DC512B not working

Hi Fabrizio,

The switching looks normal. Does it seem to be receiving the current that it is requesting?

The charging seems to stop after a short time. Did you stop the charging or does that keep happening repeatedly? Try increasing your input power - more voltage or more current. You may be power-limited at startup.

If you have a current probe, maybe take a look at the battery current.

Regards,

Zack

Hi Zack,

thanks for the answer.

Today I retried with a new test, following your suggestion and in that conditions:

• 3S battery connected and already charged at 78%. The battery requires less than 1 A in that condition. Its voltage was about 10.70v
• No load connected on the DC512B
• ILIM set to 3A, VLIM 3CELL
• No SMBUS connected
• Thermistor of 10K connected.
• Power Supply increased from 15 to 24V, 3.2 A

ACP, CHGEN high.

The behaviour is the same of yesterday. I can hear switching glitches from the coil, I can see periodic absorption on the power supply (which didn't enter in limitation, only 3-5 W of peak achieved) .

I never stopped the charging but it is happening repeatedly.
I also tried to reduce the cable length, but nothing is changed.

I have not the current probe. Can you suggest other checks?

Hi Fabrizio,

Sure, a couple more things to check.

Take a look at the INFET signal. Is it steady ON or is it turning off with this function?

Check on SDA/SCL while this happens, is there communication that stops the charging?

Regards,
Zack

Hi Zack,

today I tried to make the new test that you suggested, but after switched on the setup, the transistor Q11 burned.

I suppose the same problem of this topic: 

https://ez.analog.com/power/f/q-a/116863/ltc4100-evaluation-board-q11-fet-issue/348352#348352

Hi Fabrizio,

It could just be a hot-swapping issue with a high-voltage battery. Try replacing the FET and adding a 10k pull-down resistor on the gate.

More info in this blurb:

When a battery is inserted with no other supply present, the battery will pull up the BAT and CSP
pins and will begin to charge the inductor which also pulls up the SW node. The CLP and CLN pins
will also be pulled up via the body diode of the top FET. The exact speed and order of this
sequence depends on many factors largely influenced by parasitics.
During these events, the BGATE pin is not driven by the IC (high or low). After a hot plug event, the
IC typically requires about a millisecond or so to take active control of BGATE. Without an active
pull-down, SW pull-up energy couples through the bottom FET Miller capacitance, turning on the
FET if its threshold is exceeded. If the bottom FET turns on for an extended period of time, it can be
destroyed by the simultaneous combination of Vds and Ids.
We have been unable to replicate FET failure in the lab, but we have been able to initiate bottom
FET conduction via the capacitive coupling discussed above. Note — the likelihood of turning on the
bottom FET depends on many factors including FET threshold and battery voltage. Given the ratio
of Miller capacitance to gate capacitance, it is unlikely that battery voltages below about 10V can
cause this problem. Therefore, 1-cell and 2-cell Li-ion applications are probably safe.

Solution: A simple solution to this issue is to ground the BGATE pin with a resistor. 10k seems to
work well and has minimal impact on switcher efficiency.

Regards,

Zack

Hi Zack,
Thank you very for the explanation.
Two main topics:
1) The board is under warranty, so I think that I cannot change the component by myself.
2) This limitation should be documented in the “QUICK START GUIDE FOR DEMONSTRATION CIRCUIT DC512B” because in the real world it is completely typical to insert (or to leave) the battery inside the device, connected to the smart charger without power supply (as for example the laptop computer).

Hi Fabrizio,

I'm not sure that's the issue here, but it is worth trying as a preventative measure if you are using higher-voltage batteries.

As for the board under warranty, if you would like to send it back you may. Otherwise I recommend just swapping out for a new FET. It is very typical for customers to modify our boards for their own use.

If you are in contact with your local field applications engineer, they might just be able to get you a replacement board while you modify the board you are currently working with.

Regards,

Zack

Hello Zack,

Just a follow up to this issue. I am also having a similar issue where Q11 burns on hot-plugging battery.
I will add the pull down resistor. However, is there a recommended procedure to avoid this issue in the field, as the DC-IN will only be present only during charging and battery power will be used for most of the operation.

Thanks,
Hardik

Hi Hardik,

One thing that may alleviate this issue would be applying DCIN first before plugging in the battery. This ensures the buck charger's input caps are charged so there will be less inrush when connecting the battery. I can't say this will 100% remove the problem for all setups, but it might work for yours.

Adding the resistor is the more guaranteed method.

