Advantest/ADCMT R6581 VFD to TFT conversion

[vf33184]

Hello Ian,
thank you for very good description of different SPI buses and its function.

I have been running TEST2 HEX file again, this time more than 24 hours.  MAIN display counted 1 million and AUX display 6.5 million samples, VFD SPI report no errors (see picture). Counting is very smooth without any interruption, haven't noticed any glitches or flickers.

But with regular firmware some symbols are sometimes being replaced by "?" (shown on picture in my previous messages) just probably for single display refresh cycle, next cycle it's being corrected. Based on your description it would lead me to conclusion that:
a) no issue on communication between BluePill and LT7680A-R driver
b) no issue on communication between BluePill and TFT COG
c) issue with data being spied on SPI link between main meter CPU and Hitachi CPU on display board. If so, then I don't know why TEST2 firmware reports "VFD SPI - no errors". I don't know how do you evaluate issue on VFD SPI, if each data batch has some parity check or some other mechanism?

If there is corrupted data at VFD SPI input, BluePill replace unknown symbol by "?" and send it to TFT or BluePill sends unknown received character down to TFT and then TFT itself replace unknown character by "?" symbol?

Sorry for many questions, I am just trying to find out where is my problem. 

Thank you.

Vladimir

You've got it right.....just to add that the SPI errors thats can be reported are probably rather limited. It's a built in mechanism I am am not 100% sure of all the failure modes it can report on.
Can you post a photo of the termination of the 3 wires down onto the R6581 display board.

Ian.

Hello Ian,
here is the picture of wiring between R6581 front panel board and new display board.

There is still a question, when there are corrupted data at VFD SPI input, BluePill replace unknown symbol by "?" and send it to TFT or BluePill sends unknown received character down to TFT and then TFT itself replace unknown character by "?" symbol?

Thank you.

Vladimir

[coromonadalix]

good questions,  could be the interface pcb for the vfd who could spit ?   because there was some error ? or miss interpretation ?  drop in the spi data flow ?

noise on the supply line,  would you try to add an 0.1uf   on the 5v vcc line   could help

for the 3 small wires, could they catch some noise ?  i would have tried to match their lenght ?

[IanJ]

Hello Ian,
here is the picture of wiring between R6581 front panel board and new display board.

There is still a question, when there are corrupted data at VFD SPI input, BluePill replace unknown symbol by "?" and send it to TFT or BluePill sends unknown received character down to TFT and then TFT itself replace unknown character by "?" symbol?

Thank you.

Vladimir
[/quote]

Blue Pill code, data coming in from the R6581 is checked against a bitmap/ASCII table.
Example:
If this is received then it maps it to 0x34 which is the number 4
{{0x02, 0x06, 0x0A, 0x12, 0x1F, 0x02, 0x02}, '4'},  // 0x34, 4

If there is no match then "?" is returned and thats is what is passed to the LT7680A-R.

Can you scope V-SCK, V-SDA, V-RESTART and the +5V supply.

Also, see Post #1 at the top (or attached), there is an issue with one of the versions of PCBs that require a track cut to avoid a short on my PCB (my mistake in the Gerbers). Saying that I assume you've done that because with the short the display is blank.

Ian.

[jj131415]

hi Ian,
Maybe something is missing during the SPI character capture from R6581 front panel. I don't know if it caused by SPI interrupt and conversion time of STM32 MCU. If let R6581 run internal calibration, it's easily to see "?".

[IanJ]

 

hi Ian,
Maybe something is missing during the SPI character capture from R6581 front panel. I don't know if it caused by SPI interrupt and conversion time of STM32 MCU. If let R6581 run internal calibration, it's easily to see "?".

I hear you, but it works for many other people without this issue.
Are you running the latest firmware on the Blue Pill.....?

