ICP cert batteries depleting

[Randy Stankey]

I'd be interested in knowing where the cert battery traces end up and see if a temp battery could be attached there for the duration of the battery swap.

You're thinking the same thing as me.

Ambiguities of how the circuit board works aside, I don't think connecting a temporary battery would be too difficult for a proficient technician to do. Then the old battery could be removed and a new one swapped in.

Just from the pictures, it looks like the positive terminal of the battery is connected to the circuit trace that leads to R581 which, I assume, leads to the backup power terminal of U47. (*) I don't see a reason why it wouldn't be possible to use a hot air soldering station to tack a wire onto the solder bead on that resistor. I can't tell where the negative terminal leads to but it's got to go somewhere that a wire can be tacked onto. It would be great if you could find a via-hole, somewhere, to use.

Once that is done...assuming all the foregoing is true...it would be a simple matter to clip out the old battery, clean up the pads and/or through holes and solder in a new one.

If you made sure to wear an ESD smock with grounding straps and used an ESD-safe soldering station, I don't see why this couldn't be done.

I think of it this way. If the device is 100% certain to fail when the battery goes bad, even if you have a 50/50 shot at preventing it, you're still ahead of the game. If you are going to have to spend $4,000 when it fails, successfully replacing the battery saves you money. Messing up the repair doesn't cost you any more money than you would have had to spend in the first place.

The only difference is that you get to decide when to make that expense instead of waiting for the other shoe to drop.

In my opinion, we can do some more research to find out how this repair can be done, if it can be done, then come up with a plan of attack and do it. Have a backup plan in case the experiment fails then give it a shot.

I see this as a situation where there is little to lose and a lot to gain if only the right people could apply the right combination of brainpower, skill and work.

-----

(*) I'm sorry. I know it's obtuse but I can never see the name "U47" without being reminded of Frank Zappa. "It looks just like a Telefunken U47!"

[Mark Gulbrandsen]

I can GUARANTEE you that board is at least 4 layers if not more.It's way too complex in nature not to be at least multi layer. So where you "think" a trace may go is likely not where it actually does go. Multi layer boards have been around and in general use for a long time now.

Steve, I was told by NEC that there is a super capacitor on there for both cert and clock. Thats why they have you run the projector for a bit before hand.

[Steve Guttag]

There is no super-cap. I've changed too many clock batteries and know that the moment you pop that battery that the clock is reset. On Barco, one has to reset that clock themselves...on NEC and Christie, they pull the time from other things to automatically reset the ICP clock.

I'd be amazed if the ICP isn't 6-12 layers or more. However, with SMT...all of that stuff ends up on the outer layers (there are no chips hiding mid layer). Even with Through Hole, still everything, ultimately ends up on an outer layer. The trick is in finding it but, perhaps it stays on the outer layers. If the clock connects to a ground then it likely dives to an inner layer as it is common practice to make the two two most inner layers the rails to ensure you have power distributed as evenly as possible across the board and get a bit of a capacitive effect from the two layers separated by a thin insulator (the board). One really doesn't want to run power via traces since there can be voltage drops from one end of the track to the other as various things tap off it it.

[Carsten Kurz]

It is true that NEC has passages about a bridge/sub battery in their ICP maintenance manuals, they describe the soldered-in battery as a backup cell. But they are wrong.

Bildschirmfoto 2021-02-05 um 15.43.57.png

When the socketed battery is drained, the clock is gone. When the soldered in battery is depleted, the cert ist gone. They don't sustain each other. I don't think NEC uses special ICPs in their systems, and it would be easy to compare these areas of NEC ICPs with those from Christie or Barco projectors. I guess that somehow that documentation has been mixed with their cinema controller board (which has had it's own battery issues). There is a NEC support document 'The ICP board and you' that shows how to swap an ICP board. The ICP photos in it are bad, but they show exactly the same board and component layout that we see on all other ICPs. Both cells are BR2032, they are not rechargeable.

It is enough risk to solder a new battery onto the existing battery contacts - bottom or top layer. Trying to replace the original battery by removing it completely, just to have it look 'pretty' afterwards is not worth the effort. To me from the photos I see, it is obvious that the positive of both batteries has traces on the top layer. You can even recognise the Maxim DS3232 clock IC where the socketed cell positive goes to. Where the positive trace of the cert battery ends up, I don't care. It is too dangerous there. This is not a regular board repair, it is coronary surgery on a living species.

No need to to go there, the through-hole pad of the positive has ample estate to solder a wire to it. If at all you could scrap off some of the solder mask of the relatively broad + trace and connect a thin wire there. As the certy supply current is minute, even the thinnest wire will do (wirewrap or patchwire).
When you look at top and bottom side, it is obvious that the -/GND of the cert battery is connected to an inner ground layer. It is probably easy to find the same ground within that area on the top side as well. Again, a dead ICP would be handy to assure the process.

