BTS428L2 fails after two years of use.

[Anonymous]

Hello,

We have a product that we ship worldwide that uses the BTS428L2ATMA1 as a digital output switch to switch 24 Volts.

An optocoupler (the VO615A-8X017T) puts a switching DC signal of 0V and 2.651V on the enable pin of the BTS428L2.

When the signal is 2.651V on the enable pin, the output of the BTS428L2 is 24V. When the signal is 0V, the output is 9.6V. I don't quite understand how the output is not 0V when the enable pin is pulled to GND, but that's a sidenote.

After about two years, we received multiple products back where this IC failed, in a way that I can't seem to reproduce.

Pin 2, the input pin, is "shorted" and gives out 24V. The measured resistance between 24V (Tab) and pin 2 is 2.66kOhm. On a working one, the resistance is 8.5kOhm. The IC does not heat up much, the package temperature stays under 30C.

The rest of the resistances between pins is the same.

On the output, an SMC SY5140-5LOU-Q is attached. Note that there's no additional protection on the output pin of the BTS428L2. The SMC part has a varistor build in from what I can remember looking at the datasheet.

What could be the cause that this component fails?

Is there a possibility to ship over the PCB with the broken component for Infineon to diagnose?

[Meghana]

Hello,

Thank you for posting on Infineon Developer Community. 

Can you please share the schematic of your product and Vgs , Vds , & Id measurements through the switch on a working board?

Regards

Meghana R

[Anonymous]

From left to right, input from the MCU to the digital output on a connector.

VCC2 is 24V (which is attached to the Vds pin of hte BTS428L2)

The output of the optocoupler (pin 3 of OK5, which is attached to the Vgs pin of the BTS428L2) is either 0 or ~2.651V (normal operation).

The source, Vss, (no load connected) switches between 9V and 24V (again, normal operation).

On the failed BTS428L2 ICs, the source, Vss, is either stuck at ~17V or 24V (not switching between the voltages, it's what I measured at two independent ICs).

I did some extra measurements and checked the following:
R87 is ~2.2kOhm when measured (on both working and failed parts), which is good.
VCC2 to In (pin 2 of the BTS428L2) is sometimes ~2.5kOhm and sometimes only 15 Ohm. The resistance is correct when it's about 8kOhm.

I believe that the parts that have the 15 Ohm resistance cause the output to be at 24V and the parts that have ~2.5kOhm resistance to have ~17V on the output. I will have to confirm this.

[Meghana]

Hello @Anonymous ,

Do you have any working board? Can you please share us the below measurements on BTS428L2 from a working board both under no load and loaded conditions 

1. IN pin to GND

2. OUT pin to GND

3. Vbb to OUT pin

Regards

Meghana R

[Anonymous]

Here are my measurements between a working and two broken BTS428L2 ICs. The applied load is a 4k7 resistor between OUT and GND.

I. Working BTS428L2 no load

1. 6.5kΩ to 6.6kΩ
2. 2.8MΩ
3. Open  circuit

I. Working BTS428L2 load

1. 6.5kΩ to 6.6kΩ
2. 4.2kΩ
3. 5.5kΩ

II. Broken BTS428L2 no load (BTS428L2 is stuck at 24V out without load)

1. 895Ω
2. 15.3kΩ
3. 14.3kΩ

II. Broken BTS428L2 load

1. 854Ω
2. 3.54kΩ
3. 3.9kΩ

III. Broken BTS428L2 no load (BTS428L2 is stuck at 17V out without load)

1. 904Ω
2. 2.85kΩ
3. 2.85kΩ

III. Broken BTS428L2 load

1. 864Ω
2. 4.168kΩ
3. 4.286kΩ

[Meghana]

Hello @Anonymous ,

By measurements, i meant the voltage measurements. Sorry, it was not clearly mentioned in my previous response.

BTS428L2 is a smart switch that comes with integrated protection against over current, over voltage, and over temperature. Hence, possible reason for the damage could be the prolonged stress on the device due to high-frequency overshoots. To find out, we need the voltage measurements across the below terminals from a working board.

This is needed to also evaluate, why the output is 9V when the potential at IN pin is 0V, as 9V at the output indicates that the switch is not off but would be operating in linear region. 

1. IN pin to GND

2. OUT pin to GND

3. Vbb to OUT pin

Also, we need to confirm if the damage is on the BTS428L2 or its driving circuit. To find this out, can you please depopulate the damaged IC and measure the input and output voltages of the opto coupler. 

