BQ76952: Using Predischarge autonomously

.. and the Similar Topics column opened up suggestions that I didn't see in my forum search.

A year ago, Shyama Agrawal, asked a similar question where inrush created a short circuit event first (which I am wanting to avoid). See:

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1075634/bq76952-predischarge-corner-case-in-fully-autonomous-mode

He, like me, wants to run autonomously but I think Luis missed this (sorry Luis) and detailed how to do make this work with a host MCU.

TI has marked his post TI Thinks Resolved but I suspect now that pre-discharge is not available if you have an autonomous design. Unresolvable.

If there is an autonomous way, I really need to know.

All the best
Harry

If (and only if I have to) attach a small MCU simply to achieve predischarge in the way I describe above is the following possible ?

If I leave Settings:FET:Predischarge Timeout and Settings:FET:Predischarge Stop Delta both zero, and then use the small MCU to turn on the DSG FET will this actually turn on the PDSG FET indefinitely ?

If so, that is a way to force the condition I want (PDSG FET enabled).

Then, if I sense for Power:Sleep:Sleep Current or Power:Sleep:Wake Comparator Current have moved the state to NORMAL, I can adjust Settings:FET:Predischarge Timeout and/or Settings:FET:Predischarge Stop Delta for the predischarge conditions I -actually- want and cycle DSG FET off and back on to initiate a predischarge cycle.

When the load drops away and returns the state to SLEEP, I can clear both Settings:FET:Predischarge Timeout and/or Settings:FET:Predischarge Stop Delta, cycle the enable on DSG FET and that should restore the indefinite PDSG FET enable.

This only works if there is some overlap between the enables for DSG FET and PDSG FET. Otherwise I will briefly extinguish the pack output and there may still be a keep alive circuit relying on the 20mA that I choose as the Power:Sleep:Sleep Current.

All the best
Harry

Hi Asher

Thank you for the follow up. I have been drawn into other issues. I returned to this several days ago but I must still be missing something.

First I set everything up to use BQStudio as a surrogate external MCU (not autonomous behaviour):

Power:Sleep:Sleep Current = 20mA
Power:Sleep:voltage Time = 5sec (minimum required)
Power:Sleep:Wake Comparator Current = 500mA but also WK_SPD = 0 so less noise
Power:Sleep:Sleep Hysteresis = 10sec
Settings:FET:Predischarge Timeout = 50 (500mSec) <- this is left default initially
Settings:FET:Predischarge Stop Delta = 1000mV
PDSG_EN = 1 (was already from default OTP)
SFMODE_SLEEP = 1

Sequence 1
----------------

The PACK pins have no load applied as if the pack is uninstalled.

Now, with BQStudio doing 100mSec scans of just the Battery Status and FET Status registers I see:

DSG is ON (other FETs OFF) and in Sleep state
Send ALL_FETS_OFF() .. still remains in sleep state
DSG goes OFF as expected
Send ALL_FETS_ON()
SLEEP => 0 (transition to Normal)
PDSG goes ON
500mSec elapses (Settings:FET:Predischarge Timeout) (this is odd because, with no load, Predischarge Stop Delta should be reached earlier than Timeout)
DSG goes ON
CHG goes ON
PDSG goes OFF
10Sec (Power:Sleep:Sleep Hysteresis) elapses after SLEEP => 0 transition to Normal above
CHG goes OFF
SLEEP => 1 (transition to Sleep)
Note that it does not restore DSG = OFF and PDSG = ON

Where you explain:
"Source-follower allows the DSG FET remain off while in SLEEP mode until a discharge current is detected.",
what I do not see is transition to Sleep re-enabling PDSG so the pack is ready for a future surge from a capacitive load as the pack pins are connected to that load. This is the requirement to avoid a plug-in surge (spark) that Predrag was trying to achieve here:

e2e.ti.com/.../bq76942-state-of-pdsg-and-dsg-during-sleep-state-in-order-to-prevent-spark-upon-connection

Sequence 2
----------------

If I now apply a 100ohm resistive load across the pack, the pre-discharge sequence doesn't commence. That is, PDSG is not set for 500mS etc. Instead DSG remains on and is the same as above except, after 10Sec, the state remains in Normal since the pack has load. When the resistive load is removed, any of the remaining 10Sec Power:Sleep:Sleep Hysteresis runs and:
DSG remains ON
CHG goes OFF

So these two sequences are behaving as described for pre-discharge with BQStudio in the role as the external MCU; that is, only the MCU turning ALL_FETS_OFF() and ALL_FETS_ON() initiates a single pre-discharge sequence.

