BQ25750: REGN stuck at less than 1 volt

Hello Phil,

Thanks for working on this. I have a few testing suggestions to diagnose REGN being low:

• Is anything draining power from REGN?
• REGN is shorted to DRV_SUP, right?
• Can you measure REGN with an oscilloscope?
• Has anything had a chance to damage the IC?
• Can you send me the part numbers for the FETs and inductor that you are using?

By the way, the BTST capacitors should be 100nF. I also recommend installing an RC filter in parallel with the ICHG resistor. We recommend this when this resistor is greater than 5kΩ. You can use the EVM for reference here.

Best Regards,
Ethan Galloway

Hi Ethan,

Thank you for your prompt reply.

> Is anything draining power from REGN?
Just what is in the schematic (TS resistor ladder and DRV_SUP cct)

> REGN is shorted to DRV_SUP, right?
Correct

> Can you measure REGN with an oscilloscope?
Yes, I see it rise to 250mV about 40msec after the input goes to 20V.  It remains steady at 250mV until the input is turned off

> Has anything had a chance to damage the IC?
Unless there is something wrong with my design, I don't think so.  The I2C and ACDRV and BATDRV portions seem to be working properly.

> Can you send me the part numbers for the FETs and inductor that you are using?
Q6, Q7, Q8, Q9 use SIR880BDPT1RE3
Q4, Q5, Q10, Q11 use PSMN3R060B
L4 is IHLP6767GZER100M01

>  the BTST capacitors should be 100nF
Hmmm.... The typical application shown in Figure 9-1 of the datasheet shows 47nF.  Is that out of date?

Thanks again for your help

Phil

Hello Phil,

Thanks for the new information.

Phil Rittenhouse said:

The typical application shown in Figure 9-1 of the datasheet shows 47nF.  Is that out of date?

The capacitors should be 100nF. I'll try to update this in the future.

Can you measure the following pin voltages while the BQ25750 is trying to charging?

• VAC
• VBAT
• SRN
• DRV_SUP
• TS
• ACUV

Also, are there any other status or flag registers being set in the charger?

Best Regards,
Ethan Galloway

Above are the pin voltages.  I'll check the status/flag registers next

Here are all the status and flag register values I could find (the rest seem to be only for control and, since I am not writing them, should be set to their defaults).  As before the Power Good flag is set as is the DRVSUP out of range flag

20:51:33.188 -> 0x17: 0xc9

20:51:33.188 -> 0x1F: 0x0

20:51:33.188 -> 0x21: 0x8
20:51:33.188 -> 0x22: 0x80
20:51:33.188 -> 0x23: 0x2
20:51:33.188 -> 0x24: 0x2
20:51:33.188 -> 0x25: 0x0
20:51:33.188 -> 0x26: 0x0
20:51:33.188 -> 0x27: 0x0

Let me know if there are any I missed that might be helpful

Best regards,
Phil

Hello Phil,

Thanks for the new information.

I have a few more test questions:

• If possible on your board, can you disconnect REGN from DRV_SUP? Does REGN come up then?
• On an oscilloscope, do you any switching action on SW1 or SW2?
• If you replace the IC, does REGN come up then? I think this would be the most useful test to do here.
• What does the circuit do if you power DRV_SUP with 5V? This could be a dangerous test on the circuit by the way. I would only recommend this if you have a quick way to replace the IC.

Best Regards,
Ethan Galloway

Hi Ethan,

>  If possible on your board, can you disconnect REGN from DRV_SUP? Does REGN come up then?
Unfortunately no.  The pads are connected in the copper layer under the chip

> If you replace the IC, does REGN come up then? I think this would be the most useful test to do here.
I agree but removing and hold soldering this chip is a bit of a challenge.  I was wondering if a bad solder joint could be the cause.  I have visually inspected and all the joints looks OK but most of the pad is hidden underneath.   Are there specific pins that could cause symptoms like this if they were not connected?  I can try reflowing those to see if it helps

> On an oscilloscope, do you any switching action on SW1 or SW2?
No, they are both steady at 1.4V

Thanks,
Phil

Hello Phil,

Phil Rittenhouse said:

> If you replace the IC, does REGN come up then? I think this would be the most useful test to do here.
I agree but removing and hold soldering this chip is a bit of a challenge. 

