LCD Display Replacement Head-scratcher

[helius]

I think if you don't have a good desoldering tool, you should probably invest in a set of hollow stainless needles. The idea is that you melt the joint, then push the needle through the hole around the pin. Solder won't wet to stainless, so what this does is separate the pin from the hole. These sets were available for $1 delivered a few years ago, and might still be sold cheaply from some vendors.

(The latest trick the ebay/aliexpress sellers use to get subsidized shipping to the West is to use freight forwarding from central Asia, which still has the lowest rates in the International Postal Union.)

[tooki]

For anyone still following, it looks like my assumption about the charge pump produced negative voltage was correct. I have reopened the unit and removed the LCD, and there are traces taking the pin 19 output directly to the trim pot, then from there back over to the V0 contrast adjust pin. Not only that, but the replacement display i bought does not have the same configuration as the original one, and is missing the charge pump and capacitors necessary to provide that voltage. So what was being sent out of the LCD i put in was in fact 5v, which i think is coming from the backlight V+ on the main PCB. This means the contrast can't ever get low enough for the difference to be enough. In fact, it's worse, as the LCD is being powered from Vcc of 3.3v, which means that the potential 5v going to the contrast adjust pin is out of spec by more than a volt over the max expected. I can't get the replacement display working in my test rig either. Perhaps i busted it. Will see.

Next i'm going to hook the original LCD back in to the unit and see if it actually works or not. Just waiting for some jumper wires so i don't have to solder it in again. The thing is a total nightmare to get out. Took me almost 2 hours to remove the LCD using solder braid and some quite forceful rocking back and forth..

Many LCD modules have an unpopulated footprint for an ICL7660 charge pump circuit for exactly this purpose. See if yours does.

FYI, i really suggest against “forceful rocking” — you could easily rip out a plated through hole. If you don’t have the tools to desolder a pin header (or other multi-pin device) properly, you’re much better off cutting the header pins so you can desolder them individually.

[gnuarm]

Well, it turns out there's a negative voltage being generated on the LCD board from a small innocuous charge pump sitting there, and routed out to the pin on the main header marked as backlight anode on the datasheet. There's no mention of this pin being for that purpose, though there is a note 'negative voltage optional for 3v supply' on the datasheet, and that's all. Turns out it's perfect for setting the contrast, in particular when running the LCD on 3.3v, the datasheet that says there should be 4.45v typical between Vdd and V0. It took me ages to realise this is a differential measurement, and i was getting very confused that contrast was working even when i fed no voltage to the pin directly. It seems that it's referenced to Vdd and obviously that requires V- when running on 3.3v. So i now have an idea that the main PCB could be drawing this negative voltage from the LCD and putting it on the trimmer. I'm going to see where the pin leads off to, and see if i could have cracked and traces or something like that which could have killed the contrast circuit. Fingers crossed.

The old LCD which i pulled from the machine is definitely not damaged, in fact it's working within spec pretty well. Glad i made this test rig and stuck with it. Seems to have paid off, at least, partly!

I discovered this myself recently.  The contrast control on multiplexed, passive LCDs is the end of a chain of resistors that set the drive voltages for multiplexing the LCD segments.  The other end of this resistor chains is tied to Vdd.  In 5V modules the end of this resistor pin, I'll call Vo, is left for the user to connect to a pot between 5V and ground.  However, it is just as effective to just use a pot between this pin and ground without a 5V connection.

On 3.3V modules there is not enough drive to bias the display for optimum contrast.  So they provide a negative voltage generator which can be used in place of ground to use with the adjustment pot.  Otherwise the 5V and 3.3V modules are identical.  In fact, the 5V modules have the negative voltage generator on the design, just not populated.  I expect they change the value of the current limiting resistor for the backlight. 

I'm working on a project that is using a 5V display because they started with an Arduino.  Now everything else is 3.3V so the 5V level shift uses two more chips.  Rather than that, I want to run the module from 3.3V.  So I'm adding a 7660 chip to the board in place of the two level shifters. 

