1. Home
  2. Electronics Design
  3. Electromigration and/or dielectric erosion in FR4 substrates

Electromigration and/or dielectric erosion in FR4 substrates

I am looking for information on causes and design guidelines for using thin FR4 between DC voltage potentials. I have a couple of data points that FR4 can have punch through and arcing at voltages far below the dielectric breakdown specified for the materials.

1.) 48 Vdc punched through 4-5 mils of FR4 within 6 months continual usage 2.) 35 Vdc punched through 0.4 mils of FR4 in about 630-650 hours 3.) FR4 type was not specified, but these are credible data points.

This (in part) appears to be dielectric erosion exacerbated by Electro-migration of the copper traces/planes. More information would be great.

I am concerned by the data points because I am looking at ~125 Vdc on 4 mil thick FR4 and the product is expected to last 24x7x365.25x10 (613620 hours) minimum.

If any one could help, there doesn't seem to be a lot of information available. Thanks.

Reply to
Stefen
Loading thread data ...

Your data points were probably wet. Literally speaking that is.

Regards,

Boris Mohar

Got Knock? - see: Viatrack Printed Circuit Designs (among other things)

formatting link

Reply to
Boris Mohar

100V/0.004"=31,250 V/in = 375,000 V/ft.

I wouldn't think that you would experience dielectric breakdown across

4 mils. Even 4 mils of dry air. But I haven't ever built a board like you describe.

--Mac

Reply to
Mac

70 years, wow! Must be Western Electric or military.

Matt Roberds

Reply to
mroberds

Why not use Kapton as a substrate instead ?

Can you actually *get* 0.1 mm thick FR4 ?

Graham

Reply to
Pooh Bear

on 4

Close, but it is 10 years with 100% utilization (all day every day), not 70 years:).

Reply to
Stefen

Syntactically, it's close: "twenty-four seven, three hundred sixty-five (and a quarter) days a year" sounds almost natural, since "twenty-four seven" means "continuously" in the vernacular.

Oops! Is the vokda kikcing in alradey?

If 7-11 stores are open 24 hours a day, howcome the doors have locks?

Cheers! Rich

Reply to
Rich Grise

Because, the stores were supposed to open at 7AM and close at 11PM.

Reply to
Robert Baer

look harder. 24*7*365*10 = 70 years. there are *NOT* 7*365.25 days in a year. To put it in perspective, the "normal" working year is 2,050 hours.

Cheers Terry

Reply to
Terry Given

I have come across anecdotal evidence of low-voltage breakdown of FR4 as well. But I looked into the product in question (a commercial UPS), and found some pretty shonky layout/design/workmanship.

Firstly, the PCBs were all made somewhere in China, for the lowest price obtainable (100,000 per annum) no doubt using the cheapest, nastiest material known to man. Looking at half a dozen units, I saw significant variation in the LPI solder mask coverage and screen-printed overlay registration. I would suspect the process is neither clean nor consistent.

The design itself had acute angles, potentially allowing etchant to accumulate. bad CAD man. The copper width/thickness was also inadequate, and the high-current low voltage traces were discoloured. Standard 125C Tg (glass transition temperature) FR4 was used.

For our MIL products we stuck with a large, reputable PCB manufacturer with stringent process & quality control. We used high Tg FR4 (150C IIRC), and designed tracks for a 10C temperature rise (and considered dynamic thermal resistance too). And we very carefully cleaned then conformal coated our PCBs.

One faulty PCB was found thru hi-pot testing (17,000 commercial units) but that had a clearly visible lump of contaminant within the PCB material itself. Which came from China, but was cheap :)

Cheers Terry

Reply to
Terry Given

Okay,

I am sufficiently embarrassed.

Reply to
Stefen

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.