or did you use an external programmer? If it's self programming, hold the 3 menu buttons down and press reset to get to the recovery menu and select the 3 bit CPLD.
Note putting the 3 bit CPLD on the 6 bit board is OK although not normally useful but putting the 6 bit CPLD on the original 3 bit board won't work.
*STEVE wrote:
Wed Jun 24, 2020 2:41 am
Very much so. In fact I'd like to build up two units, one with the logic buffer front-end like this (for my B+/Master) and one with the analogue front-end to experiment with other machines (Spectrum in particular). However, I'm US, California based so was concerned at the postage and wondering if just getting them manufactured from China might be more cost-effective. If you could check on shipping cost to California (speed is not necessary) that would be super appreciated.
I've just sent you a PM.

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Thu Jun 25, 2020 4:08 am

IanB wrote:
Wed Jun 24, 2020 11:41 pm
Mode 7 will only work in 3 BPP mode as there isn't enough time to capture at 6 BPP and deinterlace it.
Have you tried adding 4.7K pull downs to the unused secondary RGB inputs and Vsync input? Floating inputs are not a good thing especially in a noisy environment. (If you actually built the latest 6/8 bit design, there are two further inputs X1 and X2 that should be pulled down as well)
Just tried that, with the pull-downs added to the secondary RGB and Vsync inputs, no difference. I already have the pull-downs fitted to X1 and X2 since I hadn't discovered the 8-bit configuration so wasn't sure what these were used for and assumed as signalling or identification inputs.
IanB wrote:
Wed Jun 24, 2020 11:41 pm
If that doesn't work, try programming in the 3 bit version of the CPLD. It is slightly different but will run on the six bit design and might behave differently under noisy conditions.

Have you wired the CPLD up for self programming or did you use an external programmer? If it's self programming, hold the 3 menu buttons down and press reset to get to the recovery menu and select the 3 bit CPLD.
Note putting the 3 bit CPLD on the 6 bit board is OK although not normally useful but putting the 6 bit CPLD on the original 3 bit board won't work.
Also tried reprogramming the CPLD to the 3 bit version (wired in for self-programming). Exactly the same behaviour although setting the sampling to 6 bits per pixel doesn't help in that case.

My only thoughts right now are to go probing around with the scope looking for noise (or more likely a little ringing) on falling edges which could be the cause. Not sure if I have enough bandwidth though... to get an idea of timing, what's the period equating to 1 pixel?

Thanks for the suggestions so far - much appreciated. I might have to just live with this until I build up some proper boards, but have a feeling it will turn out to be something simple/stupid.

Steve.
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

User avatar
BigEd
Posts: 3173
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by BigEd » Thu Jun 25, 2020 8:13 am

*STEVE wrote:
Tue Jun 23, 2020 2:16 am
If you take a look at the video in the link to my google photos share you can see noise (speckles) in the black but only adjacent to white areas on the image. https://photos.app.goo.gl/r927MDkpbLhVPPSn8
I see no speckles in the area of any coloured pixels, only in the area of white on black. And I only see the speckles with black to left and right, never attached to a white pixel, although I'm a little less sure of that.

It seems to me like a noise or decoupling problem, caused by the white-to-black transition, which is after all an all-three-channels transition.

Have you another machine to check with, in case it's the input, not the HDMI converter?

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Thu Jun 25, 2020 9:19 pm

BigEd wrote:
Thu Jun 25, 2020 8:13 am
I see no speckles in the area of any coloured pixels, only in the area of white on black. And I only see the speckles with black to left and right, never attached to a white pixel, although I'm a little less sure of that.

It seems to me like a noise or decoupling problem, caused by the white-to-black transition, which is after all an all-three-channels transition.

Have you another machine to check with, in case it's the input, not the HDMI converter?
Hi Ed,

I've checked with my BBC Master as well as the B+ and both show identical behaviour. I've also connected my B+ back to my SCART TV input and all looks good with no speckles.

You're right in that the issue is always following edges from white-black which makes me think this is stray capacitance and ringing on a signal somewhere.

My most recent experiment was to try and slug each PI pin in turn to GND via a 4K7 resistor. Whilst this showed a marked improvement when loading any of GPIO 21, 7, 8 or 17 to GND I later found that the effectiveness of this depends on the screen content. Also it only makes an improvement if I apply the load at the PI end which reinforces my theory that the problem is long wires generating unwanted impedance.

Anyway I now have some proper PCBs on the way so I can put up with this for the time being.

