CPAP Repair - PCB Coin Battery Power Loss?

[svor1988]

So when the device runs, it stops working after maybe 10-30 minutes of use, beeps, flashes a red LED, and shows the error message "E76" on the LCD. Unplugging it & plugging it back in will get it to work for maybe 1-3 more minutes, then it fails again.

Attached is the error code from the service manual, as well as the relevant circuit shown in the service manual. I think the device is 10+ years old (I bought it off ebay), it is a discontinued model. Interesting that the service manual for this error code says to replace the entire PCB for the "failed" battery, when the replacement PCB retails over $1000!

Also attached are some relevant excerpts from the AT91M55800A datasheet.

The coin cell battery measures 2.83V, and it should work for the AT91M55800A all the way down to 1.8V.

I think either C10, C11, or R32 are not working? Maybe the circuit is designed to fail this way? I am no expert, I can read these datasheets but I am very novice with circuitry. Can you help direct me at what to test/replace?

R32 measures 0.99MOhm.
C10 no resistance reading (over 2MOhm).
C11 measures 16.45kOhm. Capacitors shouldn't have resistance right?
When I measure voltage over either C10 or C11 it starts around 2.1V-1.8V and rapidly drops to 1.4V then slowly drops to 1.2V.

The coin cell battery has two electric welds holding it in place, I am reluctant to replace it because it should work.

NRSTBU activates low, and I think it is what drives that error message and stops the device from working. The use for the battery in this device is so that when the grid power goes off, it will alarm (wake the person up) in addition to stop blowing. NRSTBU appears to be used to detect when the battery is low.

I think either C10 or C11 is discharging faster than the current through R32 can charge them, thus NRSTBU lowers and activates?

Some datasheet links:
Atmel Arm Microcontroller AT91M55800A
BAT54

The service manual for this device (and dozens more CPAP machines) is on apneaboard.com, behind a registration wall and a maze of links. I believe I've attached everything necessary though. I think the standard warranty for CPAPs is 2 years, and many people buy them second hand out-of-warranty (including me, I can't afford $4000+ retail new machine & service).

[gabiz_ro]

Re: CPAP Repair - PCB Coin Battery Power Loss?

C10 no resistance reading (over 2MOhm).
C11 measures 16.45kOhm. Capacitors shouldn't have resistance right?

So you measured after remove from circuit.

Replace C11 or better both of them.

If still you get this readings in circuit then there is brocken track between C10 and C11 since you must get same value on both.

For a quick test you may go on without C11 and see if works.

[CapLeaker]

Re: CPAP Repair - PCB Coin Battery Power Loss?

the coin cell is dead at 2.8V as it cant deliver much current anymore at that voltage. That needs to be replaced. Capacitors are best checked with a propper ESR meter and out of circuit. it seems like a 25v 10uf cap should last forever at 3.3V, but caps can look good and still be bad. I'd be either properly testing or replacing C10 and replace that battery. What is the DC voltage on either side of R32? Retest that thing again and see what happens. where does battery_bu come from (fed to cr5)? To me, when the device is on, it shouldn't draw anything from that battery. There should be something feeding something like 3.3V.

[jetadm123]

Re: CPAP Repair - PCB Coin Battery Power Loss?

Try heating up the suspect capacitors with a hair dryer to reduce the time to get to the "E76" display. This should help with isolating the area causing the problem.

[redwire]

Re: CPAP Repair - PCB Coin Battery Power Loss?

Error code E76 "Failure in loss of power battery" could be for a gel-cell battery backup UPS some apnea machines can use, not the coin cell?
The MCU monitors the +12V and V_BULK rails, so this error code might be more about that.

Start with the basics measuring the power supply voltages and see if they are sagging due to problems.
I would ensure BATTERY_BU and what it powers, what its voltage is- but once the main power is up, this (coil cell) is not used, so it shouldn't be the problem.

