Decoupling caps on SSTC keep failing / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Decoupling caps on SSTC keep failing

Move Thread

LAN_403

Sigurthr

Wed Jul 25 2012, 07:05AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Hello everyone!

I've been building a few solid state non-DR coils for some time now and have come upon a bit of an issue. I've been successfully using the standard UCC37321 / UCC37322 gate drive chips in a fashion consistent with Steve Ward's SSTC designs for several coils now with no problem. However, when I started making larger higher power coils I found that the decoupling caps across the supply rail and ground at each IC kept failing after several runs. I've been using 50V 0.1uF Tantalum caps

for this function without any issue on coils less than 600VA. On my 1100VA coil they fail almost instantly at full power.

They fail as a dead DC short, which overwhelms my 12V regulator causing it to go in to thermal shutdown. Thankfully I heatsink my regulators very well and all have survived. The first cap to fail had a fiery death, the second one failed silently, the third with a pop/click, and the fourth with a hiss and then click. No pictures, sorry, these caps are far too tiny for my camera to focus on.

I tried using a WIMA MKP 100V 0.01uF cap at one point and though it took much longer to fail, it too failed - silently but warm to the touch.

I've been running the coil without any decoupling caps (at the ICs, I have lots of capacitance across each supply rail for smoothing/filtering) with no issue so far, but keeping RF off the supply rails and out of the expensive ICs sounds like an excellent idea to me, so I would ideally like to replace these caps with ones more suitable to the task. To be honest I'm not even sure if they are really needed given that I have 2000 to 4000uF of capacitance across the supply rail, but Steve Ward seemed to think they were so I'm inclined to trust his judgement. What type of caps should I look for, for this purpose? I assume the caps were over-volted but have no way to test this theory. If it is true, I woul think that I should look for high voltage low capacitance ceramic caps. If it is more of a matter of too much RF current and not an over-volt issue I don't really know what kind of cap to look for. If you have a specific cap in mind, please see if it is available at Mouser.com, I can't order from digi-key and many other places as they do not accept the special characters in my street address.

zzz_julian_zzz

Wed Jul 25 2012, 08:03AM

Registered Member #3964 Joined: Thu Jun 23 2011, 03:23AM
Location: Valenzuela City
Posts: 332

You should look for monolythic capacitors, rated atleast 50vDC 0.1uf

Dr. Dark Current

Wed Jul 25 2012, 11:51AM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

use ceramic or non inductive film caps at around 1uf, tantalum are absolutely usesless for pulse applications. im surprised they survived so long.

Steve Conner

Wed Jul 25 2012, 12:20PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I use 330uF low-ESR electrolytics located right next to the gate drive chips and nothing else. This has always worked fine, and my DRSSTCs have travelled round the world and appeared on TV, so Murphy's law had plenty of chances to do its thing.

The DC block capacitor in series with the GDT sees the same RF current as the decoupling caps, so you could use whatever cap you use for that.

If the same cap fails in the decoupling position and not in the DC block position, you have some other problem with your system.

Sigurthr

Wed Jul 25 2012, 01:35PM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

I did an agressive teardown of the first failed board today and discovered that the 100V WIMA cap

I put in place of the tantalum cap is still fine, it did not fail. Another one of the tant caps failed and I assumed it was the wima one. So, I desoldered it and put it in my parts bin.

@ Dr. Dark Current:
Aha, so it was my choice of cap type and not the ratings! That is good to know! I've never come upon an explanation of what types of caps are good for what purposes other than pulse foil/film caps for TC tanks.

@Steve Conner:
I've been using WIMA FKS-3 250V 0.1uF film caps for the DC blocking on the GDT and the antenna. They've held up to absurd strain (one early test there was corona on my antenna due to the EM field being much larger than I guestimated - I quickly corrected the placement, and nothing failed at all). I do have two left, but I'm not sure they're ideal for decoupling due to their footprint, it would require more inductance from traces/leads to make the connection. They're slightly bigger than 0.75" x 0.25" x 0.5", much larger than the DIP-8 chips, haha. I'm about as sure as one can be when one doesn't have a scope (it's in the mail) that the rest of my system is reasonably OK (probably some slight ringing on the gate waveforms). No other components have ever failed even during unfavorable conditions. Let's hope it's just my poor choice of tantalum caps.

Ok, so then it seems WIMA pulse caps are just fine for this application then, correct? If so, that is what I will use then. I have two 0.01uF 100V FKP-2 caps on hand in my parts bin that I will try as soon as I can get time to rework the board. Are these good enough or should I spend the money for 1uF versions?

Thank you very much for the replies, everyone!

Maurizio Valvo

Thu Jul 26 2012, 05:54PM

Registered Member #3219 Joined: Mon Sept 20 2010, 09:06PM
Location: LeinÃ¬ (Turin), Italy
Posts: 23

Hi, on my driver based on Steve's design I have a parallel of 100uF/25V electrolytic plus 10uF/25V tantalum plus 0.1uF/100V SMD caps on the supply rails of each of the UCC chips and have had no problem at all.

My coil is a full bridge using IXGR60N60C2D1 IGBTs running at around 130kHz, tested up to about 400Apk/1.2kW.

EDIT: my coil is a DRSSTC

HV Enthusiast

Thu Jul 26 2012, 06:55PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

Note that tantalums generally cannot tolerate too much ripple current, so that may be the reason they are burning up, especially if they are inexpensive low-cost tantalums.

For our gate drivers, we either use a 0.1uF ceramic and 10uF tantalum in parallel, or a 0.1uF and 10uF ceramic in parallel on our gate driver bypass.

But, really, i've never heard of anyone smoking the decoupling caps on a gate drive IC. Maybe your caps are just really bad, or maybe they aren't rated for the voltage you are putting on them.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob