Tektronix TDS744 CRT to LCD color converter FPGA module DIY

[TerrySt]

I finished the new design using a Spartan 7 FPGA instead of using a development board.  I was able to solder the FPGA and the other small components using a hot plate and  a hot air gun.  The FPGA uses the FTGB196 package (196 balls in a 1mm grid).  I was pretty pleased at how well the FPGA balls looked after soldering (looking in under the edges).  I still used a larger pcb than required so that it can reach the mounts on the o'scope.
I also printed a bracket and bezel for the display.  I started with the files that TERRA provided earlier, and modified them for the 3D printer.  I added standoffs instead of using spacers and also added stiffeners to the bracket since it was a bit flimsy in plastic (would be fine in aluminum).  I added an extra layer to the bezel to keep it centered in the frame.
Lastly, I added an anti-glare screen protector (https://www.amazon.com/dp/B01MQYZBX9?psc=1&ref=ppx_yo2ov_dt_b_product_details).  It can be cut to size and works great.  It cuts down on reflections pretty well.  I'll attach a picture with the bezel and screen protector in place.

Terry

[TerrySt]

Here is the Spartan 7 FPGA board I made.  Like I said earlier, this was just a learning exercise.  The Spartan 7 is probably not the best choice for this project.  The project uses a small percentage of the FPGA resources and it requires extra power supplies (3.3V, 1.8V, 1.0V and 2.5V) just for the FPGA.  Then you still need 3.3V and 12V for the LCD panel.  Also, it requires level shifters on each of the lines from the o'scope.  So it ends up being a more complicated design that it would if you used a Spartan 3 or something similar. 
It also doesn't save any money over using a development board.  The Mini S7 dev board I used in my first version was about $80 after shipping to the US, but the FPGA is $20 and the extra pcb costs to go to a 4 layer board with smaller vias and ENIG finishing for the FPGA easily costed more than the dev board.  But now I know I can design and solder using BGA chips (at least the 1mm pitch ones).

Terry

[TERRA Operative]

How did my parts fit? I can make changes and adjustments if needed

[TerrySt]

How did my parts fit? I can make changes and adjustments if needed

They were perfect except for the cutouts for the connectors on the bracket.  They didn't match up with the connectors on my LCD panel.  They were close enough that the cables still could exit through the cutouts.  Other than that, the changes I made were just to reinforce it for printing with plastic (PLA).

I will attach what I ended up with and you can compare to what you made.  Thanks for providing them.  They save me lots of time.

Terry

[TERRA Operative]

Hmm, I can't download. Seems the forum attachment bug has struck again.
Can you reupload but ignore the first attachment upload slot and use the second slot to get around the bug?

[Tantratron]

Me too, I cannot upload the last 2 attachements of Terry (message #75 and #78)

[TerrySt]

I wasn't aware of the forum upload bug.  I'll try again.
First, the step files for the bezel and bracket.

[TerrySt]

The LCD with the bezel and the screen protector:

[TerrySt]

...and the FPGA board.

[Tantratron]

Many thanks Terry for the new uploads, it works fine.
All my congrats for your BGA soldering skill and work, I could never do this myself.
Happy New Year, Albert

[TerrySt]

Here is the latest version of my VHDL code that uses the LVDS transmitters in the FPGA instead of using external LVDS converter chips.

Terry

[ltarjanyi75]

Terry, your code works wonderfully on my Spartan 3 board also!

I just had to make one modification: I reverted back to the process in your previous version, where you examined the CLK_175 and CLK_25 in the same process. After that everything works correctly: changing colors, intensity, etc. 
Without this modification I just had some blue and black pixel "noise" on the LCD. Don't know why exactly...

But anyway it is super cool!!! Thanks for your help!

[TERRA Operative]

Just made some modified mounting plates, adjusted as per TerrySt's modifications (I derped the design somehow... )

I included STEP files for a thicker 3D printable mounting plate that doesn't need standoffs, and the original that can be cut from aluminium plate with the use of 2mm tall standoffs.