It just looks like it's SPI corruption on the VFD SPI........but I guess it could be intermittent on the TFT SPI or even the LCD itself.
If you have tried another Blue Pill/LT7680 driver board then I guess I could take a look at your setup if you ship me my board c/w BluePill, LT7680 driver & TFT (at your cost both ways). I'm in the UK.

Ian.

[IanJ]

Hi all,

Just an observation……with the latest firmware and using the BuyD config I notice flicker returns when my aircon switched off and the temp rose to 24.8degC…….then stopped flickering once temps reduced……so definitely the LT7680 driver or TFT itself hates the heat.
More investigation required.

Meanwhile my 2nd R6581T had no such issue and has never, ever flickered. Rock stable.

Ian

[vf33184]

Hello Ian,
here is the picture of wiring between R6581 front panel board and new display board.

There is still a question, when there are corrupted data at VFD SPI input, BluePill replace unknown symbol by "?" and send it to TFT or BluePill sends unknown received character down to TFT and then TFT itself replace unknown character by "?" symbol?

Thank you.

Vladimir

Blue Pill code, data coming in from the R6581 is checked against a bitmap/ASCII table.
Example:
If this is received then it maps it to 0x34 which is the number 4
{{0x02, 0x06, 0x0A, 0x12, 0x1F, 0x02, 0x02}, '4'},  // 0x34, 4

If there is no match then "?" is returned and thats is what is passed to the LT7680A-R.

Can you scope V-SCK, V-SDA, V-RESTART and the +5V supply.

Also, see Post #1 at the top (or attached), there is an issue with one of the versions of PCBs that require a track cut to avoid a short on my PCB (my mistake in the Gerbers). Saying that I assume you've done that because with the short the display is blank.

Ian.

Hello Ian,
yes, I have PCB version 1.1 what is affected by that Gerber error what caused blank screen at the beginning of my display replacement process, but I already cut that shortening traces and display start working.

In a meanwhile I have added 100nF capacitor on 5V input and modified wiring between BluePill and VFD board (all three wires twisted with different routing) but it didn't fix the problem. See attached picture.
   
One of your ideas was that BluePill itself can be an issue therefore I have ordered new one from different source. It has already arrived (has different reset button and both LEDs are red) and I started with its programming. Immediately I have noticed one difference, new BluePill was programed much faster (just in 1.5s) compared to old one (3s). I am using ST-LINK V2 USB dongle and ST-LINK UTILITY software. What is your typical programming time?

There is one more difference. You previously mentioned that your BluePill has 64kB flash, but both of mine have 128kB, not sure if this doesn't indicate they are counterfeit?

Then I tried new BluePill on display board. During one day of continuous running I haven't notice any "?" symbol. So far so good.

But then I start playing with Menu and noticed that when I am moving by arrows between different Menu items (e.g. calibration, beeper, GPIB etc) sometimes very shortly "?" symbol appears in place of some letter of submenu what is blinking. Other items what are not blinking never have this issue. Then I tried to run internal calibration. During its execution there is INT CALIBRATION blinking on display. I have noticed that "?" symbol sometimes shows (again very shortly), but also that frequency of blinking is fluctuating a bit. Also sometimes whole "INT CALIBRATION" light up at the same moment, sometimes "LIBRATION" and fraction of second later also "INT CA" lights up. I recorded slow-motion video and this behavior is clearly visible. Similarly when I am measuring e.g. voltage, digits are sometimes being updated not at the same time but in different order (I mean that 7th digit is updated and then 5th digit).

Here is video I recorded showing display behavior described above:  https://youtu.be/GlcbgusXKIo

Maybe I am just too sensitive about these display artifacts and most of users will not notice it. Maybe I have some issue with hardware. Or maybe there is some imperfection in the way how data are being processed by BluePill and/or LDC driver or LCD itself.  BluePill being too much loaded? Or timing issue on one of SPIs? Or maybe there is the other way how display updates should be commanded in some bulks to have whole display updated at the same moment instead of individual symbols?