[gnuarm]

One thing I didn't see when reading the posts is what the original problem was that prevented the original display from working?  I read that you discovered the negative voltage circuit on the display which is missing from the replacement displays.  But what was wrong that prevented the original unit from working? 

[al_m]

I think if you don't have a good desoldering tool, you should probably invest in a set of hollow stainless needles. The idea is that you melt the joint, then push the needle through the hole around the pin. Solder won't wet to stainless, so what this does is separate the pin from the hole. These sets were available for $1 delivered a few years ago, and might still be sold cheaply from some vendors.

Wow, thanks for the tip. I had never heard of such a thing before! I see i can get a set for less than 5 pounds on ebay. So will buy a couple of types and see how i get on. Hopefully they could be useful for future work (i have a large 4x DIP DRAM desoldering job ahead that i have been putting off because i know it might not go well with the tools i have to hand...)

Many LCD modules have an unpopulated footprint for an ICL7660 charge pump circuit for exactly this purpose. See if yours does.

Yes, the replacement i bought has an unpopulated area with pads for a SOIC, which is filled in on the original with the charge pump. I'm going to order the components from digikey and see if i can work out the correct jumper positions to get it working. Will also contact winstar to see if they can advise, but not holding out much hope for that...

FYI, i really suggest against “forceful rocking” — you could easily rip out a plated through hole. If you don’t have the tools to desolder a pin header (or other multi-pin device) properly, you’re much better off cutting the header pins so you can desolder them individually.

Yeah, i'm against using force to remove components, but in this case i had no better option, and the PCB i was rocking was the LCD one trying to get the header pins up off the extremely tight through-holes. It worked out ok, but i do know better, despite all my seeming naïveté. I'm going to look into the needles that helius recommended, as that might save any rough stuff in future.

I discovered this myself recently.  The contrast control on multiplexed, passive LCDs is the end of a chain of resistors that set the drive voltages for multiplexing the LCD segments.  The other end of this resistor chains is tied to Vdd.  In 5V modules the end of this resistor pin, I'll call Vo, is left for the user to connect to a pot between 5V and ground.  However, it is just as effective to just use a pot between this pin and ground without a 5V connection.

It was making the connection between there being no voltage 'put' on the pin and the connection to ground working well, that made me realise it's not about voltage going in, but the difference between the two ends. A tricky point, and not obvious at all from first glance at the datasheets i've been working from (don't get me started on the winstar one, the word datasheet is a compliment to that POS).

One thing I didn't see when reading the posts is what the original problem was that prevented the original display from working?  I read that you discovered the negative voltage circuit on the display which is missing from the replacement displays.  But what was wrong that prevented the original unit from working? 

The short answer is: i don't know. It could have been something as simple as a bad solder joint. The fact the contrast trimmer had no effect made me panic and suspect LCD failure, but that turns out not to have been the case. I contacted the manufacturer and they told me to look at an SMD ferrite bead in series with the power line as a point of failure, but when i told them what voltages i'm reading they said it looks good and is unlikely to be that component. So, I'm waiting on some dupont jumpers to arrive after which i can do a solderless install of the original LCD and see if it is in fact working in circuit or not... From my readings and being able to trace the circuit, it looks hopeful. Will report back.

[Ian.M]

If the holes are tight enough that square pins are effectively a press fit, with their corners biting into the hole plating, stainless hollow needles wont help.  Short of cutting the pins, there aren't many good options.   If you *MUST* preserve both the pins and board, probably the best option is one of ChipQuik's low melting point desoldering alloys, which in typical usage will lower the melting point of the solder joint to lower than a fresh cup of tea!

Remove as much solder as is reasonably possible with flux+braid or a sucker, flux and apply the desoldering alloy with an iron, then ideally keep all the pin joints above melting point with a board preheater, hot air pencil or hotplate, all at or near minimum temperature while you gently work the pins out of the holes.   Its more of a PITA if you have to run an iron up and down the row of pins to keep the joints molten, unless you can enlist a second pair of hands to assist.

[al_m]

If the holes are tight enough that square pins are effectively a press fit, with their corners biting into the hole plating, stainless hollow needles wont help.  Short of cutting the pins, there aren't many good options.   If you *MUST* preserve both the pins and board, probably the best option is one of ChipQuik's low melting point desoldering alloys, which in typical usage will lower the melting point of the solder joint to lower than a fresh cup of tea!

Remove as much solder as is reasonably possible with flux+braid or a sucker, flux and apply the desoldering alloy with an iron, then ideally keep all the pin joints above melting point with a board preheater, hot air pencil or hotplate, all at or near minimum temperature while you gently work the pins out of the holes.   Its more of a PITA if you have to run an iron up and down the row of pins to keep the joints molten, unless you can enlist a second pair of hands to assist.

Thanks, Ian. Yes, i was wondering about that. This unit does seem to want you to use snippers to get the thing disassembled, it's bloody hard otherwise, everything is really tight, including the self fastening standoffs. I can see how the needles could be of use in other projects though, so thought it worth a punt for the low cost.

I had considered using ChipQuik, but it's quite expensive, i've never used it before, and i couldn't find a good place to source the correct type fast enough. It would probably help a lot though, as would a fresh cup of tea!

[Ian.M]

The trouble with using dykes to snip the pin, especially on straight pinned multi-way connectors is they can deliver a severe mechanical shock to the board.  Also, sooner or later, you'll encounter a tin or gold plated hard steel pin and wreck your *good* precision dykes.   Depending on access you may be better off grinding through the pins or the body where the pins emerge with a Dremel cutting disk.  Some care is needed to avoid the cutting disk wrenching unsupported pin remnants - solder bridging them all so they gain mutual support may be advisable for SMD pins.

Yes, ChipQuik desoldering alloy is expensive.  You certainly wouldn't want to use it for every part you need to remove, but you only need enough to go along all the pins of the chip, so 2" will do a medium size TQFP, and it actually works out under $0.60/inch.  Also, as has previously been discussed here, you can save the 'blob' of used ChipQuik alloy and reuse it for less critical jobs several times, dropping the price per chip removed even lower.   Not having to   scour EBAY for a donor unit because youve wrecked the PCB is priceless . . .

[gnuarm]

If the holes are tight enough that square pins are effectively a press fit, with their corners biting into the hole plating, stainless hollow needles wont help.  Short of cutting the pins, there aren't many good options.   If you *MUST* preserve both the pins and board, probably the best option is one of ChipQuik's low melting point desoldering alloys, which in typical usage will lower the melting point of the solder joint to lower than a fresh cup of tea!

Remove as much solder as is reasonably possible with flux+braid or a sucker, flux and apply the desoldering alloy with an iron, then ideally keep all the pin joints above melting point with a board preheater, hot air pencil or hotplate, all at or near minimum temperature while you gently work the pins out of the holes.   Its more of a PITA if you have to run an iron up and down the row of pins to keep the joints molten, unless you can enlist a second pair of hands to assist.

I would wonder about the residual contamination.  But I suppose that's not much of an issue and a noop for home use.  Interesting.  I've never seen this before. 

[Ian.M]

The melting point is so low you can scrub most of the used ChipQuik off the pads with brush followed by a cotton bud, and easily suck through holes or clear them with a cocktail stick.  SMD pads should be clean enough to immediately resolder for all except the most demanding applications.  As PTH pads will trap more residual ChipQuik, its advisable to 'wash' them with the correct solder alloy for the board, removing it by suction or with flux & braid, to reduce the contamination to acceptable levels before fitting a new part.

[al_m]

I'm glad this has turned into a productive discussion rather than just me grumbling on about my wayward handiwork 

If there is still residual ChipQuik remaining, is there a risk that it will lower the melting point of solder joints to the point where normal thermal transfer from electron flow would remelt the joints?

[gnuarm]

I'm glad this has turned into a productive discussion rather than just me grumbling on about my wayward handiwork 

If there is still residual ChipQuik remaining, is there a risk that it will lower the melting point of solder joints to the point where normal thermal transfer from electron flow would remelt the joints?