[TERRA Operative]

Hey, ltarjanyi75, would you mind releasing your schematics/PCB files if it is all working perfectly?



Also, how hard would it be, or would it even be possible  to get something like the Lattice LCMXO2-256HC-4TG100C working? (I have practically no programming experience however...)
I can get them in Akihabara for cheap.
https://www.latticesemi.com/view_document?document_id=38834

Where is everyone buying their Spartan 3 chips?

[ltarjanyi75]

Yes, I can share the details, I just need some more time to fix my design errors and test before that.

I quickly checked the datasheet of the family of PLD you sent and I think it could work with one of those also. Here is the brief resource usage summary of the code for Spartan 3 and these fit into the specs:
- Number of Slice Flip Flops: 51
- Total Number of 4 input LUTs: 172
- Number of bonded IOBs: 31
- Number of DCMs: 1
These devices also have on-chip flash for configuration, so no external flash would be required if I'm correct.
Maybe I will give a try later to check the development tool for this device and try to compile the code...

I bought my Spartan 3 chips from Aliexpress, about 8 USD each (+ shippping). Bought 2 of them and they work perfectly.

[TERRA Operative]

Great. If you find that you can compile code, I can draw up a PCB to test on.

Having another look at the datasheet, it seems I need to use the XO2-640U (not the XO2-256 that I can get off the shelf here) Lattice FPGA if I want dedicated LVDS output buffers and a PLL. Are these needed for this application?

[ltarjanyi75]

Yes, that is what I have just found also, because I could not add PLL to the 256 type.
640U is the smallest with 1 PLL, which is required for 25MHz to 175MHz clock multiplication. The smallest package of this is 144 pin TQFP.
(Compared to Spartan 3, which has 100 pin package)

Otherwise I could modify the VHDL code to synthesize without the pin/IO constraints... I don't know yet know how this can be compiled to a programmable binary on this platform and also do not have cable/programmer etc. to test it.
But at least it still seems to be doable even if not with the smallest chip.

[TERRA Operative]

Well, I can only get the non-PLL 256 type locally, so if I have to order chips, I may as well stick with the Spartan 3.

Do you have a link to the seller you bought from? If they are selling known good chips, I'll buy from them too.

(I just found Spartan XC3195A-3PQ208C for about 4000yen per 4 units being sold locally if they would work?)

[ltarjanyi75]

Do you have a link to the seller you bought from? If they are selling known good chips, I'll buy from them too.

This is where I bought from:
https://www.aliexpress.com/store/1102598538/pages/all-items.html?sortType=bestmatch_sort&SearchText=XC3S50A

[ltarjanyi75]

I had some time lately to further enhance my LCD solution. It was more a fun to build than it will be a frequently used feature.
The idea was to solve the brightness setting of the LCD (without opening the scope and changing DIP switch setting) which is possible by varying the duty cycle of the LCD PWM signal.

What I did:
- placed a resistive touch panel in front of LCD
- moved the PWM signal generation from FPGA to an atmega4809 chip/board (installed behind LCD) and program written with Arduino IDE
- Brightness can be set by (controlled by Atmega):
   - touching the upper right corner of screen for around 3 seconds
   - then you can set the value on the left of screen with the y value
   - you can save the value permanently by pressing save button (lower right)
   - if not saved the setting is only stored temporarily till the next restart
   - setting will disappear after not touching the screen for about 5 seconds
- FPGA code modified:
   - removed PWM signal generation
   - accept commands and brightness values from Atmega
   - necessary logic for brightness setting UI elements to overlay the pixel info received from scope during setting

The code could be further enhanced, but probably I will stop here.
(The resistive touch coordinate reading is not perfect, it has some false readings, but most of the time it works fine. And the UI is a bit ugly...)


https://1drv.ms/v/s!Am44vwFO-Vr3gsB13GQrtzPChORk2w?e=a98BJh

[TERRA Operative]

Nice!

I wonder how much of a project it would be to communicate back and forth via GPIB to actually control the scope like a modern touchscreen scope?