Half bridge driver IC: What is this failure?

Thread Starter

Chipbox

Joined Aug 13, 2018
16
Background
I had a failure on a half bridge driver IC (IRS2304) between pins 7 (HO) and 8 (VB). The driver IC is now fried. I know this because after desoldering, pins 7 and 8 were shorted together. On the PCB however, the impedance on the footprint is still very high, so there wasn’t a complete breakdown in the insulation. You can see some odd markings where the failure might have occurred between these pins. This same fault also damaged my +5V bus but that’s a different story.

Question
What is the name of this fault that probably caused this brown stuff between these two pins? It appears like a burn mark. Has anyone else seen this? Does this particular failure mode have a name?
 

Attachments

Last edited:

AlbertHall

Joined Jun 4, 2014
12,345
It really doesn't look like a burn mark. It has straight sharp edges whereas burn marks tend to blur out at the edges.
I suspect it is some smd glue that has escaped and that would be nothing to worry about.
 

Thread Starter

Chipbox

Joined Aug 13, 2018
16
@AlbertHall Hmmm, perhaps but I'm not confident. I soldered the chips on this board and others, and I haven't seen anything similar before. It also corresponded to the pins that shorted. This is why I find it intriguing.
 

IamJatinah

Joined Oct 22, 2014
136
Hi All, I do know what this is, and lets discuss this.

This is "Staking compound". This has been around for decades, starts a light yellow and turns into a dark brittle brown after years of heating or air exposure.

Get this now....that compound intended to "hold" or stake parts, traps moistures, which form small beads under the brown, which dries on top and continues to form beads underneath until the compound and moisture begins to conduct as a high resistance path. High potentials harder as imagined. It's been an issue for years, was distributed by Delta among others, and is still in use today. Gently brush it from parts, careful not to damage parts or short charged planes, but it chips off with some care, you will find the moisture under there, and at times can even see carbon-trails formed by the beads that broke-over as a current-path. That's what it is :)
 

IamJatinah

Joined Oct 22, 2014
136
I don't doubt that you believe what you say. But it doesn't line up with what experts in the field have to say.
http://www.circuitnet.com/experts/88019.html
I read the article, they have a different intrusion going on, they have moisture laden with enzymes that will eat/etch the traces, metal legs, and pads. What you have looks to be the brown potting. One way to know, touch it with a hot iron, staking will crumble apart, oxides will prohibit heating and the joint wont even wet or flow until cleaned off. Either way is should be cleaned off carefully. The picture shows the pins corroding and leaching it's way into the IC body, staking compound doesn't leach or move :)
 

Thread Starter

Chipbox

Joined Aug 13, 2018
16
Hi everyone,
Thank you all for your responses.

I should probably clarify some details of the fabrication process. The way I solder, perhaps somewhat naively, is to first wipe the FR4 pads with IPA, followed by only using a regular soldering iron and leaded solder for the actual job. For these particular cases, I didn't use any epoxy, glue, paste (flux), hot air/oven, or anything else. Perhaps applying some more standard soldering methods (flux and hot air) would be one of those good habits I still need to pick up. These are just prototypes for testing, not any sort of production runs.

That being said, I should note that very high switching voltages (500V+, 100kHz carrier, minimal ringing) also form between these two pins. So it might be plausible to suggest it to be some sort of artifact of electromigration or arcing; although a short didn't actually form when I measured the impedance following desoldering. I am speculating the cause, which is why I came here looking for opinions others may have.

If it is something like arcing for example, then I have a real pickle on my hands because an SOIC-8 simply might not do. A more careful redesign to prevent such high voltage problems might be in order then. I'm thinking along the lines of using an isolated gate driver IC and/or a pulse transformer for the gate. But I think I would still run into the problem of (potentially) inadequate pin spacing.
 

ebp

Joined Feb 8, 2018
2,332
Those gate drivers are routinely used at up to at least five or six hundred volts. HO, VB and VS all sort of "travel together" so there is never much voltage difference among them. There will be large potential difference between VS and LO.

I note that there is gunk that looks much the same to the right of the gate driver IC.

What solder, or more important, what flux in the solder, are you using?

The failure of the driver between HO and VB probably is unrelated to the gunk on the board.
 

Thread Starter

Chipbox

Joined Aug 13, 2018
16
@ebp

- Yes, I do fully agree with you that those pins are all mostly at the same voltage, and that high voltages are periodic between VS and LO. I stand corrected on my wording, and what I should have said was there is a large difference between the drain of the HI FET and the HO/VB/VS pins. However, in the case of the particular failure I had, the gate-drain of both LO and HI MOSFETs shorted, but the gate-source and drain-source of both FETs remained high impedance as they should. In the moment of the fault, I think its plausible to suggest how an instantaneous high voltage spike might have shorted HO/VB. It was also interesting to note that all other pins on the driver were good except those two.

- Back to the solder and the gunk, I am using Kester 24-6337-6422. Its 27 AWG, Sn63Pb37, #66/331 solder. The "gunk" to the right is residue from a desoldered 1206 bootstrap capacitor from its pads.
 

ebp

Joined Feb 8, 2018
2,332
That Kester solder has organic acid water soluble flux. It MUST be very thoroughly washed off within a few hours of soldering. It is corrosive and conductive.

Washing is not easy with surface mount. I used to use (past tense because I don't do any electronics anymore, not because I decided I didn't like it) water soluble flux for both through-hole and surface mount work. My cleaning approach seemed to be effective, but the truth is that without a proper analytical method, it is hard to assess.

I used a fairly powerful fine spray of tap water. I actually used a nozzle intended for agricultural spraying. It made a flat fan-shaped spray. I would carefully spray the board from each edge to the opposite, keeping the angle of the spray low to the board so as to get as much water impact as possible under parts. After doing this, I would blow the board off with clean compressed air, then I would repeat the tap water wash. Another blow-off followed, then two rinses with distilled water. Sometimes after the final distilled water rinse I would do another blow-off and a final rinse with a little fresh 99% isopropyl alcohol (isopropanol; 2-propanol). This is all rather messy, but I have a very large sink in my wet lab.

I adopted the multi-step with blow-off in an effort to avoid entrapment of contaminated water under low-profile surface mount parts. Capillary action tends to keep the first water that hits the board under the parts, so without good high-pressure water impingement, it's hard to get it out. Blow-off forces it out and allows fresh water into the gaps.

I originally assessed my method by omitting the isopropanol rinse, instead drying the board thoroughly, then carefully applying small drops of alcohol around parts and allowing it to evaporate. If I could see any evidence of residue from the evaporated alcohol, I knew cleaning was inadequate. Unfortunately, the lack of visible residue really isn't adequately conclusive. Commercial assembly houses generally use deionized water throughout, and conclude the board is clean when the conductivity of the "used" water is sufficiently high. I reject the assumption that that is necessarily meaningful due to water entrapment issue.
 
Last edited:

IamJatinah

Joined Oct 22, 2014
136
did you see the part about cleaning the board of all flux .
yep, always clean rework with 99% alcohol (pharmacy or industrial) and wipe or dab any leftovers, I've been IPC certified for decades got another email today, should be listed at IPC due to recert this next year, all levels as well as 610c, Mil., and NASA compliant via USAF PMEL since the 80's ... why might you ask about cleaning? I never brought up flux? But, flux should be used during rework and cleaned. puzzled.... and going back to not offering opinions....y'all got this
 
Top