So the MAX chip looks to be doing what it says on the tin.

I guess the next step is that I should solder on the timer, inverter, and those components corresponding circuitry?

I guess so too. If you add each of those in turn and repeat the shut-down test at each turn, hopefully that will show up which bit is causing the problem.

 

So the MAX chip looks to be doing what it says on the tin.

I guess so too. If you add each of those in turn and repeat the shut-down test at each turn, hopefully that will show up which bit is causing the problem.

Ok, I added the inverter, decoupling capacitor, timer IC, and the timer programming resistor (did 10 seconds). Everything is continuing to work perfectly. I was manually cycling the "done" pin on the timer to trigger the timer to shut down the MAX1795.

I connected the original board up to the oscilloscope, and was monitoring the MAX1795 shutdown pin and its doing what its supposed to (e.g. I can see the timer/inverter bringing the shutdown pin up and down like it should). But for whatever reason current is still flowing somehow.

 

So your second build is working ok but the original is not? That suggests the original MAX chip is defective. Static damage perhaps?

 

So your second build is working ok but the original is not? That suggests the original MAX chip is defective. Static damage perhaps?

I originally built a couple boards though and each one behaves the same.

 

Weird indeed. Clutching at straws here, but if manually cycling the 'done' pin gives ok results, perhaps when that pin is automatically controlled there is spurious oscillation, causing the MAX chip to cycle rapidly between on and off (more off than on)?

 

Well, I just figured it out! Thanks again Alec_t for the help!

It turned out to be IC4 (Analog-to-Digital Converter). Sure enough when power is dropped to the ADC, current is flowing through the channel pin (this information is not in the data sheet). It was the absolute last thing that I could imagine, so I de-soldered the ADC and it now works perfectly (aside from not having an ADC now). This also makes sense now why I noticed that the lower the battery level, the less amount of current would leak (when I was testing a single 1.75 battery).

With that in mind, what do you think the best solution would be for this ADC situation? It seems like I am going to need to add a transistor/mosfet into the equation?

Maybe have it in a default off state but controlled by the microcontroller, so when a reading needs to be made the microcontroller activates the transistor/mosfet, takes a reading, then the whole thing shuts back down?

 

Glad you got to the bottom of it.

Maybe have it in a default off state but controlled by the microcontroller, so when a reading needs to be made the microcontroller activates the transistor/mosfet, takes a reading, then the whole thing shuts back down?

That looks like it would work. Finding a FET with a low enough turn-on threshold may be a challenge, given the very low supply voltage.

 

Glad you got to the bottom of it.

That looks like it would work. Finding a FET with a low enough turn-on threshold may be a challenge, given the very low supply voltage.

Do you think I could place a diode at the VDD/REF of the ADC chip?

 

Can you afford to lose ~0.3V (if a Schottky diode is used) of an already low supply voltage?
The diode voltage drop will be temperature dependent, so Vref won't be a very stable reference!

 

Can you afford to lose ~0.3V (if a Schottky diode is used) of an already low supply voltage?
The diode voltage drop will be temperature dependent, so Vref won't be a very stable reference!

Yeah thats a good question. For a period of time, the voltage input on the ADC channel will be 3.6V (fresh batteries), with a 3.0 voltage reference with the schottky diode, I don't know if there would be any longterm problem with that. It is just a battery indicator so it doesn't necessarily have to be that accurate, but It would sill be nice to have a decent design at least thats not overly complicated.

EDIT:
Completely forgot that I have the light sensor connected to this ADC as well....hmmm... now I don't know what the best idea is.

 

Since the light sensor (T1) is supplied by the 3V3 from the MAX chip and not directly by the battery I don't see how it can draw current if the MAX chip is in shut-down?

 

Since the light sensor (T1) is supplied by the 3V3 from the MAX chip and not directly by the battery I don't see how it can draw current if the MAX chip is in shut-down?

Yeah you are right. I was thinking more along the lines of reading accuracy if I incorporate the diode into the VDD of the ADC. Maybe I will be good enough.

 

If your ADC has a band-gap internal reference you should be ok.

 

If your ADC has a band-gap internal reference you should be ok.

I found a diagram of the input channel of the ADC:


So this shows how the leakage current is traveling up through the top protection diode and onto the 3.3V rail. If I put this diode on the VDD pointing towards the ADC, It looks like it will successfully solve the problem.... but will this basically nullify the protection diode within the ADC?

 

If you have the usual decoupling cap bridging Vdd/Gnd, that would offer some protection for transients (but not excessive DC voltage).