LCD Text jumbled on 6502

[GARTHWILSON]

Dr Jefyll wrote:

Paganini wrote:

... a known-working design to run known-working code.

Design, yes. Code... maybe not so much. According to Garth, "a reliable reset and setup requires writing the function set three times as shown there; and if you don't do that, you can sometimes get problems like you're talking about."

The code is in the file he posted, so I saw the three-times thing is not there.  It's also apparently made to work without RAM, since there are no subroutine calls, and the code to feed each letter is straightlined.   (I don't know what the ".s" file extension is to stand for.  I've always only seen ".asm" for assembly-language source-code files; but since the computer didn't know what to open it with, I just tried the text editor, and that worked.)

All the intelligent character LCDs I've worked with—and that's a lot, of many sizes—used the Hitachi 44780 and support ICs, with the possible exception of a couple that had the die mounted directly on the board, with bondwires going directly to the board, and covered with a blob of epoxy, meaning I couldn't see if it was Hitachi or not.  Although it wouldn't surprise me if there are other brands, I don't recall ever seeing any others.

[pjdennis]

besser435 wrote:

Sorry, I forgot about that. The code can be viewed as plain text here: https://eater.net/downloads/hello-world.s

That version of the code is intended to be used with the slow clock. It won't work with the 1MHz crystal because it does not include logic to wait for the LCD to be 'ready' before sending each command, hence the missing characters. You'll want the 'final' version of the code instead: https://eater.net/downloads/hello-world-final.s.

[pjdennis]

Paganini wrote:

If I remember correctly, Ben's code worked on Blue April but only at power up. The power up met the LCD module's setup requirements, but a system reset did not; so it would stop working after a reset. Using the full setup sequence Garth described along with some appropriately proportioned delay loops let me run Blue April at 8Mhz (instead of 1Mhz!) and also switch the LCD to 4-bit mode.

For me, Ben's code worked fine at higher clock speeds and after power on or reset, as long as I was using 8-bit mode for the LCD. It was only once I switched to 4-bit mode that I discovered the need for the extended "3x" initialization sequence which handles the LCD being in an unknown state re 8-bit vs. 4-bit mode depending on power on vs. reset. In my experience, as long as the LCD is never put into 4-bit mode, the shorter initialization sequence is reliable since it can count on the LCD being in 8-bit mode when it runs.

[GARTHWILSON]

pjdennis wrote:

Paganini wrote:

If I remember correctly, Ben's code worked on Blue April but only at power up. The power up met the LCD module's setup requirements, but a system reset did not; so it would stop working after a reset. Using the full setup sequence Garth described along with some appropriately proportioned delay loops let me run Blue April at 8Mhz (instead of 1Mhz!) and also switch the LCD to 4-bit mode.

For me, Ben's code worked fine at higher clock speeds and after power on or reset, as long as I was using 8-bit mode for the LCD. It was only once I switched to 4-bit mode that I discovered the need for the extended "3x" initialization sequence which handles the LCD being in an unknown state re 8-bit vs. 4-bit mode depending on power on vs. reset. In my experience, as long as the LCD is never put into 4-bit mode, the shorter initialization sequence is reliable since it can rely on the LCD being in 8-bit mode when it runs.

When we had the problem at work, we were using only 8-bit mode.  The problem was solved with the 3x initialization.  I did not try 4-bit mode until a few years later.

[pjdennis]

GARTHWILSON wrote:

When we had the problem at work, we were using only 8-bit mode.  The problem was solved with the 3x initialization.  I did not try 4-bit mode until a few years later.

The Hitachi datasheets currently available online (excerpt below) state that the internal reset can fail if power supply conditions are outside the supported envelope. So that is a documented scenario in which the 3x initialization (which the datasheet calls 'initializing by instruction') is required. Of course I have no idea if that could have been what Garth experienced back then at work.

That said, I think OPs main problem (as I posted earlier) is the code they are using does not include the LCD busy flag checks - the code they posted is an interim simplified version designed for a 'slow' clock module. I know from experience it produces garbled output when run at 1MHz, which happens because at that clock speed it sends commands to the LCD controller while it is still busy executing prior commands.

[GARTHWILSON]

I dug up and examined the old circuit, and I calculate that the power should have come up on the LCD in about 0.25ms.  I don't have any of the old hardware here to measure.

But yes, it is imperative that you check the busy flag if there's a possibility of going too fast for the LCD to process the inputs.

[mort]

I know its been a while, but I've just started tinkering with the ben eater 6502 and ran into the same issue as the Op. Ended up being the LCD display for me. Replaced the green one that came with the kit with a blue one that I found laying around with the same "1602A" labeling on the PCB, and everything worked as it should. Just wanted document what I found for future people.

[GARTHWILSON]

Welcome, mort.

mort wrote:

I know its been a while, but I've just started tinkering with the ben eater 6502 and ran into the same issue as the Op. Ended up being the LCD display for me. Replaced the green one that came with the kit with a blue one that I found laying around with the same "1602A" labeling on the PCB, and everything worked as it should. Just wanted document what I found for future people.

Make sure you're using good static-handling precautions.  We occasionally get someone thinking their IC is bad, or saying that they changed a suspect IC and that made everything work; but I have been responsible for hundreds of thousands of ICs we've used at work in our products, and I have a few thousand at home in my personal inventory too, and to my knowledge, we have never gotten a bad one from a legitimate distributor.  (I say "legitimate" as opposed to something like eBay for example.)  IOW, it's not like buying produce at the supermarket where you know that once in a while you'll cut into a piece of fruit and find that it's bad inside, and that's just the way it is with produce.  The ICs' manufacturers test thoroughly and they and the distributors handle appropriately, so you can be sure the parts were good before we take them out of the tubes or other packaging.

The 6502 Primer (which Ben Eater seems to have gotten some of his ideas from) has a page on 74xx logic families and timing margins, and about 2/3 of the way down the page it also has a few paragraphs addressing CMOS static-handling precautions.  The Primer is about building your own 6502 computer, and has 22 logically organized sections which I hope you'll go through more or less in order.

[Dr Jefyll]

mort wrote:

Ended up being the LCD display for me. Replaced the green one that came with the kit [...] and everything worked as it should.

Welcome, mort. Glad you got the LCD working.

Just a reminder, though... If you haven't already, and before drawing any final conclusions, it would be good to try the original LCD again, because there's a chance it's actually not the source of the trouble.

As you may know, what can happen sometimes is, you replace a suspect part and of course in the process you wiggle or jiggle the associated connections. And it's the wiggling and jiggling that "cures" the problem, because there's a bad connection somewhere. (Bad connections are a VERY common cause of trouble.)

-- Jeff