Author Topic: Which diode to use?  (Read 1430 times)

0 Members and 1 Guest are viewing this topic.

Offline Marco

  • Super Contributor
  • ***
  • Posts: 3806
  • Country: nl
Re: Which diode to use?
« Reply #25 on: April 19, 2018, 04:12:44 am »
Nothing. The 3V3 is only present when the device is switched on.

Why feed it 3.3V from what I assume is a DC-DC converter at all? SDA/SCL are allowed to be higher than Vdd and current consumption at Fscl=0 is the same as on battery backup with inactive I2C (note 3 page 60).

So just connecting the Li-Po to Vdd and the battery to Vbat should allow it to function fine.
« Last Edit: April 19, 2018, 04:15:57 am by Marco »
 

Offline Karel

  • Super Contributor
  • ***
  • Posts: 1264
  • Country: 00
Re: Which diode to use?
« Reply #26 on: April 19, 2018, 04:34:00 am »
Nothing. The 3V3 is only present when the device is switched on.

Why feed it 3.3V from what I assume is a DC-DC converter at all? SDA/SCL are allowed to be higher than Vdd and current consumption at Fscl=0 is the same as on battery backup with inactive I2C (note 3 page 60).

So just connecting the Li-Po to Vdd and the battery to Vbat should allow it to function fine.

What I read at page 60 is that the I2C input thesholds (hi/lo) are relative to Vdd.
So when Vdd is 4.2V (max Li-Po voltage), the high-level input threshold is 2.94V minimum.
Not much room for tolerances...

The difference between theory and practice is less in theory than
the difference between theory and practice in practice.
Expensive tools cannot compensate for lack of experience.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 3806
  • Country: nl
Re: Which diode to use?
« Reply #27 on: April 19, 2018, 05:00:30 am »
Maybe pull the I2C bus up to the Li-Po voltage if the other devices can handle that?

Or put an ultra low quiescent current LDO on the Li-Po.
« Last Edit: April 19, 2018, 05:03:37 am by Marco »
 

Offline Karel

  • Super Contributor
  • ***
  • Posts: 1264
  • Country: 00
Re: Which diode to use?
« Reply #28 on: April 19, 2018, 05:06:35 am »
Maybe pull the I2C bus up to the Li-Po voltage if the other devices can handle that?

Or put an ultra low quiescent current LDO on the Li-Po.

I got the same idea but the fist chip I checked (which is connected to the same bus) doesn't like it  :(
Didn't bother to check the others.
Thanks anyway for helping.
The difference between theory and practice is less in theory than
the difference between theory and practice in practice.
Expensive tools cannot compensate for lack of experience.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 3806
  • Country: nl
Re: Which diode to use?
« Reply #29 on: April 19, 2018, 05:14:40 am »
Feeding the RTC with say a MCP1810 straight from the Li-Po would work, only run the rest of the circuitry off DC-DC.
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 1426
  • Country: fi
Re: Which diode to use?
« Reply #30 on: April 19, 2018, 05:28:06 am »
The Li-Po battery has an internal circuit that protects it from deep discharge. It disconnects the battery if the voltage is a little below 3V.
It's not a user replaceable part. Imagine the situation that the device is used till the Li-Po reaches almost 3V. The uC in the device detects
this and will warn and shutdown nicely. Then, if the user doesn't use it for 6 months, what is going to happen with the RTC...

Great that you are thinking about this specific use case. I'm sorry to say, but most do not; including the people designing the protection circuits! This is actually the most typical failure mode in all types of li-ion BMS and protection circuits I have seen, custom designed, discrete or ICs from the reputable big guys.

So, according to my experience, it's quite likely that a broken-by-design protection chip will destroy the cell (by first allowing complete discharge to near 0%SoC, which happens already at 2.9V cutoff if your load current is very small (less than, say, 0.05C)), then applying too high quiescent current in its own control logic. And then you need to replace the device with a broken battery. You can set the RTC again at the same time :).

This isn't necessarily the case, some protected batteries are OK, but I have measured quiescent currents far over 50 µA in "disconnected by undervoltage" circuits. Even if the cell survives the protection chip (i.e., it over-discharges it only slightly), the same protection chip won't probably allow the charging to happen, even if it was still safe to do (for example, the cell still reads 2.9V and would be safe to charge, or it reads 2.5V and would be safe to precharge at reduced rate). The threshold voltages of the PCM chips tend to assume voltage bounceback. In these cases, you can fix the issue by manually charging the cell, bypassing the protection chip. This was a classical failure mode in many laptop battery packs. Don't remember the exact IC anymore, but you ran it down to 0% like you normally do, then forgot it just for a few weeks, and bang, the cells were at about 2.9V IIRC, so completely OK to charge, but the pack was dead and refused to charge.

So, while I understand that a "protected cell" concept sounds like a simple task to allow you stop worrying, that is not always the case.

No one suggested not using cell protection and management, though. Usually in professionally designed fixed-battery devices, it's located on your PCB however, whereas the cell itself is provided with passive protections such as fuse against short circuit (PTC, resetting bimetal, or non-resetting classic fuse).

The reason for this is the assembly simplicity and part cost. Assembly simplicity is important; I have heard/read reports of cases where the "protection PCB" has been installed incorrectly, for example puncturing the cell or insulation sleeve in the case of a 18650 cell, causing a short before the protection chip. These incidents do happen because the actual cell manufacturers with high quality control and reputation factor do not tend to offer these "PCM'd" cells, but they are often provided by secondary sources, manufactured in small batches, main customers being hobbyists and clueless small companies/startups. Not saying this is always the case, reputable "complete integrated battery" manufacturers exist as well. Their products won't at least catch fire, but still, even the reputable products have failed in your "6 months on the table after complete discharge" use case.

The magical "protected cell" just moves this IC from your PCB to a small PCB next to the cell, and almost always lacks crucial safety controls such as cell temperature qualification (charge rate management/prevention in low temperatures), charge overcurrent trip point with different value from discharge overcurrent trip, etc., so that the designer now simply has: 1) total lack of understanding, combined with 2) false sense of security.

You may have a very good battery with a good PCM, or you may not. Just sayin'.

In any case, if you were to place the cell management on your own PCB - you could even use the exact same IC - now you have the choice! - then there would be no issue of taking that 5µA current directly off the cell - possibly with a simple series resistor such as 470R to limit fault current in case of short. This 5µA would be in the same order of magnitude as the quiescent current of the PCM (or even less, if it's a shitty PCM), so even if it ended up killing the cell by overdischarge in your scenario, that would have happened anyway...

You won't necessarily even need a disconnection switch. As long as your li-ion charger IC does the starting voltage&temperature qualification, as it definitely must, there is nothing unsafe doing your own cutoff in the microcontroller by shutting down external circuits and going to low-current sleep at low V (of course, this is not always feasible, depending on your external circuits and if they can power down easily). But guess what? This is the normal way to design these things, and exactly the reason why so many modern ICs provide power-down modes and boast about <1µA quiescent currents for battery operation. Worst case, you fail with the software, the cell is overdischarged, cannot be charged anymore due to charger IC qualification protection, and you need to replace it. Unless you have an over-the-air FW update mechanism, this is no different from any other bug.

But I guess you won't want to do that. Maybe it's paperwork, or the classical faulty "safety belief". These arbitrary constraints do happen in projects, and I understand that. But I still hope this point of view helps...
« Last Edit: April 19, 2018, 05:34:13 am by Siwastaja »
 

Offline Karel

  • Super Contributor
  • ***
  • Posts: 1264
  • Country: 00
Re: Which diode to use?
« Reply #31 on: April 19, 2018, 03:52:02 pm »
The Li-Po battery has an internal circuit that protects it from deep discharge. It disconnects the battery if the voltage is a little below 3V.
It's not a user replaceable part. Imagine the situation that the device is used till the Li-Po reaches almost 3V. The uC in the device detects
this and will warn and shutdown nicely. Then, if the user doesn't use it for 6 months, what is going to happen with the RTC...

Great that you are thinking about this specific use case. I'm sorry to say, but most do not; including the people designing the protection circuits! This is actually the most typical failure mode in all types of li-ion BMS and protection circuits I have seen, custom designed, discrete or ICs from the reputable big guys.

<Removed a big part of text here>

But I guess you won't want to do that. Maybe it's paperwork, or the classical faulty "safety belief". These arbitrary constraints do happen in projects, and I understand that. But I still hope this point of view helps...

Well, I don't want to go too far offtopic but generally, I agree with you. The problem is that I'm not at liberty to discuss all details
of the project here. What I can say is that we prefer to use Li-Po batteries that have already been tested and certified in a laboratory
for UN38.3. We don't want to do that ourselves with the complete product. Thera are already too many other tests we are obliged to do and pay for.
Second, the uC in the device observes the battery voltage and, in fact, it will shutdown before it reaches 3V. The exact level can
be set in the firmware. So, we have some more tolerance here before the battery will be damaged.
Thanks for the insight anyway.
The difference between theory and practice is less in theory than
the difference between theory and practice in practice.
Expensive tools cannot compensate for lack of experience.
 

Offline Karel

  • Super Contributor
  • ***
  • Posts: 1264
  • Country: 00
Re: Which diode to use?
« Reply #32 on: April 23, 2018, 06:54:36 pm »
I decided to use two diodes BAS116H. At 10uA the forward voltage is 0.4V according to the datasheet. So, at 2.5uA it will be even less.
Reverse laeakage current is very low: max 5nA@75V.

Thanks to everybody.
The difference between theory and practice is less in theory than
the difference between theory and practice in practice.
Expensive tools cannot compensate for lack of experience.
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Which diode to use?
« Reply #33 on: April 23, 2018, 07:59:13 pm »
I decided to use two diodes BAS116H. At 10uA the forward voltage is 0.4V according to the datasheet. So, at 2.5uA it will be even less.
Reverse laeakage current is very low: max 5nA@75V.

Thanks to everybody.

You can just use two different diodes - as the reverse leakage of the diode from the coin battery is not important, use a Schottky there, and for the diode from the LiPo use a low leakage diode (BAS116, BAV199 etc).

Cheers

Alex
 

Online Mr. Scram

  • Super Contributor
  • ***
  • Posts: 5241
  • Country: 00
  • Display aficionado
Re: Which diode to use?
« Reply #34 on: April 23, 2018, 08:09:09 pm »
Do I understand correctly that the diode is there to provide reverse current protection? Can't you use a FET to achieve something similar without the voltage drop? I'm probably missing something.
 

Offline Karel

  • Super Contributor
  • ***
  • Posts: 1264
  • Country: 00
Re: Which diode to use?
« Reply #35 on: April 23, 2018, 10:32:37 pm »
Do I understand correctly that the diode is there to provide reverse current protection? Can't you use a FET to achieve something similar without the voltage drop? I'm probably missing something.

There's almost no space. This extra backup battery has been added at the last moment  :(
The difference between theory and practice is less in theory than
the difference between theory and practice in practice.
Expensive tools cannot compensate for lack of experience.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf