Trying to figure out if a PLASMA/Neon display/driver is broken (74 series logic)

[factory]

Soldering looks fine to me. 

How much current was the PSU with reversed connections capable of outputting? There was a legend of someone doing this at work years ago, but with a high current PSU, all the ICs blew up one after another. 

For all neon based displays & lamps, current limiting is necessary, running them higher currents shortens their life, or no current limiting things tend to flash over & melt.

Did you try contacting the company that made it before you started? To see if they still had any data for wiring these up. 

David

[inse]

I have rarely seen solder joints failing on plated- through pads, only on single sided boards without plating - but who knows.
I suggest to unsolder the chip, repeat measurement on chip and board, solder a socket and place or replace the chip.

[akasaka]

factory

It's not the soldering is bad, but this kind of inconsistency is weird. I would have thought the thermal mass of the Gnd line (not plane!) could have caused that, but then it would have been on all of them. My hypothesis is that once this board, or whatever wired to it, failed — the current was high enough to heat up those particular chips and melt the solder.

So all those chips with this symptom are likely to be failed, but not all failed chips will have this symptom

Inre: contacting the company, Japanese companies are notorious for not wanting to give out schematics or service manuals for their products, especially in such industries like transportation.

inse
See above — it's not the solder joint that failed, but the chip went so hot it melted the solder.
Rather than taking the chips off (which is a pretty PITA, that one 107 took me an hour or something), I'll most likely cut the traces carefully, measure, and then mark the chip as to be swapped (or good), and join the trace back with solder.

[akasaka]

(using this message as my logbook and editing as I desolder chips and check them)

* 107-1: dead short in to out (1K to 1Q, pin 4 to 3) FAILED
* 107-2: 1J to ~1Q is 15k Ohm; 1Q/~1Q to GND is 40k Ohm, 2CLK to Vcc is 2k Ohm FAILED
* 107-3: 1Q to ~1Q is 400 Ohm   FAILED
* 393-1, 393-2: 1A input to Vcc is 2k Fault gone after removing offending 107-1/2/3 it seems..?
* 32-1: 1A short to 1B gone after removing 107-1/2/3... TBC:  3A has a weird pulse, 4A is also weird
* 04-2: 1A kinda low circuit related to 107-1, to be rechecked, Vcc to 1B is 13kOhm needs checking is the resistance of the blanking circuit
* 10-1: suspiciously low resistance to power on pins 1, 3 that's in the board, flux residue??

Looks like I only need a bunch of 107s?? Dang I wish we still had electronic parts shops here in Sapporo...
We have one, but it downsized recently, and now most has resistors, not even all kinds of them, and works Mon-Fri 11am-4pm :/
Would be a major PITA if I missed something

[factory]

factory

It's not the soldering is bad, but this kind of inconsistency is weird. I would have thought the thermal mass of the Gnd line (not plane!) could have caused that, but then it would have been on all of them. My hypothesis is that once this board, or whatever wired to it, failed — the current was high enough to heat up those particular chips and melt the solder.

So all those chips with this symptom are likely to be failed, but not all failed chips will have this symptom

Inre: contacting the company, Japanese companies are notorious for not wanting to give out schematics or service manuals for their products, especially in such industries like transportation.

inse
See above — it's not the solder joint that failed, but the chip went so hot it melted the solder.
Rather than taking the chips off (which is a pretty PITA, that one 107 took me an hour or something), I'll most likely cut the traces carefully, measure, and then mark the chip as to be swapped (or good), and join the trace back with solder.

Think about how the wattage & temperature required to melt solder, if the IC was capable of that with only a 5V supply (which is very unlikely), then it would have destroyed itself in the process.

You should probably update the thread title, now you've found out it isn't a VFD type.

David

[akasaka]

factory

I presume this was overvolted at some point due to a PSU failure or something, given the stuffed ROM on the other board. Definitely didn't happen on a 5V supply.
And given the shorts and whatnot, it did indeed destroy itself, at least some of them did :-)

Thanks for the tip, renamed the thread, didn't know I could do that

[akasaka]

So well... It's *almost* alive!