NRSTBU (pin 112) is the hardware RESET input to the MCU for its battery power section called VDDBU. This would reset the real-time clock and wake-up logic (but not used here). If it sags down to 1.2V, that's OK, this is just the RTC reset circuit. To manually reset the RTC MCU section, shorting C10 would do it.
I'm not sure if a corrupted RTC could be part of this, kinda doubt it.

[svor1988]

Re: CPAP Repair - PCB Coin Battery Power Loss?

So you measured after remove from circuit.

Replace C11 or better both of them.

If still you get this readings in circuit then there is brocken track between C10 and C11 since you must get same value on both.

For a quick test you may go on without C11 and see if works.

I didn't remove them from the circuit.
I now read 16.35kOhms on both C11 and C10, they now give the same resistance. I had to have red multimeter lead on ground side of the capacitors. When the black lead is on the ground side of the capacitor then there is no resistance reading for either of them.

R35 gives 1.96kOhms.
CR5 gives 6.16kOhms red lead on battery side, black lead on that side has no resistance reading.

I'll try testing without C11 after testing a new battery, thank you

the coin cell is dead at 2.8V as it cant deliver much current anymore at that voltage. That needs to be replaced.

You made me look at what voltage a coin cell is considered dead, and it looks like measuring a dead one "open circuit" still gives 3V. I suppose testing the voltage between the battery and some accumulation of R35, CR5, and R32 (as a load) would be more meaningful. I guess it probably is dead then I need to go buy a new one.

Capacitors are best checked with a propper ESR meter and out of circuit. it seems like a 25v 10uf cap should last forever at 3.3V, but caps can look good and still be bad. I'd be either properly testing or replacing C10 and replace that battery.

I guess it would be cheaper for me to just buy new capacitors & replace them than to buy an ESR meter.

What is the DC voltage on either side of R32? Retest that thing again and see what happens. where does battery_bu come from (fed to cr5)? To me, when the device is on, it shouldn't draw anything from that battery. There should be something feeding something like 3.3V.

There are only two battery_bu instances, both in attachments, they connect to two different pins VDDBU & NRSTBU. I agree it shouldn't draw anything from the battery when on. I'm thinking CR4 is more meant as a way for the battery to power the 3.3V rail to alarm when the power is off? I don't know. I have not looked at CR4 yet on the board.

I guess I should attach the whole schematic. But both the place I got the service manual from (for free) and this cheesy clause in there that makes me hesitant:

8.0 PROPRIETARY STATEMENT Schematics are supplied in direct support of the sale and purchase of this product.
The Schematics are proprietary and confidential. Do not copy the schematics or disclose them to third parties beyond the purpose for which they are intended.
The schematics are intended to satisfy administrative requirements only. They are not intended to be used for component level testing and repair. Any changes of components could effect the reliability of the device, prohibit lot tracking of electronic components, and void warranties. Repairs and testing are supported only at the complete board level.

I think administrative requirements means laws, they had to follow the laws of each country they sell the device in. Why would the country require published schematics for anything other than the user being able to repair what they bought when it breaks? From what I see it looks like they want to force the selling of the entire circuit board from their country's factory whenever something breaks.

[svor1988]

Re: CPAP Repair - PCB Coin Battery Power Loss?

Try heating up the suspect capacitors with a hair dryer to reduce the time to get to the "E76" display. This should help with isolating the area causing the problem.

Yes this makes sense why the machine behaves as it does, it may be heat leading to the failure.

Error code E76 "Failure in loss of power battery" could be for a gel-cell battery backup UPS some apnea machines can use, not the coin cell?

I thought might have something to do with an external battery, but the error message says to replace the PCA in the service manual. Two power sockets on this device, and they are the same connector, see attached images. Since they're the same I tried connecting the AC/DC adapter into both and still get the E76 error message with both sockets.