Yeah, replacing the chip really is a challenge. Reflowing the chip might be an easier test to check for bad solder joints.

Phil Rittenhouse said:

Are there specific pins that could cause symptoms like this if they were not connected? 

Most of the time when I see these symptoms like these, either the IC has been damaged or the FETs have been damaged.

Phil Rittenhouse said:

> On an oscilloscope, do you any switching action on SW1 or SW2?
No, they are both steady at 1.4V

It may be good to also measure HIDRV and LODRV as well.

On a different note, do you know what the quiescent current of the IC is? Checking this current is a way to try to check for damaged components.

Best Regards,
Ethan Galloway

> what the quiescent current of the IC is? 

I measure 50mA going into VSink with a 1Kohm load on the output SYSPWR.  With no load on the output, I see 30mA being drawn

Here's what I measure for all the pins from 28 to 19 (all are steady voltages)

SW1 = 600mV

HIDRV1 = 340mV

BIST1 = 320 mV

LODRV1 = 0

REGN = 280mV

DRV_SUP = 280mV

PGND = 0

LODRV2 = 0

BIST2 = 310 mV

HIDRV2 = 320mV

No idea why SW1 is different now.  I checked SW2 and its 600mV too.

As an experiment, I built up a fresh circuit with new parts but with only the parts I thought would be necessary for REGN to be activated.  See the attached schematic.  The parts highlighted in purple were not populated.

When I powered it up there was no voltage on REGN or SYS.  There was also no measurable current being drawn (<1mA).   I read the following register values from I2C:

18:39:08.067 -> 0x1F:  0x0

18:39:08.067 -> 0x21:  0x8

18:39:08.067 -> 0x22:  0x0

18:39:08.101 -> 0x23:  0x0

18:39:08.101 -> 0x24:  0x0

18:39:08.101 -> 0x25:  0x0

18:39:08.101 -> 0x26:  0xe0

18:39:08.101 -> 0x27:  0x0

18:39:08.101 -> 0x17:  0xc9

which seems to indicate that the POWER GOOD flag is not set.  I scoped the VAC, ACUV and ACOV and got 20, 20 and 0 V as expected.   Is there something else that would cause the power good flag not to be set?

Best regards,
Phil

Hi Phil,

The POWER GOOD flag checks for a voltage between the valid operating window of ACUV and ACOV. I am not aware that there is another condition for this flag. If there is one, it would likely be related to REGN meaning that it is a symptom of the problem rather than the cause. Since POWER GOOD was set initially, I do not think that this is the case.

The change from POWER GOOD to now not being POWER GOOD, and no voltage on SYS, makes me think that something else changed. When you built up the new circuit, did you resolder anything? This could be an issue where the power is not reaching the VAC pin. This would be needed to turn on the ACFETs.

Best,
Michael

> When you built up the new circuit, did you resolder anything?
Yes, i used all new parts on a new PCB

I measured 20V at the VAC pad and the solder joint does appear OK.   Does the chip get power from any other pin?  I assumed that it must have power on VAC since the I2C was working but maybe it comes from somewhere else?

Best regards,
Phil

Hello Phil,

The IC receives power from either VAC or SRN (SRN gets power from the battery).

With that mind, can the IC communicate over I2C with only the battery or with only VAC?

Also, can you measure the voltage on SYS as well?

Best Regards,
Ethan Galloway

Thanks Ethan,

I don't have a battery hooked up in the newest circuit so I2C must be getting power from the VAC

The SYS pin reads 0 Volts

Best regards,
Phil

Hello Phil,

Thanks for the updates and thanks for being patient with this behavior.

Does the circuit behavior change if you short across the ACFETs?

Also, is CE being pulled LOW?

Best Regards,
Ethan Galloway

Thanks Ethan,

Yes the CE pin is connected to GND.  Shorting across the ACFETs I see the output SYS level go to 20V as expected but REGN remains off and the I2C registers remain the same.

Speaking of the I2C registers.  Is 0x26 expected to be 0xe0?   If I am reading the datasheet correctly, this indicates that PG signal toggle detected, ACFET signal toggle detected and BATFET signal toggle detected.  Does this mean power good was true briefly?

