Problems with DC/DC converters in isolated H-Bridge design

[stube40]

I am pulling my hair out since I keep destroying the Murata NMC1215SC DC/DC converters in my CPU controlled H-Bridge design. There is no obvious signs of damage such as heat or smell, they simply stop outputting +/- 15V. Testing out of circuit against a reference device shows they truly are dead.

The H-bridge design uses the 4x Murata device to provide an 4x isolated power rails for the 4x Fairchild IGBTs and the Vishay VO3120 opto-isolated MOSFET drivers.

I have attached a pictures that shows the general idea of the schematic, as well as a detailed schematic with every component.

The design is for switching a 150V / 40A high-voltage power source, but so far I've only been bench testing with 12V into a 27 Ohm resistor.

I suppose I must be damaging the Murata's though inadequate protection, although so far I haven't been able to catch any suspicious spikes on the oscilloscope. At $20 a pop though from Farnell, these parts aint cheap and it's proving to be an expensive detective project.

My only other thought is that I'm over-loading the output of the Muratas which are only rated to 1W.

I'm beginning to run out of ideas - can anyone suggest something to try?

 

[55pilot]

I thought you had an unlimited budget for this project, so a few $20 power supplies should not be a big deal

I could not find any specs for a NMC12158C on Murata's website or anywhere on the web, so I am unable to help you till that is sorted out. I suspect you have either mistyped the part number or put down a distributor's part number. Ideally, the correct part number should be showing up on the schematic.

The schematic lacks enough information to comment specifically. Part numbers like "Zener" and "NMADCDC" do not tell me anything.

A few general comments:

Does the power supply have a minimum current requirement? Do they require some external capacitance? The 12V input actually 12V? Are the DC-DC converters isolated? What is the input to output isolation?

Have you cut a PCB or are you doing this on a breadboard? How do you know that the circuit is exactly what you have drawn here?

How are you preventing cross conduction on the bridge?

---55p

 

[stube40]

Typo - its the NMA1215SC

The budget aint an issue, but my patience is ;-))

The DC/DCs are galvanically isolated. I've had the PCB made, so I'm pretty confident it's exactly as you see there.

If it becomes critical, I'll post the data sheets for the Zeners on Tuesday when i'm back at work. I also have a tranZorb (actually a bank of parrallel TranZorbs) across the HV+ and HV- terminals to the H-bridge - other than that the H-bridge is as you see on the schematic - no additional protection.

The Murata DCs do have a minimum current requiment, although the data sheet is a bit convoluted and I was confused as to what it was. However, I found by experiment that the 22k resistor and LED combination across +/- 15V keeps it happy.

The DC/DCs also stipulate the Inductor/Cap combination to reduce output ripple current - the values are 220uH and 1uF, but I used 10uF as that's all I had available to me when I populated the PCB.

Teh 12V input to the DC/DCs is a nice 12V rail coming from a L5973 switching reg from the main PCB - however, I'm wondering if there could be some 'beating' going on between the two??

The other 12V being used to supply the HV+ and HV- to the H-Bridge is from a lead-acid battery.

 

[55pilot]

Still no luck on the DC-DC. On the MuRata website, there are no DC-DC converters that start with NMA, or have a part number ending in 12158 or are 1W. All the parts start with MPD and start with 5W. Are you sure it is a MuRata and not a knockoff?

Going on a slight tangent.....

If you are hoping to send your schematics out to other to have them review it, you need to draw them properly. That means adding all the critical information on it. I would recommend you update the schematic to reflect the part numbers so someone looking at it can see what is going on. For all I know the problem is that you are using a 5V zener, but I have no way of knowing because all I see is "zener"

---55p

 

[55pilot]

Teh 12V input to the DC/DCs is a nice 12V rail coming from a L5973 switching reg from the main PCB - however, I'm wondering if there could be some 'beating' going on between the two??

You have to be REALLY careful when driving a DC-DC with another DC-DC. The problems generally manifest themselves as the intermediate voltage getting stuck at a low voltage like 2V because the up-stream converter goes into current fold-back protection. I can not see how it can kill the downstream converter, but it could happen.

---55p

 

[stube40]

Sure, I know the schematic is crap - I drew it in a hurry with really old and crappy design software. I'm in the process of migrating to Altium, but the project timescales didn't fit in with the associate learning the curve - will rectify all the problems for the next project..........

The NMA1215SC is Farnell part number 1021441:

http://au.farnell.com/murata-power-solutions/nma1215sc/converter-dc-dc-sil-1w-15v/dp/1021441

I can upload the part number for the Zener, or anything else, on Tuesday - from memory it is a 20V breakdown component.

