DRV8220: MCU resets when driving a solenoid latch valve

Franco,

Can you post your full schematic?

Are you using the BQ2589x as an LDO or buck converter to generate the 3.3V logic voltage? Seems like that would be better served by an LDO or a buck converter, the BQ25892 is a battery charger so it might be tripping or throwing warnings if it senses its load is not a battery.

Greetings Jacob.

Thank you for your help. Regarding the SCH, due to company policies, I can only share the parts of the SCH that are related to this issue, not all the board. So, here are the SCH of the battery charger, 3V3 power supply, 12V power supply and valve controller circuit:

-Battery charger BQ25892:

-3V3 power supply:

-12 V power supply:

-Valve controller:

I apologize because there are some parts that are in spanish (I'm from Argentina).

Regarding your 2nd question, I'm using the BQ25892 as a switching power supply and to charge the battery, so it generates 4.2 V which then powers the 3V3 Buck-Boost power supply, which then powers the 12 V Boost power supply. From what I understood, the BQ25892 can be used to power a load different from a battery, or maybe I'm mistaken. Or could be that there's some internal register to configure this case?

On the other hand, I made some more tests. What I did is put a 470 uF capacitor in parallel with the output of the battery charger, and then power the input of the battery charger with my lab power supply. In this case, when I try to switch the valve, it works. So, with this test and all the previous tests that I've made, I think that the problem is that the battery charger is not supplying enough current to the load in the transients that occur when the valve is switch.

Any suggestions?

Hi Franco,

Sorry for the delay. Please allow me some time to review the schematic and I'll provide feedback tomorrow. 

Best,

David

Greetings David.

Sure, no problem. Will wait, and thank you.

Hi Franco,

This seems more to be an issue for the battery management team. Let me transfer you to them for further suggestions.

Best,

David

Greetings David.

Understood. Then, should I close this post an re-open it in the "Power management forum"? Or is that not necessary? Let me know.

Hi Franco,

David has looped us in to this thread. Your schematic looks fine. If I understand correctly the VIN_BB Net is likely connected to the VOUT_BTY_CH Net on the BQ device schematic. This would be a fine implementation.

The BQ is able to source the required, the ISYS output current has a 5A current limit which should be able to provide current as needed. As you said the issue may lay in the transient response since it looks like the current load is applied very quickly. In this case the voltage drop may be due to the delay in loop response as current is quickly pulled out of the SYS capacitor. In these scenarios, it's typically recommended to increase SYS capacitance, VIN capacitance, or VBAT capacitance to provide a larger reserve to pull current from, without reducing voltage. Can you provide a waveform of the load pulses including VIN, SYS, and BAT?

Do you see these current pulses when powered from both VIN and Battery? Or are they not seen when battery operated?

Best Regards,

Juan Ospina

Greetings Juan.

Oh, I see, so I understand that you are saying that the problem is that the BQ isn't able to provide enough current in the transients only, and that could be solved with more capacitance, right?

Regarding the signals, I've aquired some pictures of them. And yes, the problem arises when I use the BQ alone, with no battery connected. So, I will split the signals into those where there's no battery and those where I put the battery:

NO BATTERY:

-LOAD:

-VSYS (which is basically VOUT_BTY_CH):

-VIN:

WITH BATTERY:

-VBAT:

-LOAD:

-VOUT_BTY_CH (basically VSYS):

So, as you can see, the signals are quite normal when the battery is connected. According to what you said, this could be because the battery is acting as a capacitor in the transients, providing the needed current to switch the valves.

However, I think that it would be great if the circuit worked even without the battery connected. So, for that case, what do you suggest? To add more capacitance in which places? And what values do you suggest?

Best regards,

Franco.

Hi Franco,

I didn't realize the issue was present while the Battery was not connected. If battery is not connected the output voltage of the charger is limited by the input current of the VBUS source, which as you mentioned is set to a maximum of 3.25A. This might be enough depending on the input voltage, higher input voltages allow the output to reach higher currents even given a 3.25A input current limitation. The input current limit might be reduced by protection measures such as VINDPM. Do you have a capture of VBUS in a no-battery setup when the load is applied?

Best Regards,

Juan Ospina

Greetings Juan.

Oh, I see, I didn't know that if you supplied more input voltage, the current limit incremented. In my case, I'm powering the circuit using a lab power supply, set to 5V. And I would say that yes, the current consumed by the valve should be way lower than 3.25 A.

Regarding the VBUS signal, yes, it's on my previous comment, the image in the "NO BATTERY" section named "VIN". Just in case, here it is:

Best regards,

Franco.

Hi Franco,

Sorry I must have missed that. It looks like the load seems to be too much for your input source based on the voltage dips. The voltage of VIN seems to be dropping to levels in which it would trigger VINDPM or sleep, the device wouldn't be able to pull more current from the input once VIN has dropped a significant amount. If your input source should be able up to 3.25A of output current then it might transient response limitations on the part of your input source This can be mitigated by increasing the VBUS capacitor. If your lab power supply has a current limitation below 3.25A, you can try increasing this setting, or using a more powerful input source if it is maxed out. This should confirm that is the cause of the behavior we're seeing.

Best Regards,

Juan Ospina

Greetings Juan.

Oh, I see. Well, my lab power supply can provide up to 5.1 A, and it has an overcurrent protection that is configurable. So I just changed that value to the maximum of 5.1 A, and tried to switch the valve, but it still resets the ESP32. You think the transient current is going higher than 5.1 A? 

Regarding the VBUS capacitors, do you have a value suggestion? Maybe I can increment the output capacitors too. I can add that I've tried with a 470 uF capacitor in several places of the circuit, just as the output of the battery charger, output of the 3V3 power supply and input of the 12 V power supply, and in that case it worked. I chose that high value because I've seen some latch valve designs that use a 4700 uF capacitor, which seemed a lot to me. But in my previous post, Jacob suggested that that high value capacitor is only used if the power supply can't provide enough current, and if I sized the power supply correctly, it wouldn't be necessary.

Best regards,

Franco.

Hi Franco,

I see, that does seem like sufficient power so its unexpected that VIN would drop so significantly. It seems like there are two possible causes:

1. Resistance in the line between the supply and the VIN pin could cause the voltage to drop as current increases. Decreasing the resistive load between the two might improve performance.

2. As mentioned this could be related to transient behavior in which case capacitance does help on any point along the power-path or throughout the power-path. Unfortunately I can't recommend a particular value as I'm not really familiar with the load devices. One thing to keep in mind is capacitance derating at higher voltages, its important to keep that in mind when selecting your components.

Best Regards,

Juan Ospina

Greetings Juan.

Regarding the first point, I would say that, although obviously there's resistance between the input in the board and the input pin of the BQ, it's in the order of a few mOhms, because the distance is around 2 cm. So I think that wouldn't cause the voltage drop of hundred of mV that I'm seeing.

Regarding the second point, I see. I think that, given that the switching of the valves is what generate the need of current in the transient, maybe the capacitance should be placed in the 12 V power supply, which powers the valves. But I'm still confused as to why if I sized all the power supplies correctly, according to the current that the valves seem to consume in the transient when they switch, there's still a voltage drop. Because, compared to the previous version of this board, I also added some extra capacitance, for example, around 60 uF next to every valve. But I don't know, maybe the peak current is higher than I think, and that's why in the designs that I saw in the internet they put that large capacitor of 4700 uF. The problem is that that capacitor is huge in size, so that's why I wanted to avoid it.

Best regards,

Franco.

Greetings Juan.

Sorry, I'm posting again because I've made some test, and seem to have found the final solution. Sadly, I can't mark this reply as "This resolved my issue", but well, hope it helps anyway.

First, I bought a 2200 and a 4700 uF capacitor (35 V both). Then, I soldered them in different places of the power line of my board, and testes the results when switching the valve. I was hoping that, with such a huge capacitor of 4700 uF set in the 12 V line, the problem would be fixed, because it could provide the current needed by the valve in the transient. But, although in some cases the ESP32 didn't reset, the truth is that, watching for example the 3V3 line, I was still seeing a considerable voltage drop, of about 1 V. Here's an example:

In this case, the battery was connected, and I was powering the board with my lab power supply, with 5V.

Then, I thought: "well, maybe the real solution would be to mitigate the peak current in the valve, which is the real cause of this problem". So I thought I could add a resistor in series with the valve. I first tried with a 100 Ohm resistor, but with that value, the valve couldn't even get enough current to switch correctly. So then I tried different resistance values, with the help of a potentiometer, and, for example, with a value of 25 Ohm, the valve seemed to switch correctly, and the voltage drop nearly dissapeared. Here you can see the 3V3 line:

And here the 12 V line:

So, in conclusion, I think I will just add a series resistor to each valve. I'm still curious about why the huge capacitor didn't work as intended though.

Best regards,

Franco.