Board failure, with small fire

[Josh Logan]

When powering up the board tonight after flashing the latest firmware it caught on fire...

Can anyone provide some suggestions on what would cause this and if you think it's repairable?

I had the board connected to an ATX power supply, and an Ethernet cable in the SDK.  It had worked previously with this same setup.  The LEDs lit up for about 2-3 seconds before I saw and smelled the failure.

Thanks, Josh KD7HGL

[Steve Haynal]

Hi Josh,

Sorry about your fire! You definitely have the worst failure reported so far. My best theory is that a filter capacitor at the location failed and became a short. I can't imagine anything in the new firmware doing this. This may be repairable:

1. Use a brush with stiff bristles and some isopropyl alcohol to to clean that area of the board: http://www.instructables.com/id/Cleaning-up-your-PCB/.
2. Remove any fragments of capacitor or ferrite bead from that location.
3. Check for shorts. You may have to remove the 4 ferrite beads from the front and check for shorts as described in the assembly instructions.
4. If there are not shorts, use the extra capacitors and ferrite bead in the kit to replace the fried components.
5. If there are shorts, it may be that the ground/power planes shorted in this area. The boards are e-tested during fabrication, but the flexing and handling during programming may have caused a near failure to finally breakdown and fail. If this is the case, you may have to drill a hole through the board at this location... We can discuss this more on the list if it comes to this.
6. Test that the BeMicro SDK still works in test mode.  
   

Good luck and let the list know how it goes.

73,

Steve

KF7O

[Josh Logan]

The BeMicro SDK works in test mode.

Thanks for the trouble shooting steps.  I'll give them a try.

[Steven B. Dick]

Josh, most sorry to hear about your fire. If I had to guess, and not sure of its location, but it may be failure of a 10uF tantalum capacitor, BOM line item 3, AVX Corp F931A106MAA, Digikey P/N 478-8193-1-ND.  In my former life as a design engineer, we were not permitted to use tantalum capacitors for power supply decoupling unless they were surge rated.  This capacitor is not surge rated.  If the power supply is not current limited at turn-on, tantalum capacitors can fail catastrophically due to the high initial inrush current and can actually catch on fire.  We had such an incident which resulted in banning non-surge rated tantalum capacitors in applications where the capacitor can be subject to high inrush currents. Steve, you may want to look into an alternate capacitor?  Can ceramic capacitors do the job or do you really need the tantalums?

 

Regards,

"Digital Steve", K1RF  

---

From: herme...@googlegroups.com [mailto:herme...@googlegroups.com] On Behalf Of Steve Haynal
Sent: Tuesday, February 17, 2015 1:39 AM
To: herme...@googlegroups.com
Subject: Re: Board failure, with small fire

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Sid Boyce]

I powered up my second Lite with the SDK powered from the Lite.
Everything looked good and I turned my attention to software then smelt burning.

The 3.3V lead's insulation had melted fusing the GND and +5V into a congealed mess and smoke was seen around the +3.3V FB's 1, 2, 3 and 4.

I powered off and resistance checks showed very low resistance to GND (~ 5 ohms) on both voltages.

Removed the Frontend board and low resistance disappeared.

Checked the Frontend board on its own - no low resistance.

Plugged in the Frontend board - no low resistance.

It would seem that any slight offset when plugging into the Sullins connector can cause problems.

I powered up and everything appears fine but I get no signals on receive, just spurs.

Some troubleshooting still to be done.
No idea if the AD9866 suffered damage.
73 ... Sid.

 
On 17/02/15 06:59, Josh Logan wrote:

The BeMicro SDK works in test mode.

Thanks for the trouble shooting steps.  I'll give them a try.

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

-- 
Sid Boyce ... Hamradio License G3VBV, Licensed Private Pilot
Emeritus IBM/Amdahl Mainframes and Sun/Fujitsu Servers Tech Support
Senior Staff Specialist, Cricket Coach
Microsoft Windows Free Zone - Linux used for all Computing Tasks

[John Williams]

Steve, I have some of these on the PA designs, for Power decoupling. One on the input 12V and one on the 12v directly near the PA MosFET. Can SMT Electrolytics overcome this issue?  When you say surge rated, are there any parameters to pay attention to?

I am using these parts...