 

[stube40]

For all I know the problem is that you are using a 5V zener, but I have no way of knowing because all I see is "zener"

---55p

The Zener is an On Semiconductor 20V jobbie, Farnell 1431174. Data sheet attached

 

[55pilot]

I have gotten REALLY busy due to some issues coming up out of nowhere at work and have not had a chance to post here or eat or sleep or.....

First, the really dumb observation:

Have you considered the possibility that you were delivered wrong DC-DC converters and you are installing 5V input voltage devices into 12V?

How good is your 12V? Could it be over-shooting beyond the DC-DC's maximum voltage? Have you stressed it and thoroughly tested it?

Looking through the DC-DC datasheet, here are the observations:

The DC-DC is made by MuRata Power Solutions, which is not the same as MuRata Manufacturing which is "The MuRata" that manufactures electronics components, including DC-DC converters. Hence I was unable to find the datasheet.

The modules seem appropriate for your application.

There are precautions about input voltage rise time (at the bottom of Selection Guide on first page). Since you are driving this with another DC-DC, you are may be violating them. But I can not imagine that violating them would cause the device to fail. But if the DC-DC fails to start up, what happens to the downstream circuits?

Looking through the schematic, here are the observations:

You have way too many unnecessary diodes in there. What is the point of D7, D8, D9, D10? Same for diodes on the other 4 channels?

VO3120 has a minimum operating voltage of 13.5V. What happens when the output of your DC-DC sinks below 13.5V, which it may?

I have not bothered to check the pin-out of all the parts. I am assuming that they are correct. If not, all bets are off.

Final thoughts:

How do you know that the DC-DC failure is a primary failure and not a secondary failure? Could one of the drivers or IGBTs have failed and is causing the DC-DC to fail? That would be the first place I would look.

How do you know that your layout is correct? Does your software have a robust PCB to schematic DRC?

How do you know that the PCB is manufactured correctly? Did you pay for "electrical test" of the PCB? Was it a "golden board" test or a real netlist based test? Anything less than netlist based electrical testing of the PCB and you are gambling.

How good is the layout? With all these extraneous diodes, you may have a really poor layout which is causing a lot of stray inductance in the high current traces, inducing all sorts of problems. Those are more likely to cause unintended turn-on of a IGBT, causing it to fail and taking out the DC-DC as a secondary failure.

Is R26 (and its siblings) placed next to the driver where it would reduce ringing or is it placed next to the IGBT where it would promote it?

Final final thought:

While the schematic looks reasonable, your real problem may be in the layout. You are dealing with some large currents (gate capacitance and drive current are the same even if you drive at 12V) and with devices that can generate very large dI/dT. Layout is critical and your problem MAY lie there, with the dead DC-DC being a secondary failure. Unfortunately, that is beyond the scope of what I can help you over the internet.

Good luck.

---55p

 

[stube40]

Hey 55,

Thanks for the advice, some great points in there (as usual). Here are my thoughts:

They are definitely NMA1215SC, but that wasn't a dumb thing to say (it wouldn't be the first time I'd done something like that!)

The 12V line is fairly stable, based on a L5973 which is only loaded with around 100mA (a fraction of what it can handle).

Regarding the Zeners (D7, D8, D9) and the Vishay 1.5KE22CA-E3/23 TranZorb, these are here as I'm worried about the negative line of the high voltage supply moving about during transitions between postive supply to our load (ie U12 and U16 on) and negative supply to our load (ie U7 and U13 on). Our load is effectively a large inductor of 45mH and virtually zero resistance and I'm suspecting nasty things to happen to the HV negative line that might end up blowing the DC/DC and/or the VO3120s. My philosophy has always been to go overboard with the protection in the PCB design then decide later whether I want to populate those parts. Feel free to suggest some to leave out if you think they are redundant.

Good point about the 13.5V minimum voltage - I'm not sure what will happen, although I could easily setup a quick experiment on breadboard and find out - not quite sure how to force it to dip below 13.5 though. Any suggestions?

For reference, I'm suspection the DC/DC to be a clean failure since replacing the part fixes the problem. I have no reason to suspect the IGBTs (which seem rock-solid so far).

The DRC on the PCB design is normally about 95%+ reliable, so there is room for error. The manufacturer's test is not a netlist test. Saying all that, it is a fairly simple design and so far my manual inspection has found no reasons to be concerned. The quality of the layout is a different matter, hence the stray inductance may be an issue - not sure what to suggest other than to remove some of the protection diodes.

R26 is about 2.5cm from the driver and about 10cm from the IGBT. These distances were forced due to the nature of the thick PCB tracks I had to create to get the current-carrying capacity on the PCB. I think early experiments have shown that there is ringing and it may well be causing problems. I can change the distances hence, this ringing would have to be fixed by adding other components. If you have any tips then let me know.

Thanks again for your invaluable comments.

Stuart.