Regards

Meghana R

[Anonymous]

Note that for the 3rd measurement (Vbb to OUT pin) I placed the black probe on the Vbb and the red probe on OUT, hence why the voltage is negative. As there are 4 broken in total with all different results, I added the designator that corresponds to the schematic/PCB files I sent you.

I. Working BTS428L2 no load - IC1

1. 0V or 2.65V
2. 9.65V or 24V
3. -13.93V or 0V

I. Working BTS428L2 load - IC1

1. 0V or 2.65V
2. 0V or 24V
3. -24V or 0V

II. Broken BTS428L2 no load (BTS428L2 is stuck at 24V out without load) - IC6

1. 23.94V
2. 23.92V
3. -0.063V

II. Broken BTS428L2 load - IC6

1. 23.94V
2. 22.36V
3. -1.656V

III. Broken BTS428L2 no load (BTS428L2 is stuck at 17V out without load) - IC5

1. 23.73V
2. 16.26V
3. -0.016V

III. Broken BTS428L2 load - IC5

1. 23.72V
2. 0.735V
3. -23.28V

IV. Broken BTS428L2 no load (BTS428L2 is stuck at 17V out without load) - IC7

1. 6.054V or 6.104V
2. 16.3V
3. -0.022V

IV. Broken BTS428L2 load - IC7

1. 6.058V or 6.108V
2. 0.764V
3. -23.25V

V. Broken BTS428L2 no load (BTS428L2 is stuck at 17V out without load) - IC4

1. 5.952V or 5.999V
2. 17.77V
3. -6.244V

V. Broken BTS428L2 load - IC4

1. 5.945V or 5.994V
2. 8.9V
3. -15.12V

I will post the measurements of the optocouplers without the broken ICs later.

[Meghana]

Hello @Anonymous ,

If you are measuring it with multimeter, we wont be able to see the high frequency components. Can you please measure using oscilloscope and share the waveforms. 

Regards

Meghana

[Anonymous]

I will measure with an oscilloscope, unfortunately I do not have a high-end unit, it's a Teledyne WaveAce 2014 I hope this will be sufficient.

I just received another broken unit which has different results. It again is stuck at 17V without load, but the voltage at the enable pin (1) is correct, 0V or 2.65V. With load, the output has 9.5V between OUT and GND and -14.5V between Vbb and OUT.

[Anonymous]

Here are two measurements of IC4. In blue, you can see the output of the optocoupler and in yellow you can see the output of the BTS428L2.

I decreased the time step to 2.5ns to capture any high frequencies, but it's all flat.

I was able to generate a screenshot with a bigger time step.

Currently, the optocoupler is turned on and off every two seconds. I will redo this with the output fixed to high (for easier measuring).

Please tell me what settings you'd like me to use (such as timestep, etc), so that I can create better screenshots.

Edit: I modified the software for the controller so that the outputs are fixed to a high state (so that it's easier to take measurements with the oscilloscope). I did a check and the outputs are very similar to the attachments in this reply.

[Anonymous]

I removed IC4 and here are the measurements of the IC itself:

• Vbb to GND: 15.5kΩ
• Vbb to In: 3.69kΩ
• Vbb to ST: 5.6MΩ
• Vbb to Out: Open
• Out to to GND: Open
• In to GND: 19kΩ

On the PCB itself with IC4 removed I measure 11.7kΩ between Vbb and IN.

See the image below where I measured the output of the optocoupler when the signal is high (with IC4 removed from the PCB).

[Meghana]

Hello @Anonymous ,

I summarized all the details shared so far. But the observation is that there is no consistency even in the failed module measurements. For example, in the below cases, the measured values do not meet the KVL requirements. 

Measured between	Damaged board (IC populated, output -17V, with load)			Damaged board (IC populated, output -17V, without load)
	Voltage ( IC7)	Voltage ( IC5)	Voltage ( IC4)	Voltage ( IC7)	Voltage ( IC5)	Voltage ( IC4)
1-4 (IN - GND)	6.054V or 6.104V	23.72V	5.945V	6.05V	23.73V	5.95V
2-4 (OUT-GND)	0.764V	0.735V	8.9V	16.3V	16.26V	17.77V
3-2 (Vbb - OUT)	23.25V	23.28V	15.12V	0.022V	0.016V	6.244V

In IC7 (without load), if OUT- GND is measuring ~17V, then Vbb-OUT should be around ~7V as Vbb to GND is 24V. But the measured value is different. Also after applying load, the voltage has dropped to 0.764V, which is not reasonable. 

From the IC4 impedance measurements shared in your previous reply shows that there is no permanent short across Drain to source or across any pins on the device. 

Hence i suspect the failure you are observing is not just the device failure, but might be a composite result of different causes. To evaluate this further, we need below details . 

1. What are the loads applied on these switches? Is the load characteristics similar for all the switches? 

2. What is the type of drive signal applied to these loads? Is it a continuous signal or a PWM? If it is a PWM, please share the frequency and potential of it. 

3. If the drive signal is PWM signal, then please share the voltage measurements across below points using an oscilloscope. 

IN - GND

OUT-GND

VBB - GND

4. Do you have any measurements done during the initial product testing? I mean the measurements done before shipping these to your customer. Please share if available. 

5. How is the board powered up? What is the source of 24V?  Kindly provide system architecture information if available. 

Regards

Meghana 

[Anonymous]

1. What are the loads applied on these switches? Is the load characteristics similar for all the switches? 

As far as I know, the following items are attached (but I don't know on which ICs):

- SMC SY5140-5LOU-Q

- SMC SY5120-5LOU-C4F-Q

- W0215000101

- SMC CD85N16-50-B

2. What is the type of drive signal applied to these loads? Is it a continuous signal or a PWM? If it is a PWM, please share the frequency and potential of it. 

It's a continuous signal. PWM is not the case.

4. Do you have any measurements done during the initial product testing? I mean the measurements done before shipping these to your customer. Please share if available. 

This product was developed by an engineer who has left the company years ago. Any initial testing that has been done was with done limited tools (most likely just an oscilloscope and a multimeter).

Regarding product testing, we have a test machine which tests all connections. The PCB stack is tested for 2000 seals (in the actual packaging machine) to make sure that the hardware operates as it should.

5. How is the board powered up? What is the source of 24V?  Kindly provide system architecture information if available. 

The source of the main board is 230VAC or 115VAC. This is transformed to multiple voltage sources (+12V, +5V, +3V3 and -8V). The microcontroller (which controls the optocouplers) uses the 3.3V power supply.

The Digital Out section uses an external 100-240VAC to 24VDC module from MURR ELEKTRONIK with part number MCS-B 2,5 100-240VAC/24VDC

[Meghana]

Hello @Anonymous ,

As we discussed over call, kindly share us the below measurements. 

• Measurements of IN and OUT for both high and low conditions on both working and damaged circuits.
• Depopulate the two damaged IC's mentioned below. Power these IC's externally and measure the waveforms of IN and OUT for both high and low conditions.
  • IC in which IN pin is shorted to 24V and OUT is stuck at 24V/17V
  • IC in which IN pin potential is ok, but OUT is still stuck to 24V/17V.
•   Measure value of the resistor connected between Pin 4 of optocoupler to Vcc/2 in all circuits, both working and damaged circuits.  

Regards

Meghana R

[Anonymous]

Broken IC (IC4) depopulated without any load

Off condition

• Measured OUT: 17.78V
• Measured IN pin: 24V

On condition (using a power supply to put 3.3V on the input pin that's stuck at 24V)

• Measured OUT: 17.78V
• Measured IN pin: 4.76V 

Potential good IN pin (IC7), same testing conditions

Off condition:

• Measured OUT: 16.31V
• Measured IN pin: 24V

On condition:

• Measured OUT: 16.31V
• Measured IN pin: 4.88V

The resistor (R88) measured within spec, the same is for all circuits on multiple products (PCBs).

I believe that all ICs broke in the same way, it's just that some driving circuits, although theoretically the same, can pull the input pin down harder yielding a measured ~6V.

[Meghana]

Hello @Anonymous ,

Failure cause is not clear from the circuit measurements.

1. When IN pin measures 24V while OFF, we can suspect an internal short between IN and Vbb pin. But its not justifying, when IN pin measures 4.8V after 3.3V is applied externally. Since power supply is directly connected to IN pin, it should have measured 3.3V and not 4.8V.

2. In both ON and OFF conditions, IN potential will read as 'HIGH' state by the logic. In such case, the output should have been 24V, but its 17V. We need to investigate if the failure is in output stage also or not. 

To investigate these points, we would need these boards to be sent to us. We will move this query to case management system for further actions. Please confirm once you receive a notification regarding the same to your email.

Regards

Meghana R

[Anonymous]

Hello Meghana,

Thank you for the help so far. I did not receive an email yet, I'll be patiently waiting.

In the meantime, I'll check internally what the following steps will be (if we want to send over the PCBs, etc.).