Dangerous Prototypes Open Source Hardware

VR1,2,3 getting very hot in less than 20s

May 04, 2010, 01:04:15 pm

hi all,

just received my OLS from seeed and it's not working.

When I plugged it in, nothing appears on usb.
But when something goes bad, I have a habit of putting my finger on all ICs:

  Nothing get warm except the voltage regulator stage. They're getting really hot, and fast.

I have checked the wiring, and all seems correct, no shortcut.

I then checked the levels at each:
  VR1 (LB33) outputs 0.596V
  VR2 (LB25) outputs 2.508V
  VR3 (ATB23) outputs 1.202V
  V+ (USB) outputs 4.84V

What can I do next?

Re: VR1,2,3 getting very hot in less than 20s

Reply #1 – May 04, 2010, 01:14:42 pm

Hi elbeem, I'm sorry you're having a problem with the OLS.

That looks like a short in the 3.3volt power supply somewhere. You can look through the power trace areas for a solder bridge and try to diagnose and fix it yourself, or you can return it to Seeed for repair or replacement.

Since each OLS was programmed (requires 3.3volt supply) and connected to USB before shipping, I'd guess that a flake of solder came loose in handling or transit and is bridging pins on the FPGA (they're really close together). But that's just a guess.

Re: VR1,2,3 getting very hot in less than 20s

Reply #2 – May 04, 2010, 03:20:50 pm

hmm, nothing obvious to me.

The board is really dirty (it came full of finger prints on the backside + lots of dark zones like manipulated by a car mechanic).
I eye-scanned all the board with my 6X mag, no bridge.
I found two nanoballs of solder lying around, but not entraving electricity, cleared.
I cleaned the board with an antistatic brush, it's now shiny, and I confirmed all the pins are well soldered! (no pin were bent during the cleaning).
then I finished it with air duster. Now the board shines!

But VR1 still produces 0.6V and heat.
I vote for a faulty VR1

What I could do next is remove vr1 and test it separately.
Then if it's bad, and since I have no other 3.3v regulator handy, I can build a diy 5->3.3 res divider and test it.

Will seeed still accept a return If I spot the problem myself?

Re: VR1,2,3 getting very hot in less than 20s

Reply #3 – May 04, 2010, 03:53:59 pm

Seeed has been really good about repairs and exchanges for defects. If it's just a small part I try to repair it or send replacements personally because it's usually faster than waiting for a box from China. If you identify it as a bad VREG, I can send you a new one or do the repair for you.

Re: VR1,2,3 getting very hot in less than 20s

Reply #4 – May 04, 2010, 05:34:59 pm

So, I removed VR1carefully,  and powered on the board. VR2 and VR3 still get hot (hurting !).

I mounted VR1 on a breadboard, and it's perfectly sane, outputing 3.304V.

(see pic)

I will now fire eagle and try to test and confirm each pin of the xilinx.

Re: VR1,2,3 getting very hot in less than 20s

Reply #5 – May 04, 2010, 07:14:44 pm

I made a continuity check on the following signals with success,
between the source of the signal and all the pins from the fpga :

3V3
2V5
1V2
GND

all those signals are routed correctly to the fpga.

any idea what is sucking so many current?

Re: VR1,2,3 getting very hot in less than 20s

Reply #6 – May 04, 2010, 07:37:43 pm

What resistance do you measure between ground and the different voltages?

The problem could be a solder bridge under a C or R or the use of too many flux. The first thing I would check the C's which stabilize the supply voltages.

Re: VR1,2,3 getting very hot in less than 20s

Reply #7 – May 04, 2010, 07:45:32 pm

Here is what I measure:

1V2 : 460Ω
2V5: 66kΩ
3V3: 100kΩ
5V: >30MΩ

I'm a bit surprised by the low value for 1.2, but still acceptable.