Hello Phil,

Thanks for being patient with all of this debugging.

Phil Rittenhouse said:

Is 0x26 expected to be 0xe0?

This register is 0x00 on startup.

Phil Rittenhouse said:

Does this mean power good was true briefly?

I think so and that makes this problem pretty strange.

I've got a few questions to help debug this:

• Can you desolder the BATFETs? I think this will make testing easier because there will be less variables to work with.
• Does the circuit operate in reverse mode?
• This is probably unrelated to the current behavior at hand, but what diodes are you using for the BTST diodes?

Best Regards,
Ethan Galloway

The diodes are V1FM10-M3/H (not currently populated in my simplified test).

I will try your other suggestions when I return from vacation next week.

Best regards,
Phil

Hello Phil,

Thanks for working on this. Let us know what you find out.

Best Regards,
Ethan Galloway

Hi Ethan,

I tried removing the BATFETs (Q10 and Q11) but got the same results

12:36:21.533 -> 0x1F: 0x0
12:36:21.533 -> 0x21: 0x8
12:36:21.581 -> 0x22: 0x0
12:36:21.581 -> 0x23: 0x0
12:36:21.581 -> 0x24: 0x0
12:36:21.581 -> 0x25: 0x0
12:36:21.581 -> 0x26: 0xe0
12:36:21.581 -> 0x27: 0x0
12:36:21.581 -> 0x17: 0xc9

i.e. power good is not set

Can you provide more detail on how to test reverse mode with my current circuit?

Also, I was wondering if there are any I2C registers I could write that would help debug this?  Or any undocumented registers we could read?

Thanks,
Phil

Hello Phil,

Thanks for being patient with this.

Phil Rittenhouse said:

Can you provide more detail on how to test reverse mode with my current circuit?

To test reverse mode, I would recommend pulling the ACUV pin LOW and using about 10V for the battery. You can use a regular DC power supply for the battery. Then, writing EN_REV=1 should activate reverse mode.

To load the charger, you can use a regular E-load in CC mode or CR mode.

Phil Rittenhouse said:

Also, I was wondering if there are any I2C registers I could write that would help debug this?

Do you have a USB2ANY or an EV2400 on hand? This won't solve the charger's behavior, but this will make setting the registers and reading the registers easier.

I'm not sure of any more registers to test that would be helpful. Does setting EN_TERM=0 and EN_PRECHG=0 cause the circuit to charge?

I'll try to get back to you later this week with more information.

Best Regards,
Ethan Galloway

Hi Ethan,

I think I have made a little progress on this.  It looks like my simplified circuit was actually the cause of the power good issue.  I added in the shunt components across SRN and SRP and now I get the power good to go high, REGN is normal and charging is enabled.  Here are the I2C registers

18:04:22.109 -> 0x1F:  0x0

18:04:22.109 -> 0x21:  0x9  => WD timer expired + Trickle charge

18:04:22.109 -> 0x22:  0x80 => Power Good

18:04:22.109 -> 0x23:  0x2  => ACFET on

18:04:22.109 -> 0x24:  0x0  => No faults

18:04:22.109 -> 0x25:  0x49 => Device entered current regulation + WD Flag + Charge state changed

18:04:22.109 -> 0x26:  0xc0

18:04:22.109 -> 0x27:  0x0

18:04:22.157 -> 0x17:  0xc9 => Default

I then repopulated the rest of the circuit but now it looks like it's going immediately into thermal shutdown:



10:31:09.756 -> 0x21:  0x8 => WD and NOT CHARGING

10:31:09.756 -> 0x22:  0x0 => Power is not good

10:31:09.804 -> 0x23:  0x0

10:31:09.804 -> 0x24:  0x8 => Thermal shutdown

10:31:09.804 -> 0x25:  0x8 => WD rising edge

10:31:09.804 -> 0x26:  0xe0

10:31:09.804 -> 0x27:  0x28 => Batt overcurrent, Thermal shutdown

10:31:09.804 -> 0x17:  0xc9

I will double check my work for errors, but can you suggest any likely causes for this?

Best regards,
Phil

I found that the 10 ohm resistors in the SRP/SNP filter were now measuring >500ohms.  I'm guessing that for some reason a large amount of current damaged them.  I tried replacing them along with the other filter components but I still get the same results on I2C.  I suspect the chip has been damaged.