I am trying to explain as good as possible what I see. I can be completely wrong. Or it can help you to find the issue and implement some data corruption identification and correction mechanism in software, or different display handling procedure. You mentioned waiting for additional details from LT7680A-R producer, that can help too.

Thank you.

Vladimir

Note: all tests made with 26-01-25 HEX file and default BuyD setting.

[IanJ]

First thing I noticed in your photo is you have twisted the small wires together, that can cause crosstalk. Please untwist them.

The TFT should be a beautiful experience, with virtually no artefacts (? Etc). I will try INT CALIBRATING on mine later today and let you know how mine performs.

64k/128k are just two options for the Blue Pill. I have always used the 64k version.

Ian.

UPDATE: INT CALIBRATING display works perfectly, alternating blank and the text.

[jj131415]

hi Ian,
I'm using the FW is 26th Jan '25 version, but this "?" also can be seen in several previous FW during INT CALIBRATING. Thanks.

[IanJ]

hi Ian,
I'm using the FW is 24th Jan '25 version, but this "?" also can be seen in several previous FW during INT CALIBRATING. Thanks.

I don’t really know why INT CALIBRATING displays phantom “?” When other screens don’t. Might be the R6581 data timing is slightly different if the R6581 is under its own CPU load.
I’ll take a look at the acquisition code, see if any improvements can be made.

[jj131415]

hi Ian,
I take a video on it, so I can capture this momentary "?" picture, it's very fast and easily to ignore it. Thanks.

[vf33184]

hi Ian,
I take a video on it, so I can capture this momentary "?" picture, it's very fast and easily to ignore it. Thanks.

See my previous post where I have link to slow motion video on youtube I have recorded to show this behavior with “?” symbol.

[IanJ]

hi Ian,
I take a video on it, so I can capture this momentary "?" picture, it's very fast and easily to ignore it. Thanks.

See my previous post where I have link to slow motion video on youtube I have recorded to show this behavior with “?” symbol.

Have you untwisted the 3 wires as I mentioned above?
Crosstalk on the SPI could very well be the problem with your display…….which I think is a different problem here.

Ian

[vf33184]

hi Ian,
I take a video on it, so I can capture this momentary "?" picture, it's very fast and easily to ignore it. Thanks.

See my previous post where I have link to slow motion video on youtube I have recorded to show this behavior with “?” symbol.

Have you untwisted the 3 wires as I mentioned above?
Crosstalk on the SPI could very well be the problem with your display…….which I think is a different problem here.

Ian

Hi Ian,
yes, I have untwisted those 3 SPI VFD wires, still the same behavior what I have recorded on youtube video.

Vladimir

[jj131415]

hi Ian,
They are not twisted together.

[IanJ]

I’ll have a look at the code next few days, I ran INT CALIBRATING a few times and can see a screen anomaly albeit nothing like the YT demo uploaded.

Here's my own video, I don't really see any "?" but I notice the way it updates the display, sometimes in one go, sometimes you see a few chars at a time. I don't know if this is related to the "?" issue but worth exploring.
https://www.youtube.com/shorts/Zh2gXrGFRAs

Ian.

[IanJ]

Hi all,

Attached to this post is a test .HEX file.

I have revised some timing & interrupts to help the TFT have the time and resources necessary to write to the TFT.

Could anyone who is experiencing "?" being displayed randomly please test it out and report back.
(I don't really have the issue on both my R6581T's so unsure what difference it will make, if any)

Ian.

[vf33184]

Hi all,

Attached to this post is a test .HEX file.

I have revised some timing & interrupts to help the TFT have the time and resources necessary to write to the TFT.

Could anyone who is experiencing "?" being displayed randomly please test it out and report back.
(I don't really have the issue on both my R6581T's so unsure what difference it will make, if any)

Ian.

Hello Ian,
I will test new firmware during weekend, first I need to mount my meter back together.