I can see the sweep in the "dead" rows, so it's either something along the data bus being dead, or my test jig made out of two hex spinny things is not making good contact — the tube is most likely alive.

The panel plus it's boost converter eats a whoppin' 25W in this state — guess making a wall clock out of this is going to cost a pretty penny in electricity :/ Not to mention the flicker.

Also the logic alone still eats around 3W, which IMO is too much for what it is — I'll have to look more into what might be broken.
Otherwise replacing 4 of the 157 triggers and a 393 counter seemed to have mostly fixed it (and I'm not even sure about the counter, it just looked sus).

Along the way I've figured out the pinout of the data connector. Looking into the connector on the board, starting from the top right pin (where the white arrow is), counter-clockwise:

1 to 8. Bitmap data input
9. (output)
10. ~CLK
11. ~RESET
12. ~RETZ
13. ~READY
14. BRIGHT
15. SHOW
16. Common

After powering on, wait until ~READY goes low.

To draw something on the display, you set the data pins to the top left 8 pixels and pulse ~CLK low. Then bottom left 8 pixels and pulse again. (or the opposite, depending on how you rotate the tube)

To hide whatever is shown on display, you pull SHOW low. That blanks the screen while letting you draw into the framebuffer. Also you can pull BRIGHT low to reduce brightness to about half of the normal level.

If you want to start drawing from scratch, pulse ~RETZ low. This resets the pointer to the active byte to 0 and you can draw from the top left again.

If all things went sideways, pulse ~RESET low and the board initializes itself, clearing the screen in the process. Of course then you have to wait until ~READY is low again before sending any data.


Now here's a question: while drawing this picture on the display the pixels in the middle struggled to light up until they went to around 2/3 through the screen. I keep the HV supply at 160V. Should I be increasing the HV, or is something in the drive circuitry still odd?


Upd: the dead row of pixels was just a wire falling off from my hex input test jig, reattaching that fixed it. I'll buy proper ribbon cable on my trip to Akihabara next week

Looks like it's all working now! Time to buy some shift registers and start coding, sawing, gluing and whatnot!

Thanks everyone for your help!!

[akasaka]

So I have started putting together an enclosure for this, but I can't help but wonder — is this faint pattern normal, or am I missing some other wacky chip on the board?

It is highly contrasted compared to actual lit up pixels, but pretty bright in a dark room.

Or is it the "standby" discharge to keep the gas ionized to allow pixels to toggle faster? Is that even a thing in plasma displays?

(to those wondering: PIS-DOS stands for Plasma Information System DOS )

[pcprogrammer]

(to those wondering: PIS-DOS stands for Plasma Information System DOS )

Nice acronym. 

Matches the idea of every OS sucks. (There is a song made about it. Google "every os sucks")

No idea about the glow. Maybe the voltage needs lowering?

[akasaka]

The voltage as is seems to be low, the tube sometimes doesn't strike when powering on from a cold state, and big blocks of solid color also cause some flicker. I'm keeping it at 160V-ish.

Mind you I've also swapped the 5.12MHz xtal on the sweep board for a 12.048MHz one to get rid of the flickering on camera and in peripheral vision. But those small dots were there from before.

Either way, if you guys are curious how the thing looks now, here is a small video of the WIP:

 

Nice acronym.  Matches the idea of every OS sucks.

Well, given the code (https://github.com/vladkorotnev/plasma-clock/tree/develop/src) quality, I think it describes that best

It seems that I got laid off by my employer this week, so on one hand I can't buy parts to complete the physical build (acrylic is quite expensive, and I need some flat top screws). On the other hand that might give me some time to polish up the code inbetween the interviews, or at least get all the planned features coded in!

[pcprogrammer]

It seems that I got laid off by my employer this week,

Oh man that sucks. Hope you are able to find a new job quick enough.

[akasaka]

Final and last update to the thread. I have successfully built the weather station / clock I've wanted to make out of this panel.

Please check out this video overview showcasing the super cute weather icon animations, as well as the general look of the device, along with a song I composed recently on my DX7 and CX5F :-)



Thanks again to everyone for your help!

[shapirus]

Well that's awesome. Congratulations!