Best Regards,
Steve.
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

firthmj
Posts: 238
Joined: Tue May 26, 2009 9:37 am
Location: Ipswich, UK
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by firthmj » Fri Jun 26, 2020 4:24 pm

IanB wrote:
Tue Jun 23, 2020 1:26 pm
I still have blank pcbs available if you're interested.
Hi,

Do you still have some blank boards? I would probably be interested in sharing a few with some of the other StarDotters around here
Had fun at the
Image
Meeting 21st September 2019

User avatar
anightin
Posts: 537
Joined: Thu Aug 23, 2018 2:03 pm
Location: Cambridge UK
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by anightin » Fri Jun 26, 2020 5:15 pm

Just expressing my interest in x2 RGB to HDMI rev 3 8bit board kits (with SMD parts ready-soldered) when available.

Just found 2 more machines that would like to hook up to HDMI and too hot and fiddly to solder SMD in the shed at the moment :)

Many thanks

Andy

User avatar
*STEVE
Posts: 42
Joined: Fri Jul 26, 2019 8:22 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by *STEVE » Sat Jun 27, 2020 11:47 pm

*STEVE wrote:
Thu Jun 25, 2020 9:19 pm
My most recent experiment was to try and slug each PI pin in turn to GND via a 4K7 resistor. Whilst this showed a marked improvement when loading any of GPIO 21, 7, 8 or 17 to GND I later found that the effectiveness of this depends on the screen content. Also it only makes an improvement if I apply the load at the PI end which reinforces my theory that the problem is long wires generating unwanted impedance.
Just an update to say that my issues with the sparkles is now resolved. Looks like it was most likely inductive coupling between the clock lines (GPIO17 and GPIO22) into other signals that were running as parallel wires, causing ringing like this...
DS1Z_QuickPrint1.png
Shortening the wiring on these two GPIO lines between the CPLD and PI turns these...
DS1Z_QuickPrint4.png
GPIO17 before
DS1Z_QuickPrint5.png
GPIO21 before
Into these...
DS1Z_QuickPrint7.png
GPIO17 after
DS1Z_QuickPrint8.png
GPIO21 after
(note GPIO21 isn't square with the bandwidth limitations of my scope)

With this change and a re-calibration all is fixed now, looking crisp and wonderful !
Great project, thanks everyone for your help and hard work on this super device =D> . Looking forward to building up some much neater boards to use. Next job, is getting my L3 file server working.

Steve.
  • BBC Model B+ 128k
  • BBC Master 128k
  • Econet (WIP)
  • +more

ryanm
Posts: 1
Joined: Sun Jun 28, 2020 12:48 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by ryanm » Sun Jun 28, 2020 10:09 pm

firthmj wrote:
Fri Jun 26, 2020 4:24 pm
IanB wrote:
Tue Jun 23, 2020 1:26 pm
I still have blank pcbs available if you're interested.
Hi,

Do you still have some blank boards? I would probably be interested in sharing a few with some of the other StarDotters around here
Hi Ian. I am also interested in either a blank or populated board - over here in Cape Town, South Africa. Hopefully international shipping is or will become possible soon! I have one of the small SCART to HDMI converters, but its synchronisation seems to be fairly random. It looks like a purpose built construction is best.

User avatar
IanB
Posts: 582
Joined: Sun Sep 04, 2011 8:28 pm
Location: South Wales
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by IanB » Mon Jun 29, 2020 1:19 am

*STEVE wrote:
Sat Jun 27, 2020 11:47 pm
With this change and a re-calibration all is fixed now, looking crisp and wonderful !
Great project, thanks everyone for your help and hard work on this super device =D> .
Glad you got it sorted out.
ryanm wrote:
Sun Jun 28, 2020 10:09 pm
Hi Ian. I am also interested in either a blank or populated board - over here in Cape Town, South Africa. Hopefully international shipping is or will become possible soon! I have one of the small SCART to HDMI converters, but its synchronisation seems to be fairly random. It looks like a purpose built construction is best.
Yes I still have about a dozen boards left. I've sent you a PM.

riscyrobot
Posts: 6
Joined: Wed Jun 24, 2020 1:26 pm
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by riscyrobot » Mon Jun 29, 2020 1:39 pm

IanB wrote:
Wed Jun 17, 2020 10:48 pm
I've ordered a small test batch of assembled RGBtoHDMI main boards from China so if that goes OK I will have some ready made ones available in due course.
I'll start a thread in the for sale forum once I have a better idea of the final price.
Could you put me down for two of the assembled 6-bit boards, Ian, plus one assembled analogue board to suit an Atom?
Cheers.

User avatar
danielj
Posts: 8226
Joined: Thu Oct 02, 2008 5:51 pm
Location: Manchester
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by danielj » Mon Jun 29, 2020 2:43 pm

riscyrobot wrote:
Mon Jun 29, 2020 1:39 pm
Could you put me down for two of the assembled 6-bit boards, Ian, plus one assembled analogue board to suit an Atom?
Cheers.
Hang fire until there's a for sale thread otherwise it all gets very confusing for people...

d.