In a meanwhile I would ask for one clarification about previous R6581_VS_DisplayTestTFTonly1.hex testing file. You are writing that it is not using VFD SPI but just TFT SPI. I am still trying to troubleshoot my display upgrade so I have ordered another set of TFT and display driver and also populated second your PCB. However when I am testing it on the bench without connection to meter, display is initiated (backlight is on) but not displaying anything.

Should R6581_VS_DisplayTestTFTonly1.hex work without connected VFD SPI? Or it still needs VFD SPI signals being connected to trigger displaying counting even just TFT SPI counts on the display? It seems to me that VFD SPI needs to be connected to enable software to run. It behaves the same on both of my TFT+driver+BluePill sets what I am testing on the bench.

Thank you.

Vladimir

[IanJ]

Hello Ian,
I will test new firmware during weekend, first I need to mount my meter back together.

In a meanwhile I would ask for one clarification about previous R6581_VS_DisplayTestTFTonly1.hex testing file. You are writing that it is not using VFD SPI but just TFT SPI. I am still trying to troubleshoot my display upgrade so I have ordered another set of TFT and display driver and also populated second your PCB. However when I am testing it on the bench without connection to meter, display is initiated (backlight is on) but not displaying anything.

Should R6581_VS_DisplayTestTFTonly1.hex work without connected VFD SPI? Or it still needs VFD SPI signals being connected to trigger displaying counting even just TFT SPI counts on the display? It seems to me that VFD SPI needs to be connected to enable software to run. It behaves the same on both of my TFT+driver+BluePill sets what I am testing on the bench.

Thank you.

Vladimir

No, I only tweaked the existing SPI for the TFT interface is what I meant. Everything is the same.

SPI1 = VFD data capture
SPI2 = TFT LCD drive
SPI3(bitbang) = TFT LCD COG initialization.

Next, I will start looking at how I can modify MickleT's VFD capture code to help it fit better with my code and capture R6581 data more reliably......because this is where I think the random "?" problem is.

Ian.

[vf33184]

Hello Ian,
I will test new firmware during weekend, first I need to mount my meter back together.

In a meanwhile I would ask for one clarification about previous R6581_VS_DisplayTestTFTonly1.hex testing file. You are writing that it is not using VFD SPI but just TFT SPI. I am still trying to troubleshoot my display upgrade so I have ordered another set of TFT and display driver and also populated second your PCB. However when I am testing it on the bench without connection to meter, display is initiated (backlight is on) but not displaying anything.

Should R6581_VS_DisplayTestTFTonly1.hex work without connected VFD SPI? Or it still needs VFD SPI signals being connected to trigger displaying counting even just TFT SPI counts on the display? It seems to me that VFD SPI needs to be connected to enable software to run. It behaves the same on both of my TFT+driver+BluePill sets what I am testing on the bench.

Thank you.

Vladimir

No, I only tweaked the existing SPI for the TFT interface is what I meant. Everything is the same.

SPI1 = VFD data capture
SPI2 = TFT LCD drive
SPI3(bitbang) = TFT LCD COG initialization.

Next, I will start looking at how I can modify MickleT's VFD capture code to help it fit better with my code and capture R6581 data more reliably......because this is where I think the random "?" problem is.

Ian.

Hi Ian, yes, I understand this. 

My question was related to testing hex file R6581_VS_DisplayTestTFTonly1 what was released like month ago. The question was if it needs valid incoming VFD SDI data to run the code or not. 

I have BluePill+driver+TFT combo on the bench connected just to lab 5V power supply and the code is not running (nothing on display just back-light on), and my conclusion is that the code really needs valid incoming VFD SPI data for running, regardless it is not using those data at all. If so, then it explains the behavior I see, and it means that using the code for bench only testing is not possible. 

Thank you.

Vladimir

[IanJ]

Hi Ian, yes, I understand this. 

My question was related to testing hex file R6581_VS_DisplayTestTFTonly1 what was released like month ago. The question was if it needs valid incoming VFD SDI data to run the code or not. 

