Hi all.
Here are two bad news.
I got two UmTRXv2 and both with the problems of inaccurate assembling. 1st Thomas sent to me and 2nd was new and clear (say, virgin).
Thomas said about power supply symptoms few days ago. I fixed assembling error in DC/DC - power coil (L37) was 90 degrees rotated. It was too big surprise for me, because I replaced all components around coil before I saw that it soldered incorrect. Anyhow,This board began to work, but LMS2 still didn't hears SPI and didn't work. Also I found that mashine lost one 0.1uF cap (C86-1) near LMS1, but it was easily to fix. LMS1 and all other parts, seems working fine.
Second board programmed and working, but FPGA too hot and 3.3V loaded a lot.
Seems, I have to find BGA reworking station to fix both boards finally. Also needed new Spartan-6 FPGA for replace. Alexander, do we have spare on hands?
Best regards, Andrey Sviyazov.
Hi Andrey,
On Fri, Nov 9, 2012 at 9:55 PM, Andrey Sviyazov andreysviyaz@gmail.com wrote:
I got two UmTRXv2 and both with the problems of inaccurate assembling.
I wonder, do they do at least visual inspection of boards after assembly? This again reminds us that we have to create a test procedure for the fab before we start manufacturing in quantities. If you're a community member who could contribute to the test procedure - we would greatly appreciate that (mention on the web-site, t-shirt, even free hardware).
I'm willing to try a simple semi-automated visual testing - make pictures of all boards and then subtract them from a reference one. If performed well, we should clearly see all errors on the resulting picture. The only question is how hard is it to "perform well".
1st Thomas sent to me and 2nd was new and clear (say, virgin).
Thomas said about power supply symptoms few days ago. I fixed assembling error in DC/DC - power coil (L37) was 90 degrees rotated. It was too big surprise for me, because I replaced all components around coil before I saw that it soldered incorrect. Anyhow,This board began to work, but LMS2 still didn't hears SPI and didn't work.
Do you think it's a result of bad soldering and rework might help?
Also I found that mashine lost one 0.1uF cap (C86-1) near LMS1, but it was easily to fix. LMS1 and all other parts, seems working fine.
Second board programmed and working, but FPGA too hot and 3.3V loaded a lot.
You think the FPGA itself is broken?
Seems, I have to find BGA reworking station to fix both boards finally. Also needed new Spartan-6 FPGA for replace. Alexander, do we have spare on hands?
Nothing on hands. We should check with Andrew Karpenkov and his connections at InlineGroup - they should have a stock of those chips and a BGA reworking station. E.g. they repaired our blown SP-605.
-- Regards, Alexander Chemeris. CEO, Fairwaves LLC / ООО УмРадио http://fairwaves.ru
Hi Alexander.
2012/11/10 Alexander Chemeris alexander.chemeris@gmail.com
Hi Andrey,
On Fri, Nov 9, 2012 at 9:55 PM, Andrey Sviyazov andreysviyaz@gmail.com wrote:
I got two UmTRXv2 and both with the problems of inaccurate assembling.
I wonder, do they do at least visual inspection of boards after assembly? This again reminds us that we have to create a test procedure for the fab before we start manufacturing in quantities. If you're a community member who could contribute to the test procedure - we would greatly appreciate that (mention on the web-site, t-shirt, even free hardware).
I am afraid that I didn't understood what do you mean.
I'm willing to try a simple semi-automated visual testing - make pictures of all boards and then subtract them from a reference one. If performed well, we should clearly see all errors on the resulting picture. The only question is how hard is it to "perform well".
Do you mean to design some tooling and firmware/software for testing?
1st Thomas sent to me and 2nd was new and clear (say, virgin).
Thomas said about power supply symptoms few days ago. I fixed assembling error in DC/DC - power coil (L37) was 90 degrees
rotated.
It was too big surprise for me, because I replaced all components around coil before I saw that it soldered incorrect. Anyhow,This board began to work, but LMS2 still didn't hears SPI and
didn't
work.
