We are running into an L99MC6 caution (lowest caution on page 12 of Doc ID 16523 Rev 1) where our +13.8V supply is bleeding back through relay coils to source outputs and elevating the 3.3V Vcc supply before the charge pump seems to ramp up). Workaro

[DDeye.1]

Our circuit uses an L99MC6 6-ch driver in 3-ch bridged mode to drive 3 relays (Song Chuan p/n 301-1A-C-R1-U03-12VDC) with a common +12V rail connected to the relays. The relays are activated by a sink output. 

The circuit's 3.3V logic Vcc is temporarily raised to 7.5V and limited to this voltage through the resistance of the relay coils. The problem is that it subjects the 3.3V microcontroller logic to 7.5V through half a dozen spikes that elevate 3.3V to 7.5V with a 10% duty cycle.

Isolating the L99MC6 Vcc protects the 3.3V microcontroller rail but it subjects the driver IC Vcc to voltages > 5.5V (absolute maximum). We didn't see any mention of needing to sequence +Vs after Vcc while the charge pump is charging. No typical charge time was found on the data sheet either.  

I should also mention we are stepping down the vehicle +13.8V to 12Vdc through a switching supply and further down to 3.3V using an LDO. The stack up time of the buck + linear regulators plus any charge pump start up time is enough to cause issues.  

1. Do we need to sequence +Vs after the L99MC6 Vcc is stable?
2. Can we leave our circuit connected as-is where +Vs(13.8V) is connected to a relay coil+ and coil- connected to a source output?  Is this harmful to the L99MC6 elevating only its Vcc above 5.5V for a brief time immediately after power up?  

The relay coils are subjected to some decreased supply voltage while the L99MC6 portion of the circuit stabilizes (up to 5 seconds). Possibly 13.8V - 7.5V =~ 6.3V but without enough energy to toggle the relays.  

We can put a zener clamp on the L99MC6 Vcc but feel it would stress parasitic diode junctions on the die even more that isolating the L99MC6 Vcc from the rest of the 3.3V system through a low Vd Schottky to supply 2mA to the driver IC.

[SABBI.1]

​Good morning,

L99MC6 is supposed to be power up with proper sequence. When high side driver is used, drain of channel 1 I used internally as power supply (battery). If high side is not used, DRN1 can have any voltages BUT charge pump must be disabled. Internal selector for power supply use VCC when only low side drivers are used and it uses DRN1 when at least one high side is used. Selector for power supply may create a path from DRN1 and VCC and caution you mention is saying that.

So,

when at least one high side is used, sequence must be the following: Apply

battery on DRN1, apply VCC that is even doing power ON reset so L99MC6 will be

in standby and charge pump is configured active, then best practice is to

confirm charge pump configuration, then enable L99MC6.

VCC present

before battery may enable L99MC6 but without voltage on power supply.

It is

not clear to me schematic you are using. I think that only high side is enough

to drive relay but you did use half bridge configuration, so there is something

more I don't know. Then there is a way to supply limited current from relay

coil but I think that it should be grounded.

If you

may add schematic I may understand path of current.

Any

way it is forbidden to rise VCC higher than 5.5V without time limitation. You

are right, path of current is abnormal and from what I understood limited only

by external components.

I think we have solution to drive properly relay without additional external components, please show me schematic around L99MC6, mainly how relays are connected.

Best Regards

  Salvo

[DDeye.1]

Hi Salvo,

I carefully read your response and will provide answers to your questions. Forgive the long response. There is a lot to discuss.

L99MC6 is supposed to be power up with proper sequence. When high side driver is used (all outputs are used in bridge mode so the high side driver may be used in a typical system. In any case the drains DRN1, DRN2 and DRN3 are always connected to system V+ through S3M schottky diodes for reverse battery protection), drain of channel 1 I used internally as power supply (battery) True. DRN1 is connected to approximately 11.5-13.3V as power is first applied to to +Vin.

If high side is not used (not the case in our design so not relevant), DRN1 can have any voltages BUT charge pump must be disabled (also not relevant because we are using a high side configuration). Internal selector for power supply use VCC when only low side drivers are used and it uses DRN1 when at least one high side is used (yes, at least one high side is used so the internal power selector should be using DRN1 as the charge pump supply voltage). Selector for power supply may create a path from DRN1 and VCC and caution you mention is saying that.