10uF 16V Tantalum

1206

SMT

Kemet

T491A106M016AT

47uF 16V

SMT Tantalum

Kemet

T491D476K016AT

John

[John Williams]

More data ... At a minimum, I guess I should at least go to 25V or maybe 50V? The surge rating is 1.32x at 85degC and 1.2 at 125degC. Not much margin on 16V...

John

[Steven B. Dick]

A good reference on tantalum capacitor surge current and derating:  http://www.kyocera.co.jp/prdct/electro/pdf/technical/surgtant.pdf

 

The Kemet T491 series caps should be OK but derating for particular caps should be reviewed and strictly follow vendor derating recommendations.  This series is 100% surge current tested on C, D, E, U, V, X sizes. Tantalum caps have a 2 to 1 to 3 to 1 failure mechanism of a short failure vs an open failure from some papers I have read. A power supply with very high current capability can easily cause serious damage to a PC board up to and including a fire if a Tantalum cap fails short.  In fact, for hi rel applications, AVX came out with the T496 series which has a built-in fuse element to prevent serious damage in a fail short situation. These are much too costly in this low cost application, but it indicates the significance of the problem.  I'm perhaps overly sensitive to this problem because I have witnessed it first hand and it resulted in a very costly investigation before the problem was resolved.

 

Regards,

"Digital Steve", K1RF

---

From: herme...@googlegroups.com [mailto:herme...@googlegroups.com] On Behalf Of John Williams
Sent: Tuesday, February 17, 2015 10:13 AM

[Robert Nickels]

On 2/17/2015 1:24 PM, Steven B. Dick wrote:
> a built-in fuse element

Thanks for the good info on tantalum failure modes, Steve. I'm
accustomed to the old radials from the 80s which tend to fail shorted
with the accompanying smoke and sound effects, but draw enough current
to blow the equipment fuse. It seems that at least for those of us
using old PC-type power supplies with high current capacity, it would
be cheap insurance to add an external fuse on the 5 and 3.3 volt supply
lines.

73, Bob W9RAN

[Steve Haynal]

Thanks for the pointers on surge current and tantalum capacitors. It is definitely something to consider going forward. However, I do not think tantalum caps were the cause of the two recent failures. The tantalum caps are far from the burnt area on Josh's board. There are 2 ceramic caps, a ferrite bead and PCB power planes/traces at his failure spot. Sid reports melted insulation, and smoke around the ferrite beads, but not fried tantalum capacitor. We need pictures like those seen in 

https://escies.org/download/webDocumentFile?id=60884 to confirm any tantalum capacitor failure. I think the tantalum capacitors which only filter the 3.3V supply are derated enough for surge in this design.

I thought that an askew card might cause a short. I've seen this happen with the BeMicro/Hermes-Lite connection with no smoke and no ill effect, just a crash. I tried with the Hermes-Lite powered off, measuring shorts between powers/ground, and inserting the frontend card at various different angles. I could not get it to short. 

It is obvious something shorted and then the weakest link (ferrite bead, portion of the PCB) acted like a fuse. These should be repairable. Also, as we experiment with various power supplies to decide on a final onboard version, we need to include overcurrent protection.

73,

Steve

KF7O

[Josh Logan]

The board is working again!

I just needed to clean the board and replace fb5 and fix the trace going to pin 78, carrying DVDD/3.3V to EXP_PRESENT.

I could not get a clean solder solder to CN1, but was able to connect my wire to the trace and then the via near the corner.

Thanks for the suggestions on the repairs.

Josh

KD7HGL

[Steve Haynal]

Great news! I am glad it is working again. Do you think that the BeMicro was plugged in askew when the short happened? I know that has happened to me but the power supply I am using has short circuit protection that kicked in. We should at least put a resistor on EXP_PRESENT in future revisions to limit current.

73,

Steve

KF7O

On Sunday, February 22, 2015 at 11:17:37 PM UTC-8, Josh Logan wrote:

The board is working again!

I just needed to clean the board and replace fb5 and fix the trace going to pin 78, carrying DVDD/3.3V to EXP_PRESENT.

I could not get a clean solder solder to CN1, but was able to connect my wire to the trace and then the via near the corner.

Thanks for the suggestions on the repairs.

Josh

KD7HGL

[Josh Logan]

It was possible misaligned, but not much.  I don't remember trying to reseat it right after it failed to know if it just moved a little.