Regards,

Zack

Zack,  I am having the same issue with irregular behavior when hot swapping the battery.  I have several boards which we have fabricated, where the IC comes up and regulates to voltage, but as soon as I connect the battery, the IC starts to charge, ramps up current beyond the settings above 2A (SMBus programmed at 16.8V, 750mA as a conservative low charge current), and then the IC fails to set voltage again.  I am unable to replace the ICs, as they are not in stock anywhere and we have them on backorder.   I am going through my production boards at 100% failure.
After the IC fails, it no longer tries to regulate voltage.  Inductor is Taiyo NS10165T220MNA.

I have just received a DC512B eval board, and I am going to try working with that.

thanks,

-Tom

Hi Tom,

The one thing that comes to mind in your case is that we have seen instances of battery inrush causing noise on the bottom FET gate and causing it to turn on briefly upon connection. This results in the expected catastrophic failure of such a scenario.

Try putting a 10kOhm pull-down to GND on the gate of the bottom FET. This will not hurt anything. If nothing else, this will tell us whether this is your issue or not.

Regards,

Zack

Zack,

I tried putting the 10K pull down on the bottom FET, but this did not fix the issue.  This was on 2 new boards from the production run.  The battery was not connected, and the ICs worked for a couple seconds, and then stopped pulsing on the tgate and bgate signal lines.  Basically, the ICs work for a brief amount of time, then go off into a failure state.   They accept commands through the I2C interface, and the registers read back OK, but the charger section goes into a steady state.  The circuit card routing has taken the Kelvin connections to the sense resistors into account, and I just can't wrap my head around the circuit behaving in such a sensitive manner.  It basically does not give any time to evaluate what is going on before the circuit fails to a permanent state.

This circuit card we are working on has 2 LT3790 power supply circuits at 12V and 20V,  LTC3621 3.3V supply, LTC4368 & LTC4418 power path and protection circuits.   All these other functions work just fine.

The DC512B eval board I received Monday works fine with the commands I am issuing,  it regulates to 16.8V, and charges at 750mA as I command it.

Attached is image of the board layout.  Top FET has footprint issue which was corrected by deadbugging the FET package. Bottom FET has 10K attached to gate.

Zack,

I tried putting the 10K pull down on the bottom FET, but this did not fix the issue.  This was on 2 new boards from the production run.  The battery was not connected, and the ICs worked for a couple seconds, and then stopped pulsing on the tgate and bgate signal lines.  Basically, the ICs work for a brief amount of time, then go off into a failure state.   They accept commands through the I2C interface, and the registers read back OK, but the charger section goes into a steady state.  The circuit card routing has taken the Kelvin connections to the sense resistors into account, and I just can't wrap my head around the circuit behaving in such a sensitive manner.  It basically does not give any time to evaluate what is going on before the circuit fails to a permanent state.

This circuit card we are working on has 2 LT3790 power supply circuits at 12V and 20V,  LTC3621 3.3V supply, LTC4368 & LTC4418 power path and protection circuits.   All these other functions work just fine.

The DC512B eval board I received Monday works fine with the commands I am issuing,  it regulates to 16.8V, and charges at 750mA as I command it.

Attached is image of the board layout.  Top FET has footprint issue which was corrected by deadbugging the FET package. Bottom FET has 10K attached to gate.

Thanks Tom,

Can you share a schematic? I'm sending you a friend request in case you would like to share privately.

Are you certain there are no shorts after deadbugging the FET and that it is making the right connections now?

Does that make the FET below it shown here the bottom switching FET? If so, I have some doubts about the GND connection of the FET. I don't see vias making a solid GND connection. I can take a look at the layout as well if you send that.

Regards,

Zack

Zack,  The schematic is in the previous post in this thread dated 9/22/2020.  I am away from office now, but you may have found the issue.   The bottom FET source sure does not look like it is connected to ground as well as C112.  The schematic shows a ground connection, but the net may not be tied properly.  I will do a net by net checkout and let you know the results.  Thank you very much.

-Tom

I received the DC512B evaluation board, and the circuit worked properly.  I ordered the same mosfets as used on the eval board and when they arrived, I began replacing the ones selected in our design with the eval types.  It turns out that we were having 100% failure on the top mosfet, which was an IRF7424.  When it was replaced with the Si4431, the circuit started working properly.  The parameters for the Si4431 and the IRF7424 are nearly identical, but for some reason, the circuit likes the Si4431.

Hi Tom,

Thanks for the update. I would guess the cause here is the difference in gate charge. THe IRF7424 has about 6x the gate charge of the Si4431ADY. Perhaps this was causing a slow transition which may have lead to heating and eventually destruction (or some other catastrophic event as a result of the FET being on at the wrong time).

Glad you got this working, though, and thank you for reporting back.

Regards,
Zack