I have BluePill+driver+TFT combo on the bench connected just to lab 5V power supply and the code is not running (nothing on display just back-light on), and my conclusion is that the code really needs valid incoming VFD SPI data for running, regardless it is not using those data at all. If so, then it explains the behavior I see, and it means that using the code for bench only testing is not possible. 

Thank you.

Vladimir

Ahhh sorry, I misread..........(I'm busy writing .py for Home Assistant right now!)

> Test HEX files:
> R6581_VS_Display_Test2.hex
> Test writing to the TFT LCD without actual VFD SPI data but actions when a VFD SPI packet is received.
> This initial version will count each time the TFT is updated on the MAIN display, and on the AUX display it will count each time data is captured from the R6581 SPI bus.
> Any VFD SPI error codes are reported. TFT LCD refresh is set to 100ms. See R6581_VS_Display_Test2.hex attachment below.

> R6581_VS_DisplayTestTFTonly1.hex
> TFT counter only, VFD SPI is disabled.

My memory is bad.....but from my notes, R6581_VS_DisplayTestTFTonly1.hex, you should get something without VFD data (disabled VFD SPI). It just writes to the TFT.
I can't confirm as I don't have that standalone setup handy.

Ian.

[IanJ]

Hi all,

Attached to this post 3 test .HEX files.

I have further revised some timing, DMA & interrupts to help the TFT have the time and resources necessary to write to the TFT, and the VFD capture.

R6581_VS_Display_21_02_25_test1.hex
Pausing VFD SPI2 whilst writing to the TFT:

R6581_VS_Display_21_02_25_test2.hex
As above but with additional pausing TFT SPI1 whilst capturing/processing VFD data.

R6581_VS_Display_21_02_25_test3.hex
As both above but with additional disable of FIFO check when sending to the TFT.

Could anyone who is experiencing "?" being displayed randomly please test them out and report back.
(I don't really have the issue on both my R6581T's so unsure what difference it will make, if any)

Ian.

[vf33184]

Hi Ian, yes, I understand this. 

My question was related to testing hex file R6581_VS_DisplayTestTFTonly1 what was released like month ago. The question was if it needs valid incoming VFD SDI data to run the code or not. 

I have BluePill+driver+TFT combo on the bench connected just to lab 5V power supply and the code is not running (nothing on display just back-light on), and my conclusion is that the code really needs valid incoming VFD SPI data for running, regardless it is not using those data at all. If so, then it explains the behavior I see, and it means that using the code for bench only testing is not possible. 

Thank you.

Vladimir

Ahhh sorry, I misread..........(I'm busy writing .py for Home Assistant right now!)

> Test HEX files:
> R6581_VS_Display_Test2.hex
> Test writing to the TFT LCD without actual VFD SPI data but actions when a VFD SPI packet is received.
> This initial version will count each time the TFT is updated on the MAIN display, and on the AUX display it will count each time data is captured from the R6581 SPI bus.
> Any VFD SPI error codes are reported. TFT LCD refresh is set to 100ms. See R6581_VS_Display_Test2.hex attachment below.

> R6581_VS_DisplayTestTFTonly1.hex
> TFT counter only, VFD SPI is disabled.

My memory is bad.....but from my notes, R6581_VS_DisplayTestTFTonly1.hex, you should get something without VFD data (disabled VFD SPI). It just writes to the TFT.
I can't confirm as I don't have that standalone setup handy.

Ian.

Hello Ian,
I have got in trouble. I made many changes to find that random issue, but now my setup stopped working at all.

