TI-84 Plus CEs made within the last few years have a tenancy to get stuck in a boot loop when their battery dies. I didn't understand why this was so I did some investigating.

I thought the charging circuit was completely independent of the OS but this doesn't seem to be the case. When the OS is loaded the charge limit is 500ma. However, whenever the OS isn't loaded (i.e. when validating OS, holding RAM reset button, or on the OS reinstall screen) charging is throttled to less than 100ma.

I measured how much current is pulled (when the OS isn't loaded) on few different TI-84 Plus CEs:

• L-0315: 97ma (This may not be accurate, read below)
  
• L-1215A0: 88ma
  
• L-0122W: 87ma

Left: L-0315 | Right: L-0122W (click for full size)
Not pictured: L-1215A0

For such a low powered device, I figured ~90ma would be enough to let the calculator do whatever it wanted and still charge the battery. However, I observed that just as the OS starts to load (after validating OS) the calculator stops charging completely for 1-2 seconds. This means it's likely the battery is getting charged, but the calculator booting just takes too much power and drains what little charge was just put in.

---

My theory is that if the screen is unplugged, the calculator will put more power into the battery than it takes to boot. To see if this theory has any hope of working I needed to test the power draw from the battery with and without the screen.

Testing L-0315
Date: March 2015
Revision: pre-A
OS: 5.3.0

This calculator is heavily modified. Based off the current draw results above, I don't trust any tests to be accurate.

---

Testing L-1215A0
Date: December 2015
Revision: A0
OS: 5.3.1

With the screen: Spikes up to 68ma then settled at: 52ma.
Without the screen: Spikes up to 39ma then settled at 24ma.



The screen is using around 68 - 39 = 29ma.
That's 29 / 68 = 42% of the battery's current draw when booting and over 29 / 52 = 55% when idling.

So of the 88ma of USB current, the battery is getting:
With the screen: 88 - 68 = 20ma or 20 / 88 = 22% of the USB current.
Without the screen: 88 - 39 = 49ma or 49 / 88 = 55% of the USB current.

---

Testing L-0122W
Date: January 2022
Revision: W
OS: 5.6.0

With screen: Spikes up to 53.8ma then settled at 51.0ma
Without screen: Spikes up to 20.8ma then settled at 17.7ma



I have to commend TI's engineers for lowering the overall power draw overall significantly! However, clearly it's still not enough to charge a dead battery.

The screen is using 53.8 - 20.8 = 33ma.
That's 33 / 53.8 = 62% of the battery's current draw when booting and 33 / 51 = 64% when idling!

So of the 87ma of USB current, the battery is getting:
With the screen: 87 - 53.8 = 33.2ma or 33.2 / 87 = 38% of the USB current.
Without the screen: 87 - 20.8 = 66.2ma or 66.2 / 87 = 76% of the USB current.

---

I verified the calculators pull the same amount of current from the USB port no matter whether the screen is present or not.

Therefor, I conclude these tests show the battery charge rate when booting can be doubled by removing the screen. Since, without the screen, over half the power is being put into the battery instead of booting the calculator this might allow some calculators to revive themselves. I haven't intentionally caused the boot loop problem on my calculators so I don't know for certain that removing the screen will solve the problem, but I have high hopes for this theory.

What happens if you give power to the two docking pads on the bottom of the calculator rather than through the USB cable- is there any difference?

The USB1.1/2 standards require the devices only initially pull 100 mA and cannot more power until authorized by the host. Yes, this rule is widely ignored, but TI appears to be adhering to it, with the boot code simply omitting the power negotiation.

If a lithium-ion cell doesn't have an internal protection circuit that's locked itself out, you can manually trickle-charge it with some resistors and a cut-up USB cable. I'd suggest starting with 1 kΩ and monitoring the current going into the battery, staying around 10-20 mA until the voltage reaches 2 V, and under 100 mA until 3 V, changing the resistor out from time to time. You can, of course, use lower trickle-charge currents, but it'll take forever to get the battery voltage back up.

Do not attempt to charge a cell showing more than -0.05 V of reverse voltage (i.e. (+) terminal reads negative relative to (-) terminal), as it may literally catch fire. Do not attempt to charge multicell packs because they can generate dangerous reverse voltages when overdischarged.

I've posted a request for volunteers on Reddit: https://www.reddit.com/r/calculators/comments/1etziui/fixing_ti84_plus_ces_that_wont_charge_volunteers/
Hopefully I'll get some data on whether this theory actually works.

Something I'd like to test is whether the OS validation screen takes more power than getting to the OS recovery screen. Perhaps if the user holds down 2nd+del while charging the OS validation will be skipped?

Can the lithium ion battery pack be swapped out for a set of NIMH AAA batteries in a plastic battery holder?