User avatar
IanB
Posts: 582
Joined: Sun Sep 04, 2011 8:28 pm
Location: South Wales
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by IanB » Sat Jul 04, 2020 11:45 pm

While ordering the main CPLD boards from China, I looked at the possibility of getting the analog board made by the same company. However their quote was quite high so I decided to redesign the board to use components stocked by JLC instead as they will only build boards using their stocked components and that works out much cheaper than the other suppliers.

Fortunately they stocked suitable components, so I had some made up by JLC and they arrived a couple of days ago:
new analog.jpg
Top is the original DIY design and bottom is the new JLC built design.

It worked first time as I was powering the Pi using a separate power supply but had a problem later when powering it from a computer as the protection diode had been fitted backwards due to an error with the pick and place file. Fortunately removing and replacing that was an easy fix and I've updated the file for any new orders.

This new board isn't suitable for DIY construction because one of the chips is not readily available outside of China and it also uses smaller resistors, capacitors and ICs so the previous design will remain the DIY option (they are functionally the same).

Doing it this way worked out at roughly a quarter to a third of the cost of the previous quote.

I've also managed to get 4 level RGB and YUV working with a CPLD update plus fitting the extra chip (U7) on the back of the analog board and this allows all colours to be detected on the TI99/4a :
ticapture0.png
ticapture6.png
Unfortunately the results aren't ideal as the video levels on this computer drift with temperature so it isn't 100% stable for the first 15 mins or so after switch on but it seems to remain stable after that.

This also means that 4 level / 2 bits per pixel / 64 colour analog RGB works (compared to 3 level / 27 colour as standard) although I'm not aware of any computer that uses that. However it might mean that it will produce an acceptable though not entirely accurate picture when connected to a Nula.

aotta
Posts: 267
Joined: Fri May 26, 2017 9:57 am
Location: Italy
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by aotta » Sun Jul 05, 2020 11:10 am

Great work IanB! But the cpld update works with the old analog board (with bottom ic added) too? Else if, i'd like buy one of your new jlc board (hoping it will cost less than the ones I bought from pcbway! :wink: )
And, new sw versions works with other TI graphics video chips too (I.e. Msx, sega sc-3000, etc.)?

User avatar
IanB
Posts: 582
Joined: Sun Sep 04, 2011 8:28 pm
Location: South Wales
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by IanB » Sun Jul 05, 2020 2:33 pm

aotta wrote:
Sun Jul 05, 2020 11:10 am
Great work IanB! But the cpld update works with the old analog board (with bottom ic added) too?
Yes it works with the old analog board
aotta wrote:
Sun Jul 05, 2020 11:10 am
And, new sw versions works with other TI graphics video chips too (I.e. Msx, sega sc-3000, etc.)?
In theory it should but I don't have any way of testing them at the moment. The DAC levels will probably need to be adjusted.

aotta
Posts: 267
Joined: Fri May 26, 2017 9:57 am
Location: Italy
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by aotta » Sun Jul 05, 2020 4:33 pm

IanB wrote:
Sun Jul 05, 2020 2:33 pm
aotta wrote:
Sun Jul 05, 2020 11:10 am
Great work IanB! But the cpld update works with the old analog board (with bottom ic added) too?
Yes it works with the old analog board
I've a Max9144 left from one of your old pcb version, it seems having same pinout and characteristic of Max9108, do you think i can use it for adding the IC7 to my analog board?
I must change jp1 and jp2?
The new cpld firmware is already on github repository?
Well, a lot of questions... but i'm really interested in testing your progress!

User avatar
IanB
Posts: 582
Joined: Sun Sep 04, 2011 8:28 pm
Location: South Wales
Contact:

Re: RGB to HDMI using a Pi Zero and a small CPLD

Post by IanB » Sun Jul 05, 2020 5:19 pm

aotta wrote:
Sun Jul 05, 2020 4:33 pm
I've a Max9144 left from one of your old pcb version, it seems having same pinout and characteristic of Max9108, do you think i can use it for adding the IC7 to my analog board?
No.
Either of the MAX9144 and MAX9108 chips can be used on the DIY board but all chips on each board must be the same. If you mix chips on a board it messes up the calibration as they have different propagation delays (I tried it).
aotta wrote:
Sun Jul 05, 2020 4:33 pm
I must change jp1 and jp2?
No.
aotta wrote:
Sun Jul 05, 2020 4:33 pm
The new cpld firmware is already on github repository?
Not the finalised version yet, I will be posting a beta version to test in a few days.

Post Reply

Return to “8-bit acorn hardware”