The MCU monitors the +12V and V_BULK rails, so this error code might be more about that.
Start with the basics measuring the power supply voltages and see if they are sagging due to problems.

I think the voltages are fine because the machine has error messages about the voltages (attached), but only E76 appears.
I have not tried testing the voltages yet because I need to take the machine apart more to get space for the probes.

I would ensure BATTERY_BU and what it powers, what its voltage is- but once the main power is up, this (coil cell) is not used, so it shouldn't be the problem.

That zener diode and the 3.3V rail puzzles me.

NRSTBU (pin 112) is the hardware RESET input to the MCU for its battery power section called VDDBU. This would reset the real-time clock and wake-up logic (but not used here). If it sags down to 1.2V, that's OK, this is just the RTC reset circuit. To manually reset the RTC MCU section, shorting C10 would do it.
I'm not sure if a corrupted RTC could be part of this, kinda doubt it.

Good idea, I'll try shorting C10 and see if the error pops up!

Thank you so much everyone!

[Curious.George]

Re: CPAP Repair - PCB Coin Battery Power Loss?

So when the device runs, it stops working after maybe 10-30 minutes of use, beeps, flashes a red LED, and shows the error message "E76" on the LCD. Unplugging it & plugging it back in will get it to work for maybe 1-3 more minutes, then it fails again.

The coin cell battery measures 2.83V, and it should work for the AT91M55800A all the way down to 1.8V.

You're looking in the wrong place. Does the machine "remember" any nonvolatile settings (or time-of-day clock?) when this event happens? (If so, then the coin cell is doing its job).

You can verify this by shorting the NRSTBU input to ground through a 1K resistor. CR4 & CR5 form a "switch" that selects the greater of the ~3.3V and "coin cell" voltages to provide to the BATTERY_BU signal (which goes elsewhere!). The current draw on the NRSTBU pin for the microcontroller is miniscule -- hence sourcing it through the ~1M R32. Even with the NRSTBU pin grounded, you'll only present a 2 microamp load on the BATTERY_BU signal (which will be sourced from the 3.3V power supply when the device is running)

How the software interprets this is anyone's guess; if the coin cell is not an FRU, then it may consider this a "can't happen" scenario (happens only when the battery is first installed IN THE FACTORY).

NRSTBU activates low, and I think it is what drives that error message and stops the device from working. The use for the battery in this device is so that when the grid power goes off, it will alarm (wake the person up) in addition to stop blowing. NRSTBU appears to be used to detect when the battery is low.

NRSTBU is used as a power-on-reset signal for the portion of the electronics in the microcontroller that run off the VDDBU power. It should only "go low" when you change the battery while the main power is off (cuz if main power is on, then the 3.3V supply will ensure NRSTBU remains "high" due to the action of the diode switch)

[Curious.George]

Re: CPAP Repair - PCB Coin Battery Power Loss?

I thought might have something to do with an external battery, but the error message says to replace the PCA in the service manual. Two power sockets on this device, and they are the same connector, see attached images. Since they're the same I tried connecting the AC/DC adapter into both and still get the E76 error message with both sockets.

Prior to the descriptions of the error codes, does it FIRST suggest "removing/replacing external backup battery" from the circuit (to see if the error goes away??).

[Did you, perhaps, acquire the device without a battery? Or, is there a configuration option to enable/disable battery operation? Note that the external battery will undoubtedly be large (compared to the dinky coin cell) as it will have to power the pump in the case of primary power failing while you're asleep]

I think the voltages are fine because the machine has error messages about the voltages (attached), but only E76 appears.
I have not tried testing the voltages yet because I need to take the machine apart more to get space for the probes.

The software most probably stops when it encounters the FIRST error that it detects. And, it detects errors by examining things in a particular sequence (known to the software developer). So, if it checks THIS condition before checking the others, then a failure here says nothing about the others! (it hasn't looked at them, yet).