The only failure condition in which I could see the 10 ohm resistors carrying a lot of current would be if the shunt resistance was high but it still reads less than 100 mohm (which is the limit of my meter).

After removing the filter components I noticed that the resistance from SRN or SRP to GND was now ~20 ohms.  Checking on a fresh chip I see infinite resistance between SRN or SRP to GND.  I suspect that the same event that damaged the resistors has also damaged the chip.

I have looked over the design again but I'm at a loss to explain what could have caused the high current through the battery current sense circuitry.  Do you have any thoughts on how to safely test this?

Thanks,
Phil

Hello Phil,

Thanks for the new information. I'm glad to hear that you are getting the power good signal ON now.

For your tests, I have a few questions:

• What was your input and output voltage setting?
• What was the charge current setting?
• Did you have a load on the battery output?

Phil Rittenhouse said:

I found that the 10 ohm resistors in the SRP/SNP filter were now measuring >500ohms.  I'm guessing that for some reason a large amount of current damaged them.

The SRN and SRP 10Ω resistors can be destroyed with a sufficiently high current through the battery current sense resistor. If the current through the sense resistor is greater than 100A, the voltage across the sense resistor can cause voltage to flow through the SRN and SRP resistors and then through the IC itself. The IC is most likely damaged after this.

This is most commonly seen if the BATFETs turn-on with a large voltage across the FETs. We have an FAQ for the BQ25750's power path switchover that may help with this.

Phil Rittenhouse said:

Do you have any thoughts on how to safely test this?

I recommend doing the tests with the input and output voltages below 25V. Let me know if you were already testing this way.

Best Regards,
Ethan Galloway

> What was your input and output voltage setting?
I'm not sure what you mean by setting, but I was using 20V input and a 25.2V (nominal) battery.

> What was the charge current setting?
I am currently using  200K on ICHG to GND to give 250mA max charge current.  I also have 3,9K on ILIM_HIZ to limit the input to 5A

> Did you have a load on the battery output?
No.  Do you recommend one?  Is so, what do you recommend?

I will take a look at the FAQ

I might have a lower voltage battery I can test with if you think 25.2V is a problem.

Thanks,
Phil

I was looking at the demo board and I noticed that the battery shunt resistor is much larger (physically) than the one I am using.  I sized mine based on the expected charge current being less than 1A  so it's only 1/8W vs. 5W on the demo board.  I'm guessing that the high inrush current is too much for my little resistor and the voltage across it spikes high enough to damage the other components.

I checked the datasheet but didn't see a recommended minimum power rating for the sense resistors.  Is 5W the recommended size?

Best regards,
Phil

Hello Phil,

Phil Rittenhouse said:

I checked the datasheet but didn't see a recommended minimum power rating for the sense resistors.  Is 5W the recommended size

I would recommend a larger sense resistor. Although, keep in mind that a 5W sense resistor was chosen to handle the max current of 20A. You may not need a resistor this size.

Phil Rittenhouse said:

I'm not sure what you mean by setting, but I was using 20V input and a 25.2V (nominal) battery.

This is the information. I was looking for. Thanks.

Phil Rittenhouse said:

> Did you have a load on the battery output?
No.  Do you recommend one?  Is so, what do you recommend?

I wouldn't recommend another load. The battery already functions as a load for the charger.

One test you might consider is removing the BATFETs from the circuit. This will allow you to isolate whether or not the inrush current is coming from the BATFET turn-on or not.

Best Regards,
Ethan Galloway

Hi Ethan,

I have built two more charging circuits using the new 5W shunts.  All components are installed except the BAT FETs.