Edit:
I've removed C1-C3, nothing wierd under those.
I've then resoldered C1-C3 (My soldering skill can't beat a reflow oven :/ ).
and still a constant 0.6V instead of 3.3 and "calor".

Re: VR1,2,3 getting very hot in less than 20s

Reply #8 – May 04, 2010, 09:52:23 pm

This is a long shot but is the Xilinx chip facing the right direction? The Xilinx should be upside down and on the bottom part of the board.

Jack.

Re: VR1,2,3 getting very hot in less than 20s

Reply #9 – May 04, 2010, 10:06:00 pm

yes, the fpga is correctly placed, dot2dot.

Re: VR1,2,3 getting very hot in less than 20s

Reply #10 – May 04, 2010, 10:18:23 pm

The smaller dot is lined up with the dot on the pcb right? Just double checking because you are seeing the same thing that took me a couple days to figure out when I first built a board using the Spartan 3E chip. The alignment is confusing.

Re: VR1,2,3 getting very hot in less than 20s

Reply #11 – May 04, 2010, 10:35:51 pm

when I read your name on the board, "xilinx" is upsidedown.

the smaller dot is align with the /!

The pic, eprom and the 16245 are correctly aligned too, based on the brd document.

This picture is totally confusing, I admit :


someone must detroy it before someone else try to solder the spartan the wrong way, and the pic too. I wonder who made the picture.

Re: VR1,2,3 getting very hot in less than 20s

Reply #12 – May 04, 2010, 11:02:00 pm

Haha, I made that picture. Funny thing is that the Xilinx logo is generated with the Xilinx Logo correctly facing downwards. But I modified the POV code to turn it around so it would look pretty. When I made the 3d drawing it was before any boards were built and we just wanted to show what it would look like and get a visual idea that everything looked right. I didn't think forward to it causing confusion.

It looks like you have correct orientation then. Attached is a picture of the OLS with the 3.3V net highlighted. That might help you better understand where to look for any bridges etc.



Jack.

Re: VR1,2,3 getting very hot in less than 20s

Reply #13 – May 05, 2010, 08:37:20 am

yea, i've used this process to check the VRx lines and GND.

remember that when I unplugged VR1 (3.3V), all those chips were disconnected from 3.3V
and VR3 (1.2V) (or VR2?) continues to draw so many current, being untouchable in less that 20s.

I've scoped the three VR rise time (with VR1 and C1-C3 resoldered, ofc) :

3.3V , a rise time to 0.6V > 3ms
2.5V, a rise time to 2.5V of 12µs
1.2V, a rise time to 1.2V < 18µs

if the problem is around the 1.2 plane (why a 1.2 plane btw?), all the components could be implied, sic!

Re: VR1,2,3 getting very hot in less than 20s

Reply #14 – May 05, 2010, 11:36:06 am

New tests:

1/ lift up VR3 output (no more 1.2V on board)

  I let the board plugged in for 10mn without any hot place.
  but, VR1 still outputs 0.6V instead of 3.3
  and VR2 began outputing a regular 2.5V then when I  rechecked a few minutes later, it was outputing a singular signal
    (saw signal, oscillating between ~ 2.4V and 2.5V, starting from 2.4V then progressivly going to 2.5, then falling back to 2.4... sorry no sshot).
  At this time, I took a current measure at VR3, since it was lifted up: 12.7mA

2/ lift up VR2 output too (only 5V and ?3.3 on board).

  still 3V3=0.6V instead of 3.3V
 
  if I press reset: 3V3 = 380mV
  if I press update: 3V3 = 380mV
  if I press both: 3V3 = 240mV

3/ lift up VR1 (only 5V on board, no more lower voltages)

  measure at Vr1 output (lift up) : 2.45V (better but still not 3.3)

  since when I put it before on a breadboard, i got 3.3, 2.45V is not acceptable (same usb source btw)

  with the VR1 output lift up, i can measure current:
       660mV and 30µA (!!!)

I can't go further.