LCD design, is this good enough?

[oPossum]

(BTW, is there any way to have an inline image where the image is an attachment rather than a link to imgur which can go away at any moment?)

It is a two step process. Make the post with attached images, and then edit it to use the attachment URLs for inline images.

[robertbaruch]

I also asked to change the operating voltage from 5.2V to 2.6V. I mistakenly thought that applying an AC waveform between -2.6 and +2.6 would result in an operation voltage of 5.2V, but it is actually 2.6V -- the segment is on at 2.6V, but you flip the polarity.

[kony]

Word of advice - if someone dun fucked up something mechanical, yet provided proper 2D drawing for reference, always do the corrections graphically as overlay (or better yet in DXF if the original is provided) with marked and updated values of wrong dimensions and added new ones. Especially if there is languague barrier. Text directions can be ambiguous, and much harder to navigate.

[Refrigerator]

Oh boy, this thread is some good stuff.

[wrickert]

That is a great idea. I have been playing with the idea of a smartwatch as a conference badge. I have been struggling to find a cheap display for cheap enough. This might be the perfect solution. I can't wait to hear more.

[robertbaruch]

It is a two step process. Make the post with attached images, and then edit it to use the attachment URLs for inline images.

Works, thanks!!!

[ebastler]

Great to see how you are following through with this, Robert!

But I am also impressed by the professional support provided by your "supplier B", DDX. You are getting good value for your 320$ tooling and NRE fee. If the final product is as convincing as the dialog to plan and specify it, the result should be really neat!

Supplier B replied, saying that I really wanted STN rather than TN, I said that was fine. They also were willing to meet my price of USD 2.50/pc at 200 pcs. The "mould" fee would be USD 320. This is definitely within my budget!

[robertbaruch]

Last round of changes are in, and it's looking good! I signed the first page and sent it back, and now we wait 4-6 weeks for the samples!

I have learned:

1. Sending a multi-page spec is better than a single page.
2. Using a matrix for the pins instead of a pinout is apparently standard for LCDs.
3. Don't spell anything else out in tables. Put all the measurements on the drawings.

[james_s]

Even if the final product is junk, there's probably $320 worth of education in this thread. Prior to this I had no idea custom LCDs were even remotely feasible in volumes less than tens of thousands.

[krasin]

This. I am surprised that getting a custom LCD is projected to be cheaper than a custom injected molded case.

I really hope this project succeeds and the experience is documented, so that others can follow.

[robertbaruch]

Thought I'd post a description and pic of the test jig I'm going to use for testing the LCD samples. First, I'm just going to set up a simple square wave and test one segment, see what the threshold is, that sort of thing. Then I'm going to use this jig, which has 144 pogo pins on it. The pogo pins are 0.65mm in diameter, and the tips are rounded. I did check to see if the tips would scratch the LCD pads on my dead LCD, and they didn't seem to do any damage! Finally, I'll test one of the LCDs in a real TRS-80 PC-1.

Why pogo pins and not an elastomeric strip? Well, I don't have any extra elastomeric strips, and I don't find them easy to use. They're a bit fiddly.

So I 3D printed this holder for the pogo pins, and I have two circuit boards. One is a more general driver, and the other is the adapter for the specific LCD.





The general driver uses a single PCF8545B LCD driver from NXP. This driver is "universal", being capable of up to 8 commons and up to 40 rows (i.e. 320 segments), with driving voltages from 1.8 to 5.5v, biasing of 1/2, 1/3, or 1/4, multiplexing of 1:4, 1:6, or 1:8, and a range of scan frequencies from 60-300Hz, AND two types of output waveforms.

Sounds awesome, right? Well, sadly, unlike a lot of their other chips, this one is not cascadable, meaning you can't add rows. If you wanted to add rows, you'd have to make sure the chips are synced in their commons, and that's what cascadable chips can do. Not this one, so you can't just use four of them, their commons wouldn't sync. Sadly, NXP does not make a chip with 8 commons that is cascadable. If you happen to know of one, let me know!

So instead, I'm resigned to testing only one portion of the LCD at a time. My LCD has 4 "sections": 3 sections of 40 rows for the character display, and 1 section of 9 rows for the indicators. I connect a FSAL200 quad analog multiplexer to each row, and I set up the test jig to have 4 sections of 40 rows. That's 160 rows, but 40 quad multiplexers. The multiplexers switch either the signal from the LCD driver, or 1/2 the LCD voltage, which is the "off" state of a segment. In this way, I can switch the LCD driver to drive any one of the 4 sections of 40 rows.

The whole thing is controlled with, of course, an Arduino (Pro Mini, 5V). There are four buttons, a bunch of LEDs, and a 4-digit 7-segment display. With that, I can program the LCD driver to the right bias, LCD voltage, multiplexing, waveform, and frequency, and then step through the sections to test. I decided to use a DAC instead of a low-pass filter on the Arduino's PWM output because I wanted a more stable voltage.

Anyway, the factory says they'll have the samples ready by the end of the month, and theoretically it will take 1-2 weeks after that to get here.

In other news, I attempted to set up a CrowdSupply crowd-sourcing campaign. You have to fill out a form with all the details of your project, how much you want to crowdfund, what you're going to spend it on, and so on. I think it's a higher bar than Kickstarter or IndieGoGo. Reasonable fees, too, and they handle fulfilment. However, that was two weeks ago, and they claim to get back to you in a week. So... I don't know how that's going to turn out. I do want to set up a crowdfunder for this.