One error may make the reporting of the next error problematic

[Ask yourself how the device could possibly report 30 different errors, simultaneously, given its limited display capabilities -- every error code could potentially need to be reported at the same time!]

That zener diode and the 3.3V rail puzzles me.

It's part of a diode switch with the other diode in series with the coin cell. In effect, it says: "feed the higher of the voltages on our cathodes to the BATTERY_BU signal".

[Curious.George]

Re: CPAP Repair - PCB Coin Battery Power Loss?

I think administrative requirements means laws, they had to follow the laws of each country they sell the device in. Why would the country require published schematics for anything other than the user being able to repair what they bought when it breaks? From what I see it looks like they want to force the selling of the entire circuit board from their country's factory whenever something breaks.

Few manufacturers expect the folks who bought the device to actually repair it. OTOH, certain types of users (think: hospitals) may have their own QUALIFIED service departments. Should they be prevented from repairing a piece of equipment just because there are no schematics available?

From the end user's perspective, the manufacturer wants to ensure it isn't held liable for the actions of the end user -- potentially making the device "unsafe" because the user didn't understand the importance of a particular design feature and "broke" it by an inadequate repair. Someone suffers a loss (life?) and wants to hold the manufacturer accountable:

"Hey, the board had been tampered with by an unlicensed repair facility (or person). It's not our fault the user died!"

[Hospitals/clinics tend to be aware of the consequences of the repairs that THEY opt to do -- and the liabilities they incur as a result! And, have their own legal teams to defend against their patients claiming that a defective piece of equipment caused them to suffer a loss. And, those same legal teams can pass the problem along to the manufacturer -- via a lawsuit -- if the basic design is flawed and has resulted in the hospital incurring a loss (in litigation)]

[CapLeaker]

Re: CPAP Repair - PCB Coin Battery Power Loss?

With out seeing the whole schematic, I'd say there is something wrong with that Battery_BU input next to that CR4 3.3V Zener. So you can remove a leg of R32 and feed somehow 3.3V to that IC and the unit should work. Question still remains where that Battery_BU is coming from (next to CR4). If that is some external backup battery pack, if the power fails, maybe it is bad, or missing all together.

[Curious.George]

Re: CPAP Repair - PCB Coin Battery Power Loss?

With out seeing the whole schematic, I'd say there is something wrong with that Battery_BU input next to that CR4 3.3V Zener. So you can remove a leg of R32 and feed somehow 3.3V to that IC and the unit should work. Question still remains where that Battery_BU is coming from (next to CR4). If that is some external backup battery pack, if the power fails, maybe it is bad, or missing all together.

BATTERY_BU isn't an "input", there (except to the NRSTBU pin on the MCU). Rather, it's an off-page (?) connection to the battery backup POWER pin on the MCU (the NRSTBU pin is just a PoR input for the circuitry that is powered by that BATTERY_BU power source)

Note the NRSTBU signal has an added ~1M series resistance so it must be a really high input impedance. By contrast, the BATTERY_BU directly (low impedance) feeds the VDDBU (power) pin.

[CapLeaker]

Re: CPAP Repair - PCB Coin Battery Power Loss?

R32 is a current limiting resistor to me, not high impedance. After all it is just a reset pin.

[Curious.George]

Re: CPAP Repair - PCB Coin Battery Power Loss?

R32 is a current limiting resistor to me, not high impedance. After all it is just a reset pin.

The impedance of the input pin must be >> the source impedance, else the signal AT the pin will be affected.

Imagine Zin is 1K ohms. What will you see at the input pin if you have ~3V hiding behind 1M ohm?

Now, imagine Zin is nearly infinite (>> 1M). No current flows through the 1M (because Zin is so high) so the input pin sees the ~3V "signal". I.e., you can insert a high series resistance (effectively increasing the output impedance of the "source") because the input impedance is even higher!

By contrast, the VDDBU pin is free to draw whatever current the MCU requires directly from the BATTERY_BU signal.