In both cases the Power Good state is set but in the first case the battery charge current is much high than it should be (2.2A vs 500mA with ICHG resistor = 100K).  I get the same result with or without the filter across the ICHG resistor.  Here are the registers for this case:

21:16:00.117 -> 0x1F:  0x0

21:16:00.117 -> 0x21:  0xb => WD, Fast charge

21:16:00.117 -> 0x22:  0x80 => POWER GOOD

21:16:00.117 -> 0x23:  0x2

21:16:00.117 -> 0x24:  0x0

21:16:00.117 -> 0x25:  0x0

21:16:00.117 -> 0x26:  0x0

21:16:00.117 -> 0x27:  0x0

21:16:00.117 -> 0x17:  0xc9

In the second case I get no charge current and the battery overvoltage fault.  I tried scoping the voltage on FB and found it was normally 

1.42V with periodic brief spikes to 2.6V.  Here are the registers for this case:

23:20:30.790 -> 0x1F:  0x0

23:20:30.790 -> 0x21:  0x8

23:20:30.790 -> 0x22:  0x80 => Power good

23:20:30.836 -> 0x23:  0x2  => ACFET on

23:20:30.836 -> 0x24:  0x10 => VBAT OV fault

23:20:30.836 -> 0x25:  0x49

23:20:30.836 -> 0x26:  0xe0

23:20:30.836 -> 0x27:  0x12 => VBAT OV fault, DRV_SUP pin fault detected 

23:20:30.836 -> 0x17:  0xc9

Do you have any idea what could be the cause in either of these cases?

Best regards,
Phil

Hello Phil,

That's strange that both of your boards have different behaviors.

For the 2nd board, what happens if you short the FBG pin to GND?

Also, I'm pretty swamped right now and I'll try to answer your question in more detail later on this week.

Best Regards,
Ethan Galloway

I tried shorting FB to FBG but that let the magic smoke out of the chip after a few seconds.  Before that though the fault did not appear to be happening.  stat1 was blinking on briefly while stat2 was off indicating it was trying to charge before stopping.

I scoped the voltage at the battery shunt on the first board and it had about 12V peak to peak high frequency noise from 23-35V concentrated at about 28V.  

I scoped the voltage at ICHG and saw a noisy sawtooth with a peak at about 1.5V and undershoot ringing down to -1V.

It seems my circuit has a lot of noise.  I noticed that the demo board has a lot of extra caps.  Could that be the issue?

The demo board has a 100nF in addition to the 1uF on the VAC input.  I also noticed that the 2.2uF filter? on VAC has a 10 ohm resistor in series on the demo but I have a 2 ohm.  I'm not sure what the purpose of this filter is?  Would 2 ohms be a problem?

Any other suggestions or things to try would be greatly appreciated.

Hello Phil,

Phil Rittenhouse said:

I tried shorting FB to FBG but that let the magic smoke out of the chip after a few seconds

I wouldn't recommend doing this. This is going to tell the IC to output the highest voltage possible to try to raise the voltage of the FB pin.

Phil Rittenhouse said:

I scoped the voltage at the battery shunt on the first board and it had about 12V peak to peak high frequency noise from 23-35V concentrated at about 28V

That's not good. Can you send a layout / picture of your board so we can see how the capacitors by the battery current sense resistor is arranged? When you measure the voltage at the battery shunt, are you measuring at SRN / SRP or on the battery shunt itself.

Best Regards,
Ethan Galloway

Hello Phil,

Phil Rittenhouse said:

I also noticed that the 2.2uF filter? on VAC has a 10 ohm resistor in series on the demo but I have a 2 ohm.  I'm not sure what the purpose of this filter is?  Would 2 ohms be a problem?

This RC filter helps protect against hotplugs. It probably won't effect your circuit that much.

Phil Rittenhouse said:

I scoped the voltage at ICHG and saw a noisy sawtooth with a peak at about 1.5V and undershoot ringing down to -1V.

It seems my circuit has a lot of noise.  I noticed that the demo board has a lot of extra caps.  Could that be the issue?

I think this could be because of low capacitance or a layout issue. Can you send us your layout so we can review it?

Best Regards,
Ethan Galloway

Thanks Ethan,

Here are screen captures of the relevant section of our design's front and back copper layers.  

Let me know if you would prefer PDFs of the layers or some other format.

Best,
Phil

In the meantime, could you suggest what caps I should try adding to see if they help?

Hello Phil,

Can you send me the PDFs of your layout? That will help me see the circuit better.

With so much voltage ripple on the battery shunt, I would suspect something to be up with the current sense resistors. I can't find them on your schematic though. Are you using kelvin sense points for the current sense resistors?

Also, make sure to route HIDRV traces close to the SW traces and route LODRV close to GND or a ground plane. SW is the return trace for HIDRV.