[robertbaruch]

I have the sample LCDs! Running my fingers along the contacts (grounding one of the commons) shows all the segments seem good and in the right place. The size is also correct.



There's just one problem. The PCF8545 LCD driver doesn't seem to be behaving as expected. I can program it via SPI, no problem. The common pins ("BP", or "backplane" in the datasheet) have the correct waveform. However, the segment pins don't behave! They are supposed to vary between 0v, 1/2 VLCD, and VLCD. In fact, at VLCD = 2.5, they vary between 0v, 3.3v, and 2.5v. Yes, every time the segments are supposed to be at 1/2 VLCD, they go to 3.3v!



This is a problem because there are protective diodes going from the segment outputs to VLCD, so they conduct! And yes, the chip gets hot!

I changed VLCD to 4.0v, but now the chip outputs 4.6v when it should be at 1/2 VLCD. So we have a situation where segments at 1/2 VLCD result in an output of 0.6-0.8 volts above VLCD. It's suspicious.

I'd ask NXP what's going on, but historically they have never answered a question about any of their PCF85xx chips.

There was a suggestion on Twitter that maybe NXP reversed the pins on VLCD and VCC.

[robertbaruch]

Without a functioning LCD tester, I just decided to go for broke and install a new LCD into a known working Sharp PC-1211 (aka TRS-80 PC-1). It works! At least, all the segments are active. However, I need to adjust the contrast. If that doesn't work, then I'm pretty sure that the threshold voltage is incorrect. I can see that when I press the mode switch, the mode segments do change slightly.

[robertbaruch]

And it turns out that for the PCF8545 driver, VLCD must be greater than VCC, which cannot be done here.  So back to the drawing board for an LCD tester.

[robertbaruch]

I damaged the wiper on the contrast pot because it was all painted down. So with the wiper removed (pot at max resistance) I just decided to see what the display looked like.





So close! Those last few matrices have something odd going on with them. I'm tempted to call it flaw in the LCD manufacture, but I really need to replace a second one to see if the effect is replicated.

[Refrigerator]

awesome stuff, these sharps seem quite prone to LCD failures.

[robertbaruch]

I opened the thing back up, placed a paper shim in the bezel to make sure the LCD didn't move from where I thought it should be, cleaned the pads and the elastomeric strips with alcohol, made sure not to touch them, put them in place, and buttoned it all up again. And....



A qualified success! Qualified because (a) I still need to reproducibly replace an LCD without several tries, and (b) you have to replace the contrast adjustment with a 240k resistor.

What's interesting is that on the original, the LCD was placed into the bezel with an adhesive strip, so once it was in the bezel, you couldn't move it. And the original LCD does have a little play in the bezel. I'll have to look more closely at an original to see if they glued the LCD in the same place as I put it.

[barbeque]

Congratulations! That looks amazing.

I would have been tempted to slip in an easter egg in the "mode" line. I remember as a kid holding a lot of LCDs up to bright lights to see what modes I haven't managed to enable yet/weren't offered on my model.

[HwAoRrDk]

Looks great. As Dave would say: "Like a bought one!"

[robertbaruch]

Final video! I put this one up on diode.zone first -- I want to support a non-YouTube video streaming instance.

[coromonadalix]

Man  Impressive work  

Has others wrote,  some  HP Agilent stuff could / can be saved with guidance like this.

[Keatah]

This is really really great news! I have 11 PC-1 units between myself and friends that could use such a replacement! And I wouldn't mind paying about $25 - $30 per display (if that's a fair cost and allows for some personal profit)..

I was almost almost getting ready to cannibalize other units to make 1 working. But now with this project that may not be necessary.

Can you please update us on how this is coming along and if and when some will be available?

---

I got to thinking and wondering if it's possible to specify a more yellow-tinted color instead of the grey/green shown in the videos? For authenticity.. Or might it be as simple as retaining the yellow filter already present?

I also believe there are more than just a couple hundred of these pocket computers out there. I see one or two move for between $30 and $100 per month.

But there's a lot of delicate work involved in the repair. And watching your second video is going to be required. If any obstacle to selling these replacement displays exists it is that and likely not the amount of units out there. And the ones that are still working are likely to fail sooner or later.

[robertbaruch]

I've ordered the run of 200 from the manufacturer, and that should hopefully arrive by the end of May. At that point I'll put up a shop on Tindie and start selling them. I did apply to get CrowdSupply to make this a crowd-sourcing project rather than just a store, but CrowdSupply really dropped the ball on this one.

It's not likely that I'll meet my goal of 2.5x the cost. I probably can't put two of the same thing on Tindie, one for, say, $20, and the other for $40 ($20 for the LCD plus $20 support).

I've found that there are some of the PC-1/PC-1211 that can't be fixed by just replacing the LCD -- there's something more fundamental going on, so an LCD swap isn't a guaranteed fix for *everything*

[trompczynski]

Just came across this when searching alternatives for my PC-1211 once again. I had my head on using a standard N-character LCD and perhaps some low-power LEDs for the other indicators, but man! This is just what I've been dreaming of. I'll probably get a couple of screens when they're out.

We could organize ourselves into a pocket computer enthusiasts group after they're all working, what do you guys think? I've been dying to make this little machine communicate to the outside world through the old cassette interface. Perhaps figure out some sort of really basic modem program using the load and save commands, and audio in/out on the other side of the cable.

Greetings from Brazil, where the PC-1211 was built in rather large numbers. Can't wait to fix mine!

[rkindig]

I'm very curious: how is this project coming?  Almost ALL Tandy PC-1's have the LCD issue.