ST1S14PHR failure

[Roberto Giovinetti]

Hello everyone

We use ST1S14PHR to reduce from 40V input to 15V output. 15V are used to drive Mosfet (BLDC Motor Control). Lately we had a lot of problem with ST1S14PHR, because some parts failed during on field operation. We cannot understand what is the problem, because when parts came to us, we can only see ST1S14PHR is damaged, with pin 7 (Vin) burned, and we don't know what's happened.

40V Vin cames from a battery. We use a electolitic capacitor and a pair of ceramic capacitor in the input stage, as the image below can show.

 

We tested products on field for some months before say it was ok and before start production. After we start production we get negatives feedbacks, and discovered the problem on ST1S14.

Battery in the input should avoid high Vin transient, so we cannot understand why pin 7 burn.

Anyone has expenrienced something like that?

Any ideas?

Thank you

Best Regards

Roberto

[Peter BENSCH]

Hi, Roberto,

the ST1S14 will be damaged when it gets an input voltage higher than the value given in the maximimum ratings (52V).

Currently you are trying to protect the ST1S14 by D1, but have selected the wrong device with the SMBJ48A, as this TVS only limits at 48V at a very low current. The voltage on D1 is much higher when an incoming spike pushes high energy into Vin+. Depending on the applicable pulse test standards, you can easily get a clamping voltage of 77.4 V (10/1000µs pulse), which of course blows your ST1S14.

Instead you might consider a SMBJ30A, which provides a VCL=48.4V for the 10/1000µs pulse, but is having a nominal VRM=30V, which is probably to low.

If you cannot limit the energy of incoming spikes using other techniques, consider another buck regulator like the L7987 that can withstand higher input voltages.

When your question is answered, please close this topic by choosing Select as Best.

/Peter

[Roberto Giovinetti]

Hi Peter

Thank for your answer

I know the intrinsic limit of ST1S14 of 52V. I know the danger a spike can cause to input pin, but our application, as power supply, use a large battery, and so spikes shouldn't move Vin value: battery should absorbe motor energy when an overcurrent phenomenon turn off bridge.

Effectively SMBJ48A have an high clamp value, but battery can goes high to 41V after a complete charge session, so we can choose, at least, SMBJ40A with VCL=64.5V instead 77.4V, that is a better value but not safe for ST1S14.

The problem is that we have some parts mounted (about 500) and we need to find a way to protect this devices that are on the field: maybe we should add a suppressor on the wires that cames from motor or from battery...

Any suggestion?

Roberto

[Peter BENSCH]

Hi, Roberto,

well, having a large battery between the generator and your device does not necessarily prevent from destroying your device, as many influences are possible as to why a spike can bypass the battery: the battery impedance, the layout/structure of cables and wiring etc. The main problem remains the same: the voltage difference between the possible VCL and the maximum input voltage is too small, so that TVS with their better, but still quite steep characteristic curve cannot really protect.

I only see one (very small and maybe expansive) chance you could try to still use your existing parts: add one or two low pass PI filters between the generator and your power supply and maybe add an SMBJ40A in parallel to every filter capacitor. This PI filter consists of an input capacitor, a power inductor (with a saturation current of at least the input current of the supply) and an output capacitor. Of course, you can combine two caps in case of two consecutive filters. The values of the inductor and the caps depend on your expected pulse form and frequency, but I guess it'll be in the range of Henry and Millfarad = bulky.

To avoid this, it is be better to use a buck regulator with suitable input voltage as mentioned before.

Sorry not having a more pleasant answer.

When your question is answered, please close this topic by choosing Select as Best.

Good luck!

/Peter

[Peter BENSCH]

Hi, Roberto,

after some internal discussion I realized that I misunderstoud your setup.

We're pretty sure your problem is caused by the BLDC, by feeding energy backwards into the ST1S14 while braking/decelerating. That energy has to flow away/burned somewhere and will likely charge the ST1S14's input capacitance by reverse current, raising the voltage to dangerous levels.

To protect the ST1S14 you could block the energy from the motor using e.g. a diode/Schottky from the 15V net to the BLDC drive and additionally burn the energy with an SMBJ15A (or SMCJ15A, depending on motor size) from the bridge voltage to ground. You could also slightly increase the 15V by increasing R11 depending on the forward voltage of the inserted diode. It might also be helpful to add some 100µF or even more over the bridge voltage, i.e. after the blocking diode and in parallel to the SMBJ15A/SMCJ15A.

Additionally you can add an SMBJ40A as close as possible to the input (pin 7) of the ST1S14 to avoid voltage stress.

/Peter

[Roberto Giovinetti]

Hello Peter!

Thanks for your useful reply.

So, if i'm not wrong, you suggest to protect 15V net, that can cause the injection of energy through (I think) L1 and C6 from above image, and to do it you suggest to use schottky diode and a TVS. So power stage could be as in the following image, in which 15V for the drivers are obtained after blocking diode.

Unfortunately I don't have large space on the PCB to insert big capacitance (100uF 35V) but I've to try to insert it.

A fact for me not so clear is: do you think energy came into 15V net from the drivers L6398? I've never seen protection components on driver power supply in another examples... 15V net is not "connected" to BLDC voltages...

Maybe could be that I've not understand in the right way your suggestion.

Roberto

[Peter BENSCH]

Before that, I was assuming you are powering the BLDC voltage, i.e. the power bridge with 15V. This (very well known) schematic doesn't explain the burning of the ST1S14, as the 15V just powers the gate drivers, which should not feeding that much energy back.

Where is Vin+ supplied from?

In the original schematic you will also find a TVS (SMBJ48A) at Vin+.

/Peter

[Roberto Giovinetti]

If you check both images I've posted (power supply schematic the first, power stage the second), you can see Vin+ is the same input for ST1S14. Vin+ cames from battery. Near ST1S14 there's TVS and large capacitor (220uF 100V). Battery is 40V when completely charged.

Circuit board is very small (40mm x 87mm) double side components, and high current tracks are separated from low current traks, and ground for ST1S14, bridge and logic, meet in a star connection on input electrolitic capacitor.

I've notice the burning of ST1S14 , as I told you, sometimes (not often, but sometimes) when there's an over current event and the bridge is powered off by the microcontroller. Unfortunately I could not get screenshots via oscilloscope during this events on +VIN pin of ST1S14.

Overcurrent si set to 30A.

Roberto

[Peter BENSCH · Accepted solution]

OK, then we are indeed back to my first assumption, as the motor is switched off, decelerates and pumps a lot of energy into Vin+. That means (I repeat):

if you cannot limit the energy of incoming spikes using other techniques, consider another buck regulator like the L7987 that can withstand higher input voltages.

For the systems in the field it might be worth to think about replacing D1 (SMBJ48A/SMBJ40A) by an SMCJ40A and to add an SMBJ40A as close as possible to pin 7 of ST1S14 - but such a big motor delivers likely too much energy to burn it using TVS.

/Peter

[Roberto Giovinetti]

Thanks Peter!

The exchange of information and reasoning is always stimulating and allows to deal with problems from different points of view ;-)!

In the future, maybe, I will reply to this post if I'll get some new informations.

Roberto