Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post your project vehicles here! One thread per vehicle please.
Post Reply
User avatar
Fred
Moderator
Posts: 15431
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by Fred »

Excuse any perceived tone, nothing personal meant, just smashing out a message before bed as quickly and clearly as I can:

No such thing as too much solder. You HAVE to use too much, then you HAVE to wick it off. This has to be your plan up front. Wicking it off should be done carefully, as should the initial flooding. Never persist for a long time in a single session. If it's not going well, let it fully cool, then start again.

Depending on which pins and traces got fucked by severe overheating you may want to order a new CPU and PCB and start again, transferring what you can salvage. Traces only lift when you get things very very hot. Pins only bend when you apply physical force to them with the tip while the solder is soft enough to allow movement. Neither of these things should ever happen.

If you put that much heat into the FTDI IC it will die. They have strong thermal conductivity to the core from the legs and need to be soldered correctly quickly. The CPU is more tolerant, but don't push your luck.

How I do it:
  1. Wash the board with alcohol if it has been corrupted by fingerprints etc. Must be perfectly clean and dry.
  2. Carefully position and align the CPU such that all four sides are correct and have no more than 10% or so misalignment.
  3. Tack a pin or two on opposite corners, ideally just two pins, one on one corner, the other on the diagonally opposite corner
  4. If necessary, carefully melt one at a time and walk it around to align it better. Step by step. One corner, one rotation around the other. Repeat, carefully. This step should be avoided by nailing step 2 the first time.
  5. Once happy with alignment, flood one side that has no solder on it (so as to not free it up to move). I don't use flux, just the rosin cored solder is enough to wet it out with a fresh clean CPU straight out of dry packaging and a gold plated board that has not been contaminated by fingerprints etc.
  6. Let it cool
  7. Flood the opposite side.
  8. Let it cool
  9. Flood one of the remaining two sides
  10. Let it cool
  11. Flood the remaining side
  12. Let it cool
  13. Using the side of a medium tip to get maximum length per heat cycle, wick away the excess solder from one side only. This takes 3 or 4 or 5 passes, trimming the soaked wick off in between each.
  14. Let it cool
  15. If you are left with a few bridged pins, reflood them so there's plenty of solder, don't be afraid to wet out the ones around them, either. Not an issue. If it's doable afterward in one pass, this is better, if not, no big deal anyway.
  16. You guessed it, let it cool
  17. Wick away the reflooded bridged section.
  18. Let it cool
  19. Repeat for other sides, I typically do opposites in pairs with cooling in between.
You now have a perfectly soldered CPU with coarse tools and no fine pitch tips required at all. In fact, using a fine pitch tip for flood/wick is a fail as it's too slow to get heat into it, and the CPU soaks hot before you've even melted the solder.

When flooding a side, do it with thicker solder, 0.7mm or more, and use the heat to wet it out, don't try to spread the solder forcefully with the tip, move the tip and flow the solder in, and repeat. This way you avoid raping/bending pins. Be sure to do this quickly, though, to keep core temperatures low.

Hopefully this makes some sense in the context of your recent experiences?

Now, post up some macro shots of the mess and let's see if you can get away with keeping this PCB, or not.

Fred.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
Andreas
LQFP112 - Up with the play
Posts: 102
Joined: Mon Sep 09, 2013 11:11 pm

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by Andreas »

Thanks for all of that info, Fred! It is very helpful.

I was using 1mm solder. I found that when I got a tub of flux and used that it helped transfer the heat to the solder. When I tried to do it without, I had to use pressure to get the solder that was already on the pins to melt. It also makes the bridges magically disappear like this video: https://www.youtube.com/watch?v=Iu4rfutdm9g

I think both the board and the MCP are toast...although those two pins don't need to be attached.
Image

I have another PCB, but I'll have to order the new components....which take a week to get to me. I might also get some kind of USB microscope or something. It's really hard to check how things came out otherwise.
ivan141
LQFP112 - Up with the play
Posts: 148
Joined: Sat Sep 29, 2012 9:16 pm
Location: Rotterdam, the Netherlands

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by ivan141 »

Unconnected traces for fine pitch footprints almost always lift when soldering manually without lots of experience (you need to get it right in 1 go).
They cant conduct heat anywhere so they get hot very fast. It doesnt look great, but lifted blind traces cant do any harm to the functionality of your CPU.

You're best bet is to use this type of soldering tip:
Image

And preferably use liquid flux in moderation. Youtube has plenty of good instructional video's, but these fine pitch packages really take a lot of practice to get right.