So,

when at least one high side is used, sequence must be the following: Apply

battery on DRN1 (this is the case in our circuit), apply VCC (Vcc will ramp up after +Vin is applied and after the LM46002PWP start up time. The Vcc supply ramps up with soft start capacitor C503 0.1uF) that is even doing power ON reset so L99MC6 will be

in standby and charge pump is configured active, then best practice is to

confirm charge pump configuration (I don't think we do this. I will check our firmware and add a charge pump wait loop) , then enable L99MC6.

VCC present before battery may enable L99MC6 but without voltage on power supply. Please clarify "without voltage on power supply." Internal L99MC6 power supply?

It is not clear to me schematic you are using. I think that only high side is enough to drive relay but you did use half bridge configuration, so there is something more I don't know. Then there is a way to supply limited current from relay coil but I think that it should be grounded.

If you may add schematic I may understand path of current. See attached.

Any way it is forbidden to rise VCC higher than 5.5V without time limitation (this cycle only lasts a few seconds but 3.3V Vcc swings wildly up to 7.5V. Not good for the L99MC6 or any of our 3.3V microcontroller logic) . You are right, path of current is abnormal and from what I understood limited only by external components. I think we have solution to drive properly relay without additional external components, please show me schematic around L99MC6, mainly how relays are connected.

Thank you,

Dennis

[DDeye.1]

Hi Salvo,

Attached are scope images of the L99MC6 DRN1 voltage rail, 6.0V switching regulator and the 3.3V microcontroller rail. The first scope capture is a normal power up sequence with open half-H outputs. Nothing connected to the outputs. This shows the L99MC6 DRN1 voltage comes up first and reaches 10.5V as the 3.3V regulator starts to output voltage. 3.3V comes up 60ms after the DRN1 supply.

The 2nd capture shows an 18 second long sequence that damaged the microcontroller on the board. In this setup all 12 relays were connected to the 12 half-H outputs and the relays were connected to the +Vin rail.

The 3rd capture shows a brief Vcc voltage fault ~ 2 seconds long. The microcontroller is not active at this point.

The 4th capture is a 10x view of the 8V spikes on the 3.3V supply and 7.8V steady state period. This entire event lasts about as long as capture #3 (about 2 seconds).

If needed I can send a circuit board and relay load setup to you. My next steps are to reduce the number of relays connected to the +Vin supply and look at the changes in the waveforms. I am wondering if the oscillations have something to do with the high impedance DRN1 supply voltage provided by a reverse polarity diode. The only load on the DRN1 supplies are a pair of 680uF capacitors. If DRN1 had a much lower impedance supply like a vehicle battery or 20A 12V supply it might change these oscillations.

Nothing should elevate the 3.3V rail unless something is feeding this node other than the 3.3V LDO.

I look forward to your reply.

Thank you,

Dennis

[SABBI.1]

Thank you for sharing schematic.

I see that your solution requires relay directly connected to +Vin. This is unusual for safety but for sure allowed. Relay connected to +Vin do not requires half bridge configuration on L99MC6. Low side configuration for all relay is, for my understanding of your application requirements, the proper solution. Even diodes from GND to RLYOUTx will be redundant. Diode from RLYOUTx just one help but L99MC6 can manage without diodes.

Coming back to the issue. Battery for L99MC6 (DRN1) is protected by diode S3M. This node has also big capacitor (680µF). During power up (L99MC6 in standby) and when half bridge is OFF (I assume you use half bridge register configuration), DRN1 is supplied by relay connected +Vin through SRC1 because at higher voltage. This is short to battery condition. Short to battery is on all SRCx when half bridge is OFF this is because of drop on S3M versus simple coil resistance on relay. High side drivers are never really managing current.

Pictures are showing that DRN is rising (3rd picture) with steps but microcontroller is not active, so steps on DRN1 are not defined by loads and even pulses should not be induced by driving event, may be something from switching converter. It is difficult that is created by abnormal biasing path due to long timing of repetition.

What is really abnormal I voltages on VCC and BUS on 3rd picture at the beginning of sequence: Voltages are less than 2V, flat, while +Vin is at least 12V. On first picture, without relays, this is not present. It is present again on last picture but, the most important, with +Vin at zero. Where is coming this bias?

I think something is missing and investigation point should be on DRN1 waveform shape.

Let me suggest fast trial to prove hypothesis of reverse bias from SRC1. Using same hardware and software configuration, disconnect SRC1, SRC2 and SRC3 from all L99MC6. In this condition you should have full functionality. If it is confirmed then we may set better configuration.

Best Regards

Salvo

[DDeye.1]

ok. In separate cases I will remove the S3M, I can monitor DRN1's current during this event, and temporarily disconnect SRC1,2,3 from all L99MC6s. I will get back to you once I try these tests.

Thank you for your answer.

Dennis