It is possible, that the two-pin jumper/header in the picture below is connected to the cert battery +/-. However (quite unusual for such a header), it is SMD mounted and you can't see any traces, so, it's connected on the inner layers. On a dead ICP, I would try to measure it against the battery contacts. Yes, it is possible that there would be some passive components between like a resistor or diode, but one should be able to find that out. This header could be an official means to feed a backup supply.

Bildschirmfoto 2021-02-05 um 16.10.13.png

[Leo Enticknap]

These NEC instructions are garbled. I think they are getting confused between the functions of the clock battery and the cert battery. I've encountered an NEC with a totally flat ICP clock battery, and the projector continued to work OK. If I remember correctly there was an error, and that persisted until the battery was replaced and the clock had actually been reset, but it didn't stop the ICP from processing DCP data for playback. The bad clock battery certainly did not cause a card cage tamper.

On a Series 2 Barco, in contrast it might. Again, this was a long time ago and I can't remember all the details, but the bottom line was that the flat battery caused a mismatch between the ICP's clock and the media block's that was so great as to prevent DCP playback (alternative content through DVI to the CCB still played).

I have had one NEC with a dead ICP cert battery (in the light of these discussions, I wish I'd kept it to offer to ship to anyone who wants to experiment with surgery on it: I will keep any others I replace from now on). It gave an ICP communication error, and the media block (Dolby cat862 to Enigma, the battery in which was still good) gave "Error connecting Show Store to Projector."

[Marcel Birgelen]

I agree, those NEC instructions are really badly informed. It figures that even their own engineers seemingly don't exactly know how this whole battery thing on the ICP works.

We're over here in Europe, so shipping us dead ICPs is probably going to be pretty costly. But it would be interesting to trace some of those leads, without the fear of blowing up a pretty expensive piece of hardware. I'm especially interested in what those headers do. I guess some of them are part of some kind of JTAG-style interface, that's also being used during the initial certificate loading. If they provide both the ground and +3.xV on one of those pins, one could easily design a small daughter board that hooks up to those pins, even no soldering required...

[Steve Guttag]

I happen to have a dead ICP (wouldn't ever recover...years ago). As most would guess, the "+" of the battery does go to R587 (see Carsten's picture further up).

The "-" terminal does seem to be on a digital ground of sorts so all of those connectors on its face (USB, for example) are tied to the "-" terminal of the battery. My guess is that one could tack a lead to one of those, scrape off some solder mask for that largeish trace going to R587 from the battery and tack a "+" lead to that and put in a temporary battery for the duration of the permanent change without too much risk (presuming an isolated tip soldering iron)...once the replacement is in place, remove the temporary.

I would think that removing the battery may be challenging on the negative since it is likely going to a ground plane fill (plus the battery itself is a heat sink). Hopefully, the person that laid out the board did a good job on the thermal reliefs. If not, cut it out and tack on but I'd prefer real metal-to-metal contact rather than just solder.

So, do you feel lucky, punk? Well, do ya?

[Marcel Birgelen]

Yes, the + trace seems to go to the top of R587, where you could tack on a temporary + lead for a temporary battery. From there it seems to go to the tiny capacitor C409 and then probably disappears to an inner layer of the PCB. If the negative side of the battery is connected to a common ground that also serves the ports on the front, then it will be easy to find some exposed metal surface

I expect the ground plate to be a similar affair as the + side, also tacked on with a bunch of soldering points. So, in order to reach them you need to first remove the "+" clamp, then you probably need to bend the battery about 90 degrees ever so carefully to expose the ground plate. The problem is, if you bend those things too far and break them off, then you're in a different world of hurt. Although there are similar style battery mounts available online, so maybe it's best to be prepared to replace the entire holding assembly with it.

[Carsten Kurz]

I wouldn't try soldering on that tiny resistor. If it comes off during that process only for a fraction of a second - boom. There is more than enough space at the + battery terminal. Because of the through-hole contact and the large pads, it is very robust there. Forget about desoldering the battery - just clip off it's leads and use them to attach a wired replacement battery.

[Marcel Birgelen]

I agree that you could potentially break contact for a short period, while soldering something onto the SMD resistor. The tiny capacitor next to it doesn't look like it's going to be much of a buffer in such an incident. You could try to scrape some of the masking off of the trace. Still, is this the only +3V trace going out from the battery before it hits the resistor?

I don't think that attaching the temporary battery to the existing battery terminal itself is the most practical thing, especially when you want to clip it off though.

Isn't there a way to feed the ICP via the backplane while desoldering and replacing the battery? What if you just feed the thing with stand-by power over the relevant pins?

[Carsten Kurz]

The trouble is, we don't know much about this. For some of these clock or cert keeping circuits, manufacturers document that the circuit is backed up by system supply during operation. For the ICP, it is simply not documented. And it is a very high risk to measure around it on a working ICP. Even if you'd manage to take some measurements on an ICP that is powered up (not easy), you wouldn't know where to detect the supply backup on the board, because it is very probably isolated from the battery with a least a diode. Not an option without a circuit diagram.