I'd still try to see if the CPU has survived the soldering process.. the FTDI is most likely toast if it had the same treatment.. I fried one myself and I used to do this for a dayjob many years ago.
FABRICA MI DIEM, PVNC!
User avatar
Fred
Moderator
Posts: 15431
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by Fred »

ivan141 wrote:Unconnected traces for fine pitch footprints almost always lift when soldering manually without lots of experience (you need to get it right in 1 go).
Well, at the least, you need to never take them past the temperature of the adhesive under the copper, ever. Hmm, we didn't ask what temperature the man is running. How hot? I typically run hotter than usual to do this, so I can do it fast. But the hotter you go, the faster you have to do it. And the colder you go, the harder it is to go fast. It's a balance.
It doesnt look great, but lifted blind traces cant do any harm to the functionality of your CPU.
Very true. I couldn't make out enough from the above picture to determine if it's whoa or go. How about four in-focus pics, one of each set of legs?
I'd still try to see if the CPU has survived the soldering process..
+1
the FTDI is most likely toast if it had the same treatment.. I fried one myself and I used to do this for a dayjob many years ago.
Touchy little bastards, aren't they!

Fred.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
Andreas
LQFP112 - Up with the play
Posts: 102
Joined: Mon Sep 09, 2013 11:11 pm

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by Andreas »

I did not try to mount the FTDI chip yet. It's good to know that the unconnected traces are prone to do that and it is not much of a problem.

What are the minimum components to install to check if I fried the MCP?

Image

Image

Image

Image
ivan141
LQFP112 - Up with the play
Posts: 148
Joined: Sat Sep 29, 2012 9:16 pm
Location: Rotterdam, the Netherlands

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by ivan141 »

I would expect the power supply circuit, xtal and decoupling caps at the very least.
Two pins on that last picture might be cause for concern, looks like you lifted some traces there. Will still work as long as they didnt break, but you can imagine
that they wont be very reliable. Do get rid of the solder bridges and bent pins in the second and last pics.

The scale of the picture might skew my judgement a bit, but to me it looks like you splurged the flux around way too royally.
Broken traces can be fixed with rework wire should it be necessary. You would be better off testing and practicing your soldering on this board
so it is worth saving.
FABRICA MI DIEM, PVNC!
User avatar
Fred
Moderator
Posts: 15431
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by Fred »

First pic:

Ensure those two pins aren't touching each other or the adjacent ones.

Second pic:

Left pins ok. One pin slightly bent but looks clear, straighten with sharp thing to be sure. Two pins bridged, clear this up. Three pins bridged, clear these up. Two pins bridged, flood this, and wick these two alone.

Third pic:

Looks OK but hard to see with reflections.

Fourth pic:

Looks OK except for broken and/or bridged ones in the middle. Carefully clean this up until it's obviously right.

Repost pics once you've cleaned up the four sides and are comfortable no bridges exist, etc.

Re min parts, you can probably get some love from it with a BDM with almost nothing. Don't expect a burn to work, though. You can install decouplers with their legs long so as to be able to reuse if it's a fail, or sink into the board more if success. XTAL + resistors might be required. PLL should not be required to play with BDM, only with firmware. USB and/or BDM can supply 5V without PSU parts. You'd somewhat need to validate inputs and outputs, though. Otherwise all you're proving is that it's alive at the core.

To be clear, unused pins that are bridged can cause issues depending on pin state etc. Make sure they're all isolated from each other.

And don't forget: viewtopic.php?f=9&t=1952 it'll make the pics clearer.

Fred.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
Andreas
LQFP112 - Up with the play
Posts: 102
Joined: Mon Sep 09, 2013 11:11 pm

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by Andreas »

Thanks for the responses. I'm waiting for when I receive the USB microscope I ordered to arrive today before I fix these issues. I can't see well enough with just reading glasses for this kind of work. I want to make sure the chip isn't completely dead before I proceed and then I can load the firmware....then populate and test LEDs to see if the chip's outputs are working.
User avatar
masterkorp
LQFP112 - Up with the play
Posts: 171
Joined: Tue Jan 17, 2012 11:21 am
Location: Ponte de Lima, Portugal

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by masterkorp »

Andreas wrote:And so it begins......

Image
Nice, where did you get this base/stand for the Jaguar?
Andreas
LQFP112 - Up with the play
Posts: 102
Joined: Mon Sep 09, 2013 11:11 pm

Re: Jaguar 0.7 alpha build for 1972 VW Karmann Ghia

Post by Andreas »

From here ($13) http://www.amazon.com/gp/product/B00Q2TTQEE

I'm taking a break right now. I realized I needed to get a machine shop to do the hole enlarging.....OR buy a milling machine. :D

I'm working on getting approval...and on other projects.
Post Reply