DAC8775: Surge issue

Hi Davide,

We are reviewing this and will be able to respond with more information soon.

Does your circuit match the TIPD216 schematic exactly?

And to clarify, you didn't see surge damage when the device's power supply was off? Only when the device was powered?

Thanks,
Lucas

Hi Lucas,

Here the main differences:

• before reaching a connector, the output passes through a relay closed during the tests

• I used SMBJ33CA instead CDSOD323-T36SC

• I used PMEG6010CEJ instead MBRX160-TP

• Some capacitors have different footprint but same value and voltage rating

• The 10ohm and 15ohm resistors have a 0402 footprint and 63mW power instead 100mW

• I implemented a TVS diode (SMBJ100CA) in parallel with a varistor (PV95K4032R2) between analog GND and EARTH. The surge was applied with refer to EARTH

I confirm that the device was damage only with power supply ON, but we performed only few tests because we had limited modules and each test caused a permanent fault.

Furthermore, we found the first failure once the complete test had been performed (therefore after 5pulse positive and negative); then we repeated the test step by step, and the falut was detected at the first negative pulse (1kV peak).

We also tried to remove the protection diodes towards the power supplies (BAV99-7-F) but in that case the failure occurred after the second positive pulse.

If you need I can share the schematic with you with email (I try to attach in this chat but . If you find it useful we can organize a call.

I tried attaching images here but it requires me to insert a web link and if i simply past the image the resolution is bad.

Thanks for your support,

Davide

Hi Davide,

Thanks for the additional information. Lucas will review and provide some suggestions soon. Can you save your schematic as a pdf? You should be able to attach pdfs in the forum response box by dragging the file directly into the box or by using the Insert > Image/video/file option. 

Best,

Katlynne Jones

Hi Katlynne,

here attached the schematic.

cAO4.pdf

Best regards,

Davide

Davide,

Going back to one of your original questions, I'm not sure why the IEC61000-4-5 test is not mentioned. The original design of the board predates me and the other applications engineers in our group. I did look in some of our old folders about the tests and couldn't find any information on the original results.

I did look at the schematic, and in general, it looks similar to what we have in the design. Because I don't know what was damaged in the surge test, it's hard to make a recommendation on what to change. However, I did have a couple of suggestions to try.

First, I'd look at other TVS diodes for the ones that you have at the outputs. You mentioned that you have an SMBJ33CA TVS, but I'd find one that has a flatter voltage response. The breakdown voltage is 36.7V but the maximum clamping voltage is 53.3V at the maximum pulse current and that seems to me a little high. If the surge occurs, then the clamping might be too high, or not fast enough.

Another thing to try would be to replace the Schottky diodes with ones that have a lower forward voltage. The BAV99 has a max forward voltage of 0.715V at 1.0mA. For the DAC8775, it has a similar structure at the outputs that use basic diodes with a lower forward voltage. In that case, any surge or over voltage current goes into the diodes at the device pins instead of the external Schottky devices. Using lower forward voltage diodes would better protect the device. BAT54 diodes may work for this purpose and have a forward voltage of 0.25V to 0.4V. However, they do break down at a lower voltage (~30V) and the leakage current may be several uA (which may be seen as an error in current output mode).

I did notice the TVS in parallel with the varistor. I'm not sure if that helps or hurts in the surge setup. My first reaction would be that earth ground would have been a good place to shunt excess current, but with the TVS, it would likely be fine.

Joseph Wu

Hi Davide.

I have seen similar issues, and come to the conclusion that DAC8775 is quite sensitive against surge. You could consider using lower voltage TVS. While you are not using the tracking mode of the DAC, and are fixed at +/-15V you could go as low as 16V on your TVS.

The datasheet and EVM guide also describes series resistor as pass elements. Though they suggest 15ohm, we have passed with as with as 150ohm. That solution is in performance testing now.

24mA in 250ohms gives 6 volts on the load. This means you are dropping 11V from the supply rails through the DAC. This gives a dissipation of 275mW per channel. So just around 1.1W excluding buck-boost converter losses. I guess your DAC is running a little hot too? Increasing the external pass elements help reduce the power dissipated in the DAC and moving it to the pass elements. But you must consider what is the highest resistance in the load.

BR Henrik

Hi Joseph and Henrik,

thanks for your suggestions. We perform detailed test on the boards and I have some news to share with you.

Here a recap of the two boards tested for surge useful for the topic:

• Board 1
  • channel 1 tested without powering the field ➝ OK
  • channel 2 tested with field power ➝ KO (we carried out the entire test, 5 positive and 5 negative pulses)
    notes: all DAC power supplies were KO
• Board 2
  • channel 2 tested with field power and stopping the test after each pulse (500V)
    • positive pulses ➝ OK
    • negative pulses ➝ KO after first pulse
      notes: only supply of the second channel were KO but all the outputs were KO

Since we noticed the supply were damage in different ways, we try to supply the DAC with an external DC source to see if the problem were the interna buck-boost or the output circuit

• Board 1
  • with bipolar supply we see a big amount of current supplied by the DC source and all output didn't work
  • with unipolar (positive) supply we see a valid current supplied by the DC source and the output of the channels had wrong values (probably due to different dynamic) but linear behaviour
• Board 2
  • with bipolar supply only on the channel n.1 (which buck-boost works) we see a valid current supplied by the DC source and the output were OK!
    notes: Only the second channel circuit is damged.

I think the problem is related to the negative supply. From datasheet I see a negative limit of -20V but the TVS have a rated voltage of 33V. As Henrik says I think a TVS of 16-18V should be implemented.

The use of 150ohm series resistor is also a valid improvement and for our purpose a typical load of 250ohm is fine but  we would like an operative range as similar as possible to those indicated in the datasheet (625ohm) because we don't sell the board with a specific application (the company works in different fields).

Henrik, did the solution with 16V TVS and 150ohm series resistor pass the 1kV surge test as described in IEC61000-4-5? The SMBJ16CA would be ok? What is the power rate and footprint of the 150ohm resistor?

As an alternative we are thinking of using an external power supply but we would like to use the features of the device.

Thank you,

Davide

Davide,

I can't reveal too much, as this is part of a customer project. Partly from experience of previous projects we made a two-stage protection circuit, combining TVS at the connectors, limiting resistors, another TVS and then series resistors from the Vout / Iout as TI describes, but with higher values. This made us pass > +/-2kV and we are now evaluating functional performance.

We have also seen the Buck-boost converters getting fried on the tested channel. Also cases we Vout is done on all channels, but Iout may work on other channels except the tested channel.

/Henrik

Hi Davide and Henrik,

Joseph is out of office today, and responses might be a little slow next week due to the Thanksgiving holiday. Feel free to keep discussing in this thread, and Joseph will review and provide some ore comments next week. 

Best,

Katlyne Jones

Hi Joseph,

we modified the board with some of your suggestions by mixing a lower value TVS with an additional series resistor.

We will do the surge tests in January.

Thanks

Thanks Davide, 

Let us know if you have any other questions. 

Joseph Wu