While not as much density as a lithium ion pack, they seem more resilient and swappable if a problem occurs.

On all my triple A powered calcs, I simply use rechargeable NIMH. They don't leak like alkaline and are easy to charge.

I have a variable power supply, so I'll try and get back to you with the results. (Though my calculator's hardware revision is AA, not sure if this will affect the test.)

EDIT: The OS validation screen uses ~10 mA more (total ~50 mA) than normal (~40 mA.) It's a negligible difference, but it could still be the smoking gun.

EDIT 2: The OS recovery screen is also drawing ~50mA. The calculator's code is L-0123AA if you were wondering.

On the back of TI calcs there is what I think is a born on date: 0700 K
So I believe that calculator was build in July 2000. Is the K the revision? I now have numerous thrift store Z80 TI calculators and they seem to follow this pattern.

cbusch let's please keep this thread focused. You're welcome to make a new topic about such a project though! Just make sure you don't try to use the calculator's charger to charge AAA batteries since their chemistries are very different.

HydrideGS tested the OS reinstall screen for me and it also draws 50mA. That's enough info to tell me the OS reinstall screen wouldn't be a viable way to make the calculator charge more.

The results from people on Reddit who tried my theory aren't optimistic. So far 1/4 people had success.

I tried intentionally discharging a battery but despite my best efforts I couldn't get the battery below 3v before it shut itself down and wouldn't discharge any further. Charging the battery for about 10 seconds would revive it. Maybe TI's batteries have been sitting around so long they've naturally discharged even further and can't be revived.

Yes. The letter to the left of the datecode is the factory letter.

This probably isn't going to help, but the calculator doesn't start up unless at least 3.6V are present. In fact, if the voltage is lower than this threshold, the calculator will refuse to turn on automatically even when sufficient power is applied. Pressing "on" will turn it on again, though.

I did end up testing this and the calculator seems to charge slightly slower with the charging dock. The calculator also seems to prioritize charging from the dock and will stop charging from the USB port when dock power is applied.

I did discover how to replicate the boot loop. If you connect the battery negative and temperature lead but leave the positive lead disconnected while charging, the calculator enters the OS validation loop.

USB Power:
While it's looping, the battery isn't being charged at all (0.1ma) and power also isn't being pulled from it. Power seems to be coming directly from the USB port.

Charging Dock Power:
When you disconnect the battery positive lead while charging from the dock, the calculator doesn't even try to power itself with just dock power and enters a much faster boot loop. The battery also only gets (.1ma) of charge during this process.

On a different note, I reached out to one person on Reddit who tried my battery recovery method without success. It turns out their battery was inflated so it was never going to come back to life. I'll be reaching out to others to see if their batteries are also inflated. (I updated the guide to warn redditors never to charge inflated batteries).

Temperature lead? Is that the middle pin? When I was working with dumbphones, the group I was in always called it "BSI"... guess now I know what it is.


Yikes, inflated batteries? Already? These newer calculators are only a few years old at this point, they shouldn't have inflated yet.

As for me, I've experienced similar bootloops during the time that I worked with dumbphones. Unfortunately, I don't have much knowledge that could potentially help, but here's something I do have -- bootlooping may not necessarily be caused by the battery losing voltage, it could also be caused by the OS crashing. I'm unsure how a TI-84 Plus CE would react to an OS crash, but on the dumbphones I worked on, they immediately rebooted. Is this a good lead?

So this thought came to me recently, how about the wireless charging mod that you made? (TheLastMillennial)

I know this won't be easily accessible with calculators that have the boot loop problem as some people wouldn't want to give up the ability for charging dock power and some wouldn't have the skills and materials required for the project, but would wireless charging charge the battery? You probably have already tested it but just wondering.

From watching the video, I think that the power from the wireless charger is sent to the dock ports (correct me if I'm wrong), so it might not make a difference at all, and give the same results as charging from a dock.

The wireless charging pad is connected to the charging dock contacts so you're right it wouldn't make a difference.

I asked multiple people on reddit and discovered about half of the dead batteries were inflated in some capacity. Unfortunately this mostly likely means all the dead batteries are toast and there's no safe way to revive them. Even someone who tried to charge their battery with a bench power supply (not recommended) couldn't revive their battery.

A new topic brought to my attention there's two battery hardware revisions: A and B. I believe all my batteries are revision A so there may be a defect in revision B batteries?

There may not be a way to fix existing dead batteries, but maybe people can intentionally try to purchase revision A batteries if we can prove they won't cause a future boot loop. This would be especially useful for those without a warranty!

As much as I'd love to test that theory out, I don't have a quick and easy way to drain the battery. I have a sacrificial revision A battery to test this out with.

Also worth mentioning is this: