AMC110 dead, second time

Hello Gerhard,
There is no specific power supply sequence requirement for the AMC1100.
I do not fully understand where the LM7705 is connected or in which output pin you see the 2.55 V after power up. Could you please include a schematic diagram? It would be interesting to see the input signal, the voltage between GND1 and GND2, and what voltages are present in VDD1, VINP and VINN when the failure occurs.
Best regards,
Jose

Hi,

the LM7705 is used to have a true Zero Output after diff/single conversion. I have to feed a UDC3138 Controller, which have single ended Inputs.

Unfortunately the error occures from time zo time, so ist hard to have Infos about the voltages in the Moment of damage.

The supply voltages supply also other circuity so if there will be a over voltage Situation, there should be other dead Chips too.

I also limited the voltage between the InP and InN with two anti parallel diodes, but I didn't Limit the voltage against InP/InN and Gnd or Vcc, maybe this is an issue?

With best regards

Gerhard

Hi,

now the third part is gone. No idea whats happen. I include a pdf of the whole Project.

This time it happens during the highest current range was used. And it seems, that it happend without a power cycle.

I use this circuit to form a isolated control Loop for a high voltage fly back converter. The Output of the converter is limited by a transil with 160V standoff voltage. So it cant be high voltage. It also cant be a Problem with the iso supply, cause the chip U4 is damaged, never another chip.

Any idea?

Thanks a lot for helping.

With best regards

Power Source FB Dev.PDF

Gerhard

Hello Gerhard,

Thank you very much for including the circuit schematic; it is helpful.

What is the maximum voltage difference between the HV-P and HVGnd nodes? Also, what is the maximum voltage difference between the HV-N and HVGnd nodes?

Is it possible that (when Rel1, Rel 2 and Rel 3 are commutating) HV-N is connected directly to VINP while the resistance between the nodes equals R12+R13+R14? If so, is it possible that:

-          VINP of U4 is taken above VDD1 by more than 0.5 V? or

-          VINP of U4 is taken below HVGnd by more than 0.5 V?

 

As a side note on your PCB design (and unrelated to the failure): the rectangular copper fill directly above U4 (where R13 is located) might be violating the creepage distance of the AMC1100 package thereby reducing the isolation rating of your design.

Best regards,

Jose

Hi Jose
there were two anti parallel diodes (D1 and D2) across the resistor chain, so Maximum voltage between HVGnd and HV-N should be +-1V. Voltage between HVGnd and HV-P can raise to 160V, but it is the current sensor which makes the Problem not the voltage sensor.
I switch a resistive load, no inductor there, so it is very unlikly to get HV-N beyond HVGnd.

I will place protection diodes at the Inputs, maybe that helps.

This board is for eval purposes only, so creepage isn't in the Focus of the design.

With best regards

Gerhard

Hello Gerhard,

Thank you very much for bringing my attention to D1 and D2.

Let me make sure that my understanding is correct on the following point:

I had assumed that there is a bidirectional current between the terminals HV-N and HVGnd. In other words, the following happens when the signals coming into J1 are first connected:

1)      The 5V Iso supply is connected to U3 and U4.

2)      Relays 1, 2 and 3 stay in the position shown in page 2 of your posted schematic (with the C terminals effectively shorting HV-N into HVGnd).

3)      For half a cycle, a current flows from pin 15 of J1 into HV-N, then it is shorted through REL3C, REL2C and REL1C into HVGnd and finally exits at pin16 of J1.

4)      For the other half cycle, a current flows from pin 16 of J1 into HVGnd, then it is shorted through REL1C, REL2C and REL3C into HV-N and finally exits at pin15 of J1.

Is my understanding correct? Is the current flow bidirectional?

Or is my understanding incorrect? (In which case the current always flows strictly out from pin 15 of J1 into HV-N into the relays or R12, R13 and R14; into HVGnd and then to pin 16 and NEVER in the opposite direction).

From your statement “…it is very unlikly to get HV-N beyond HVGnd.” it would seem to me that the current through R12, R13, R14 is only unidirectional and therefore VINP is always positive with respect to GND1. Could you please confirm which is the case: bidirectional or unidirectional?

Thank you and best regards,

Jose

Hi Jose,

the current should flow unidirectional. I used this unit as Feedback amplifier for a flyback converter. The converter works as constant current source (isolated) and the Output voltage should be limited to 140V in case of open circuit.

The current should be selected in the range between 50µA and 750mA, so I Need 3 ranges, thats done by the relays and the three shunt resistors.

The regulated current flows through a MOSFET Bridge and my test object. The Bridge is needed to Switch the current on and off very quicly and changes direction of the current through my probe. The current out of the source never changes direction (hopefully).

Hope this helps.

With best regards

Gerhard

Hello Gerhard,

Thank you for the clarification.

The only alternative explanation I can think of is if U4 had an input greater than 0.5 V while the 5V Iso node is at 0 V. This case seems unlikely since REL3C, REL2C and REL1C are initially shorting the input to U4.

At this point the next step I can think of is performing a failure analysis on the parts. Perhaps we can get some hints as to what may have caused the damage. Let me know if you would like to pursue this option.

Best regards,

Jose

Hi Jose,

I do some Software Changes now. I first Switch on the power path and than start the PWM of the flyback and I first Switch of the PWM of the flyback and than Switch off the power path. Now it runs for more than 24 h without Problems.

I further see a very high spike at the beginnung, Output of the AMC1100 is 2.5V, nominal should be 220mV

I do the same test with the 40mA range and there were the same spike and some Oszillation at the Output. As the power was < 500mW I start using a R decade instead of my artificial load and .... everything is clean. No spike, no oscillation!!

Maybe there is some Problem with my artifical load? I Thing, it works correct, more or less, but have some design flaws with dynamic signals, I guess.

I will try the 2mA range next, here the AMC1100 dies immediately. Will see, if this is solved now.

With best regards

Gerhard

Hello Gerhard,

Understood.

Let me know if I can provide any additional assistance.

Best regards,

Jose

Hi Jose,

it runs now for nearly 24h switched to the lowest range and no Problems. I used the resistor decade not my artificial load!!

Seems that it works now. Maybe the artificial load produces something which caused a reverse current?

I will get a resistor with 100W Rating and than I can do some tests with the high current range and see if there were Spikes or not.

With best regards

Gerhard