Sequence 3
----------------

I pressed on to try to achieve autonomous pre-discharge sequences. With no load on the pack, I set:
Settings:FET:Predischarge Timeout = 0  .. which initiates ..
SLEEP => 0 (transition to Normal)
DSG is OFF
PDSG is ON
CHG is ON
Send ALL_FETS_OFF() (still not autonomous)
SLEEP => 1 (transition to Sleep)
All FETs go OFF as expected
Send ALL_FETS_ON() (still not autonomous but what happens?)
SLEEP => 0 (transition to Normal)
PDSG goes ON
Short delay
CHG goes ON
10Sec elapses
--No transition back to SLEEP--

I cannot see a way that source follower is enabled and PDSG_EN=1 such that the state transitions back to Sleep as it should in the absence of discharge current.

Is autonomous pre-discharge meant to remain in Normal state and the trade-off is higher power consumption in Normal state ?

Is the PDSG sequence only initiated if a capacitive load pulls the ToS to PACK voltage delta *larger* than FET:Predischarge Stop Delta ?

Sequence 4
----------------

I continued and attempted to use a resistive load to cause the pre-discharge to complete when ToS to PACK voltage delta is smaller than 1000mV (Settings:FET:Predischarge Stop Delta).

I have a 10S stack at 36V and changed from 170 ohms to a 1 ohm pre-discharge resistor.
In the unloaded situation above the ToS to PACK delta is already 0V (pack voltage supplied through the 1 ohm resistor)

I connect a 100ohm resistive load.
ToS to PACK delta becomes 0.66V (less than FET:Predischarge Stop Delta)
There is no change to FET enables, that is, PDSG remains ON and CHG remains ON
In this state the PDSG resistors are likely to be damaged by carrying continuous load indefinitely because there is no transition to DSG = ON.

I cannot see a register setting that allows Sleep and PDSG = ON to be reached so an unconnected pack is ready to run a pre-discharge sequence when a capacitive load is attached.

I also acknowledge that this can only be used in situations where the connected load has a capacitive surge and then continues to use current (higher than sleep current).

For the alternative where the capacitive surge current is followed by a small quiescent current (lower than sleep current) then going back to sleep and re-enabling PDSG could be inappropriate if the load expects full available current from the pack from this point forward.

So autonomous PDSG isn't for every situation but I am missing how to make a PDSG sequence work autonomously in any situation.

All the best
Harry

Dear Harry,

I will go into our labs tomorrow to work through the sequences you performed to see if I can verify them. I will get back to you after that.

Best,

Asher Nederveld

Thank you Asher

I can post more of the configuration if you need it.

All the best
Harry

Dear Harry,

After some testing and discussing with the systems engineer for the BQ76952, I am seeing similar results that you had. PDSG is meant to handle full turn-ons where all FETs are OFF prior. Enabling source follower won't be enough to re-enable PDSG as it doesn't completely disable the FET, but uses the BAT pin's voltage to power it. I apologize for the misunderstanding which I had on that  earlier for it re-enabling PDSG on entering SLEEP.

To go back to your earlier question about how to configure that manually, you would need to put the device into TEST mode. This is done by setting the setting the Settings:Manufacturing:Mfg Status Init[FET_EN] bit low. Once this is done, the FET control will be manual but you will gain access to the PDSGTEST() and PCHGTEST commands, which toggle the FETs ON and OFF.

Another option would be to program the MCU such that whenever the device goes into SLEEP (by monitoring the WAKE bit in the status register), you would send an ALL_FETS_OFF() command and an ALL_FETS_ON() command once a load is detected. This would be easier to implement on the software side then the previous suggestion.

Best,

Asher Nederveld