I think for professional work that might be a concern, but as Ian says, good practices can make it workable.   I would not be concerned about normal use.  Letting your devices run at the temperature of boiling tea is a bit more than expected for most devices.

[al_m]

I suppose so. And we're also talking about a case in which that specific product is only being used in a rework/repair scenario. If it was that critical it may just be better to do a full board replacement, where possible at least.

[helius]

The lead-containing ChipQuik is Cerrolow 136, at a markup of around 12x for turning the bulk alloy into wire solder. The "other metals" in 136 are bismuth and indium, which will not ruin a joint if present as impurities (indium is compatible with most alloys, while too much bismuth can make joints brittle). So while the used ChipQuik should be cleaned away, it won't prevent components or pads from being re-used. A good practice for surface-mount pads is to put a drop of normal solder on the iron and "scrub" lightly over the pads to re-coat them with the correct alloy.

The same alloy is also sold as LowMelt by Zephyrtronics. As Ian.M alluded to, there are ways to re-use the low-melting alloy as well. If kept warm, it can be pushed aside with a swab onto an area of soldermask, and then lifted away.

[al_m]

Well chaps, my dupont cables arrived today, and i hooked up the original LCD as planned. All is working fine. So what the heck went wrong is pretty much an unanswerable question now. Possibly a bad solder joint? Or something else. Really not easy to say, and since i did try the contrast trim pot at the time and nothing happened, i would guess at a bad connection being likely somewhere. Unfortunately i was in a bit of a panic and decided that the worst must have happened rather than taking it slowly and methodically tracking down the fault.

Anyway, thank you all for your help. My next step will be installing ultra-low profile pin headers to the LCD and control pcb so that i don't have to go through that desoldering madness again when i want to install another display, and then eventually converting the replacement i bought to work on 3.3v and putting that in there. All the parts have been ordered and are on their way. Hope i can get it done next week.

[YetAnotherTechie]

Some other effects that are confused with "burn-in" are tunnel vision, caused by degrading polarizer film; and "ant tracks" which are solid agglomerations of crystals. Both can be fixed with the proper tools and techniques.

Can you please elaborate on how to fix the "ant tracks"? it's the first time i read about it being possible.

[helius]

It may not be possible in every case. The idea is to pulse the polarity of the display at around 1 Hz while heating to about 50°C. Liquid crystals are by nature amphoteric, and if opposite charges get too close together they can "solidify" into an ionic bonded solid. What you want is to loosen those bonds by adding kinetic energy (heating) as well as pulling the charges around to separate them (pulsing).

[YetAnotherTechie]

Would it also require a drive pattern that "moves" the cristals across the panel away from the rat-track, or would they naturally disperse after de-bonding ?

[helius]

I did not experiment with that, but from what I understand it would not help. The crystals are normally adhered to the (electret charged) inner surfaces of the glass and should not be migrated horizontally or they will not work properly.

[al_m]

Just a final note to say i got the original display reinstalled using low profile sockets and header strips, and it's working as well as it did before i started the work on the unit. I have no idea what went wrong originally. But anyway, i have been sent schematics from Winstar as to how i should do the conversion work to add the negative voltage rail for the contrast offset to my replacement display, so can get on with that when time permits, and all that needs be done is to pull out the old LCD and put the new one in, solderless install (except for soldering the header to the LCD board).

I also worked out a very effective way to remove the standard header strips by snipping the plastic body of the header at the last pin and heating the joint on the underside while pulling the pin with pliers. Worked down pin by pin from either end, and then all that was left to do was to remove old solder from the empty mounting holes, which was fairly easy with my limited tools. It seems a much gentler approach, and it was certainly easier, if more destructive of the header strip.

Thanks for all the advice and help guys.

[tooki]

Glad to hear it’s sorted out!

As for cutting up the header to desolder it: so, like, what I told you to do back in reply #26?

[al_m]

As for cutting up the header to desolder it: so, like, what I told you to do back in reply #26?

Ah, yes, heh.

I don't think i quite got what you meant, or wasn't really paying attention when i read that reply. It was a very helpful technique though, so you're welcome to get the credit for the idea!