Why would it be a problem to use the battery contacts after clipping the battery off? You can leave them long enough to have ample space for soldering. Even if you wiggle off the battery to access the negative lead, it leaves much more for soldering than anything else there. The through hole connection with relatively large pads on both top and bottom is the best you can get for attaching a patch wire. Then you could buy or make one of these:

image_787.png

Maybe Steve can take the 'clip off +' and 'wiggle off -' approach on his dead ICP. Not much to lose there.

Steve, did you measure the cell voltage on this ICP? Is there a manufacturing date visible on it?

[Mark Gulbrandsen]

Marcel, I don't think those batteries take a charge from incoming power. Not sure those types of batteries can even be charged without leaking and doing bad things. If there is a super capacitor as NEC claims, then it gets charged separately.

Also, Leo, the ICP clock on an NEC should reset itself automatically during first boot. All the ICP's I have changed batteries on did that.

[Steve Guttag]

I would tack my temp battery to the trace heading towards R587 (in that 45-degree section). Just a little scraping on the soldermask should expose copper for the job.

As for desoldering versus tacking the new battery to existing, clipped, leads, I would want to see how difficult desoldering that negative terminal would be. The positive one will be relatively easy as there is very little copper there to soak the heat (the battery itself would be the main sink). When I get some time, I'll see how well it was thermally relieved. For a copper fill, I always used a set of four rectangles to make the relief and it worked reasonably well. For outer layers, I used 4-spoke patterns. Again, worked well.

The thing to do is to re-wet the solder connections on the battery and see how easy it is to remove the solder. Either it comes out easy enough or it doesn't. For a live unit, you really don't want to mess with it too much and you need to use something insulated (e.g. an orange stick) because it is too easy to short + and - on a coin battery. As an amusing story aside...this past week, I was on site I was, among other things changing the batteries in a GDC SX-2001 server mediablock and while I was in there, I changed the BIOS battery on the mother board (sits right below the mediablock). The battery hold on the motherboard is a good one in that you just have to release the metal tab and the battery pops up...no fussing or prying or bending anything. Well, the spring on that holder was a good one and upon releasing the catch, the battery pops up...flips and lands dead centered back in the holder! So much for the clock and those settings! First time I've ever had that happen and I had just done the same procedure the week prior on a different server...the swap was fast enough that no settings were lost.

Anyway...any such accident on an ICP certificate battery is a $5000+ accident (Barco being a particularly expensive version).

I do agree that it should be presented to customers as a "no-fault" service. You should have a spare ICP with you and if it goes poorly, the customer gets to spend for the ICP a year or two early. If it goes well, they saved $1000s per projector. If we get some successes, then we can more accurately predict the success rate.

There is probably some value to doing battery Voltage measurements on this though one would need to know the threshold at which a battery is in a danger zone as well as what the curve is...how fast does it go from say 3.0 to 2.97 to 2.92....that sort of thing. The cliff may be VERY steep. I know, with CAT745s...they seem to go from fine and dandy to dead with no warning.

[Marcel Birgelen]

No, those batteries aren't rechargeable... well, you can recharge them to a certain degree, but you indeed risk premature corrosion and leaking. But, following simple logic, eventually, this battery and the standby/operational power will feed into the same bus, powering the circuitry that keeps the certificate alive. There will probably be a diode somewhere, blocking the standby current flowing into the battery. Maybe even a somewhat smarter voltage regulator somewhere. So, as long as there is external power, the drain on the battery should be minimal.

Like Carsten indicated, the whole thing is obviously a black box, so we can't be sure that it works this way. But you'd expect that if you power the "standby bus" of the ICP, you should be able to replace the battery on the board, without the security storage losing its power.

Regarding clipping the old battery off and "hotglueing" the new one in place... It feels like an ugly hack... a bit like how your average cheap Chinese gadget from Wish.com or AliExpress is being assembled... Part of me doesn't mind it, the other part does. Also, I guess a procedure that could be performed by a larger number of people in the field would be preferable. Not everybody is comfortable clipping stuff off of such an expensive board, especially if it isn't your own.

[Marcel Birgelen]

As for desoldering versus tacking the new battery to existing, clipped, leads, I would want to see how difficult desoldering that negative terminal would be. The positive one will be relatively easy as there is very little copper there to soak the heat (the battery itself would be the main sink). When I get some time, I'll see how well it was thermally relieved. For a copper fill, I always used a set of four rectangles to make the relief and it worked reasonably well. For outer layers, I used 4-spoke patterns. Again, worked well.

image_14035.jpg


Like I described above, I don't think there will be any copper fill below the battery and the battery will not be mounted to the PCB directly. The bottom of the battery will probably also spot a metal clip, tacked onto it. That's why you probably need to bend it to access it.

What worries me though, you can buy those batteries with those clips on there. I hope they're soldered on and not spot-welded...