Actual behavior:
1. When I start my meter, I can see (using oscilloscope) communication on LCM.SCK, LCM.MOS1, LSM.MIS0 and LSM.CS. I guess this is SPI3 bus for COG on TFT.
2. After like 2 seconds communication stops and LCD backlight is on, and there is horizontal and vertical white line on screen (see picture). As you described in one of your previous messages this should be proper behavior as display just needs to be initiated once.
3. and that's it, nothing more displayed

I am probing different signal:
1. SPI VFD data (V.SDA, V.SCK and V.RESTART) - I can see communication on all 3 lines
2. L.SCK, L.DIN and L.CS (this should be SPI2 = SPI TFT) - there is no communication at all and this is the reason why LCD is blank (except those two while lines), also green LED on BluePill is not blinking. Should it be blinking when receiving valid VFD data?

I have two suspicions what has happened:

1. my ST-LINK V2 programmer is not working properly and either programmed BluePill incorrectly or even killed that partially - I already have 3 pieces of BluPills and they behave the same way now
2. I tried service menu on R6581 - during turn-on hold HOME button. And then I initiated display test what was working. But during next turn-on I was not able to initiate this service menu any more and after few tries to reprogram BluPills, now display is not working any more

I have ordered another BluePill and also ST-LINK to try again, this should address suspicion 1.

Can you try service menu described in suspicion 2 if it works properly for you, please?

Sorry, I am not able to test those new HEX files you created with different TFT timing until I resolve my issue.

Thank you.

Vladimir

PS: I have two sets of display-driver-your board and 3 BluePills, all combinations behave the same.

[IanJ]

When you initiate Power on with HOME, you need to release the HOME button quickly after power up, if you hold the HOME button on too long and release it then the DIAG menu does not appear.
I presume it does this with the original VFD also.

For you main problem, you are checking things properly.....all connections in the right place: 3 signals +5v, GND. etc.

I presume you have checked ST-LINK Utility app settings incase you changed anything by accident?

You could try updating the firmware in your ST-LINK V2. I do remember I had to update the firmware in my V3 before it would work properly.

If you really get nowhere and have tried everything then you can send your LCD, R6581 Pcb & BluePill to me and I will try it on my R6581......and return to you.

Ian.

[vf33184]

When you initiate Power on with HOME, you need to release the HOME button quickly after power up, if you hold the HOME button on too long and release it then the DIAG menu does not appear.
I presume it does this with the original VFD also.

For you main problem, you are checking things properly.....all connections in the right place: 3 signals +5v, GND. etc.

I presume you have checked ST-LINK Utility app settings incase you changed anything by accident?

You could try updating the firmware in your ST-LINK V2. I do remember I had to update the firmware in my V3 before it would work properly.

If you really get nowhere and have tried everything then you can send your LCD, R6581 Pcb & BluePill to me and I will try it on my R6581......and return to you.

Ian.

Hello Ian,
in a meanwhile I did following steps:
- ordered another Bluepills (I have already 5 pcs) and ST-LINK v2
- programmed Bluepill with new ST-LINK v2 - I have tried both STM32CubeProgrammer and ST-LINK Utility software
- tried different BluePills with different firmware versions and TFTs and driver boards
Unfortunately it still doesn't work. 
 
I believe that firmware is being properly programmed to BluePill because TFT is being initialed (TFT back-light gets on after turn the meter on). I can in oscilloscope that VFD SPI data are coming on input pins of BluePill. Unfortunately TFT SPI is absolutely silent, not signals coming out. I am just curious how this can happen? I am not sure how TFT SPI is being initialized on your code, if it checks validity of input VFD SPI data and if those are seeing not valid, then TFT SPI is silent?
On Oscilloscope I can't recognize if VFD SPI data are valid, I can just see that signal is coming in.

I am thinking about last test what I can do. I can solder back (it won't be simple) VFD display (it is weak but still a bit visible). If it works, then VFD SPI data must be valid and problem is somewhere in BluePill-driver-TFTdisplay). If VFD wont work, than something got broken within front panel Hitachi CPU of main CPU board.

Would you be so kind and check your R6581_VS_DisplayTestTFTonly1.hex code if it is really running even without presence of valid incoming VFD SPI data?

 Thank you.

Vladimir