Phil Rittenhouse said:

could you suggest what caps I should try adding to see if they help

I recommend increasing the capacitance around the battery current sense resistor first.

Also, could you send me the waveforms that you are seeing?

Best Regards,
Ethan Galloway

Here are the top and bottom layers as PDFs.  Hopefully that's easier to see.
3phase-F_Cu.pdf3phase-B_Cu.pdf

Here is the scope capture for the voltage at the battery shunt

>  Are you using kelvin sense points for the current sense resistors?
Yes.  The shunt for the battery sense is R72 and R60 for the input

I tried adding an additional 10nF and 100nF caps in parallel with the 1uF C74 at VBATT but I didn't see any improvement.

Adding the 4.7uF, 10uF, 10nF and 100nF in parallel with the 1uF C73 on the other side of the shunt also showed no improvement

Let me know if there are other spots you think I should add more caps or if there are different values that would be helpful.

Best regards,
Phil

Hello Phil,

Thanks for the improved files. I'll review the layout and I'll get back to you next week. Thanks for being patient with this.

Best Regards,
Ethan Galloway

Thanks for all your help with this Ethan, much appreciated.

Hello Phil,

Thanks for being patient with this. Here are my suggestions for the layout:

• Route the SWx and HIDRVx traces close to each other. SW is the return trace for HIDRV.
  • Likewise, the LODRV traces should be routed close to GND. Please don't route the HIDRV traces close to the LODRV traces. This may cause the signals to leak over to the other trace.
  • I don't recommend routing LODRV2 directly underneath the inductor. You might want to move this trace to the bottom of the PCB or to the side of the inductor.
• Can you make the inductor pads bigger and attach the SW traces closer to the inductor pads? The IC uses the SW traces to measure the voltage across the inductor.
• I'm not sure if I mentioned this before, but the BTST capacitors will need to be 100nF in size.
• I see REGN / DRV_SUP are connected to a trace that goes away from the BQ25750. Where does this trace go too?
• I recommend using kelvin sense traces for the battery current resistor. This will improve the charge current accuracy.
• The IC should have capacitor on the VAC pin close by. Do you know where this capacitor is? I can't find it.
• You'll probably want to make the main power traces (like VAC, VBAT, or VSYS) significantly larger. This will reduce the resistance and inductance of the traces and reduce potential ringing in the circuit. You may want to use polygons instead of traces here as well.
• For your high power components, I wouldn't recommend thermal relief pads. Using C61 as an example, I recommend a much thicker connection to the drain of Q5 and a much thicker connection to GND.

Just for reference, here is the layout and schematic checklist for the BQ25750.

Best Regards,
Ethan Galloway

Hi Ethan,

To clarify, when you say to use kelvin sense traces, do you mean use a differential pair for the sense lines from the shunt and move the filter components closer to the SRP/SRN pins?

Hello Phil,

Phil Rittenhouse said:

do you mean use a differential pair for the sense lines from the shunt and move the filter components closer to the SRP/SRN pins?

Yes, that is correct.

Best Regards,
Ethan Galloway

Hi Ethan,

One other question:  My design is currently 2 layers with the all components on the top layer and the bottom layer largely ground.    Do I need dedicated ground planes for stable operation?

Hello Phil,

The IC can be stable with 2 layers if the IC is routed well and the bottom layer is used for primarily for GND.

You don't need 3 or 4 layers, but the extra layers acting a ground planes makes routing the IC easier. Higher layer boards also act as a heat sink and reduce the boards thermals.

Best Regards,
Ethan Galloway

Hi Ethan,

I have been testing our new board after applying your suggested changes to the layout etc.  Everything was working well - the charge current into the battery was the expected 700mA and the battery and REGN voltages were also good.  The status pins indicated it was in charging mode.  This continued without change for a little over an hour before the charging stopped and a charge fault was indicated on the status pins.  As far as I can tell there was no outside event that triggered the fault.  The temperature of all components were cool or just slightly warm to the touch.

I tried cycling the input power supply and it would start charging for a few seconds before going back to the fault state.  I noticed too that the REGN voltage was now extremely noisy (it was a fairly clean 5V when I started).

Do you have any idea what might have caused this?  Let me know if you have any suggestions for things to try.

Best regards,
Phil