128 X 64 mystery LCD identify

[MLXXXp]

Any way, most commonly used controllers are SSD1306 and SH1106, which is almost a clone of SSD1306, but with minor differences.

SSD1306 and SH1106 are for OLED displays. If the Roland has an LCD display then it won't be one of those controllers.

It could have a similar instruction set but the init sequence would probably be different.

[ataradov]

Yes, those SPI pins carry a very specific protocol (a series of register writes). Just knowing what register addresses are used and what values are written is enough to compare against the known controller datasheets.

[mkiijam]

Yes, those SPI pins carry a very specific protocol (a series of register writes). Just knowing what register addresses are used and what values are written is enough to compare against the known controller datasheets.

Well, I got the logic analyzer on the way. I'll update when it gets in.

[ataradov]

SSD1306 and SH1106 are for OLED displays. If the Roland has an LCD display then it won't be one of those controllers.
It could have a similar instruction set but the init sequence would probably be different.

But the display that was already tried and partially working is OLED, so the one in Roland should be also OLED.

There is no way regular LCD initialization code would even remotely work for OLED.

Also, there is no backlight  voltage on the connector, so it is 100% OLED.

[MLXXXp]

There is no way regular LCD initialization code would even remotely work for OLED.

The OP mentions a backlight and the photo in this post appears to confirm that. An OLED wouldn't have a backlight. Also, the mention of a contrast control would further imply an LCD. The OLED displays have "contrast" commands but they're not very effective.

[mkiijam]

It uses a separate backlight on another two wire cable. The backlight can be dimmed in the user menu.

[amyk]

SSD1306 and SH1106 are for OLED displays. If the Roland has an LCD display then it won't be one of those controllers.
It could have a similar instruction set but the init sequence would probably be different.

But the display that was already tried and partially working is OLED, so the one in Roland should be also OLED.

There is no way regular LCD initialization code would even remotely work for OLED.

Also, there is no backlight  voltage on the connector, so it is 100% OLED.

The OLED controller manufacturers realised that having a command set nearly identical to existing LCD controllers would decrease design-in effort, so they did. For example,

SSD1306:
0x = set column start address LSB
1x = set column start address MSB
20 xx = set memory addressing mode
21 xx yy = set column address
22 xx yy = set page address
40~7f = set display start line
81 xx = set contrast
a0, a1 = segment output direction
a4, a5 = normal, pixel test (all on)
a6, a7 = normal/inverted mode
ae, af = display on, display off
bx = set RAM page start address
c0, c8 = set COM output direction
e3 = nop

S6B0724:
0x = set column start address LSB
1x = set column start address MSB
20~27 regulator resistor select
40~7f = set display start line
81 xx = set reference voltage (contrast)
a0, a1 = segment output direction
a4, a5 = normal, pixel test (all on)
a6, a7 = normal/inverted mode
ae, af = display on, display off
bx = set RAM page start address
c0, c8 = set COM output direction
e2 = reset
e3 = nop

The former is the common OLED controller that's been discussed in this thread previously, the latter is an older LCD controller from the late 90s/early 2000s that doesn't offer windowing/scrolling capabilities. Another one similar to the latter is the Sitronix ST7565P.

[mkiijam]

So the obvious question is, Do I really have any hopes of finding an after market LCD screen that will work?

[ataradov]

So the obvious question is, Do I really have any hopes of finding an after market LCD screen that will work?

I would say, yes you do. You may not find the exact physical form factor, but you should be able to find a display that works.

I still don't see any evidence of an a actual backlight, so focus on OLED displays. And there are not too many types to try.

[mkiijam]

Here is the backlight

[ataradov]

Oh, ok. Then you should be looking for the actual LCDs, not OLED ones.

The name of the common controller for that was already mentioned.

Still, knowing the commands that are used will help clarify things. And the worst case scenario - you can just use a small MCU to act as a translation layer for an OLED display, since that already works more than it should.

[amyk]

Thanks to that picture, there's another part number that shows you can buy a replacement, but it is rather expensive:

https://www.merktron.es/es/roland/5251-lcd-spd-sx-cmf2p0791-ec3700007r0.html

It also suggests the same part fits the TD-25.

When you get your LA you can see the commands it sends, and that should narrow it down enough to find a replacement. My bet is still on ST7565/S6B0724.

[mkiijam]

Thanks to that picture, there's another part number that shows you can buy a replacement, but it is rather expensive:

https://www.merktron.es/es/roland/5251-lcd-spd-sx-cmf2p0791-ec3700007r0.html

It also suggests the same part fits the TD-25.

When you get your LA you can see the commands it sends, and that should narrow it down enough to find a replacement. My bet is still on ST7565/S6B0724.

Yeah... the part is available from Roland as well... like $100 or so. Where is the fun in that?

Okay, I've got the logic analyzer... NOW WHAT!?!? Just kidding (kinda) I'm jumping into the world of figuring out how to hook this up without any software or docs! YEA!

[mkiijam]

Okay, I've recorded the logic signals coming from PINs 1-5....

AND?

[ataradov]

First of all, in the Saleae software add an SPI decoder and make sure you can see bytes decoded without the errors.

You need to record very specific parts, not just random signals. You need to catch very first commands it sends to the LCD after power up.

[mkiijam]

First of all, in the Saleae software add an SPI decoder and make sure you can see bytes decoded without the errors.

You need to record very specific parts, not just random signals. You need to catch very first commands it sends to the LCD after power up.

I'm using PulseView.

I have added the SPI decoder and it seems like it's capturing something. I captured when it first turned on. I also asked it to decode with ST7735, but there were errors.

What is a ST7567? That Chinese LCD that seemed to have the same 12 pin functions says it uses that CHIP.

[ataradov]

You have 1 MHz sampling rate selected. This is way too low. Select the highest it will go (24 MHz) and hope that the actual clock is slow enough to meaningfully capture it. You need to zoom in on a single byte and make sure that you see periodic clock.

And then attach the actual capture file.

 ST7567 is the controller IC it can be used in 100s of different displays. But all those displays will have a compatible command set because of that.

[madires]

Or possibly an ST7565 which is quite common. BTW, the ST7735 is meant for color displays.

[mkiijam]

Pulse view seems to glitch at 24MHz

Here is a file at 12MHz

[ataradov]

Sometimes the capture looks like the real SPI, but occasionally there seems to be some strange noise.

If you look at D1, which I assume was CS. You can see that for the first couple transfers it is fine and it aligned at 8 bits. But then just a few transfers in, it is just a random noise not aligned to anything. The data is also repeated on D3 trace.

So double check your connections to the logic analyzer.

But you are on the right path. You can start to see the commands being sent. Now you can start matching that to controller datasheets.

[mkiijam]

Connections look right.

I don't see D1 and D3 as the same, but not sure what you mean.

[ataradov]

Look at the chunk at +3172170us in your log. In this part the signals look corrupted.

Do multiple captures and compare them .See if you get the same results each time.

[mkiijam]

This seems  different..

[ataradov]

This one has issues with the CS on  the second and third transfers (+2919594us and +2929596 us).

Figure out what is wrong with your setup. CS must remain low for the whole duration of the clock.

Just double checking. Do you have ground of the analyzer connected to the ground of the device?

[mkiijam]

...Just double checking. Do you have ground of the analyzer connected to the ground of the device?

Oops...

How about this one?