Do you think it's a result of bad soldering and rework might help?
I'll try to fix LMS2. It is doable for me because required just warm up board and to push LMS2 a little bit.
Also I found that mashine lost one 0.1uF cap (C86-1) near LMS1, but it
was
easily to fix. LMS1 and all other parts, seems working fine.
Second board programmed and working, but FPGA too hot and 3.3V loaded a
lot.
You think the FPGA itself is broken?
If everything work, then extra current goes inside FPGA chip. So, yes, I'm sure, almost.
Seems, I have to find BGA reworking station to fix both boards finally. Also needed new Spartan-6 FPGA for replace. Alexander, do we have spare on hands?
Nothing on hands. We should check with Andrew Karpenkov and his connections at InlineGroup - they should have a stock of those chips and a BGA reworking station. E.g. they repaired our blown SP-605.
It would be fine, if InlineGroup can repair this board. I know that without FPGA repair station I can easy to damage PCB and I want to do it. At least we need to order FPGA with lead solder balls to make repair at lower tem.
Best regards, Andrey Sviyazov.
On Monday, I'll ask about the availability of FPGA in stock of the InlineGroup. After that, I'll ask the man to help us to resolder FPGA. FPGA with lead-free ball would come faster for 2-3 weeks. Then it can be reballing (with lead solder balls) and soldering.
Regards, Andrew Karpenkov
2012/11/10 Andrey Sviyazov andreysviyaz@gmail.com
It would be fine, if InlineGroup can repair this board. I know that without FPGA repair station I can easy to damage PCB and I want to do it. At least we need to order FPGA with lead solder balls to make repair at lower tem.
Hi Andrew.
Sorry for disturbing. How can I send broken PCB to you or directly to InlineGroup?
Today I tried to fix LMS2 on 2nd board, but no results while. No answers via SPI from LMS2. Tommorow I'll remove LMS to test PCB traces under it. I think it is last chance.
Alexander, do we have spare LMS chip?
Best regards, Andrey Sviyazov.
2012/11/10 Andrew Karpenkov plddesigner@gmail.com
On Monday, I'll ask about the availability of FPGA in stock of the InlineGroup. After that, I'll ask the man to help us to resolder FPGA. FPGA with lead-free ball would come faster for 2-3 weeks. Then it can be reballing (with lead solder balls) and soldering.
Regards, Andrew Karpenkov
2012/11/10 Andrey Sviyazov andreysviyaz@gmail.com
It would be fine, if InlineGroup can repair this board. I know that without FPGA repair station I can easy to damage PCB and I want to do it. At least we need to order FPGA with lead solder balls to make repair at lower tem.
On Sat, Nov 10, 2012 at 8:36 PM, Andrey Sviyazov andreysviyaz@gmail.com wrote:
Hi Andrew.
Sorry for disturbing. How can I send broken PCB to you or directly to InlineGroup?
Today I tried to fix LMS2 on 2nd board, but no results while. No answers via SPI from LMS2. Tommorow I'll remove LMS to test PCB traces under it. I think it is last chance.
Alexander, do we have spare LMS chip?
No,, we don't.
Actually I'd ask the fab whether they have an X-ray machine and if yes, then send the broken PCB back to them for inspection. It's much more important for us now to understand what's wrong with production then to repair one more board. If they don't have X-ray machine - I think we could find one in Moscow. What do you think?
-- Regards, Alexander Chemeris. CEO, Fairwaves LLC / ООО УмРадио http://fairwaves.ru
Hi Andrey,
On Sat, Nov 10, 2012 at 7:16 PM, Andrey Sviyazov andreysviyaz@gmail.com wrote:
Hi Alexander.
2012/11/10 Alexander Chemeris alexander.chemeris@gmail.com
Hi Andrey,
On Fri, Nov 9, 2012 at 9:55 PM, Andrey Sviyazov andreysviyaz@gmail.com wrote:
I got two UmTRXv2 and both with the problems of inaccurate assembling.
I wonder, do they do at least visual inspection of boards after assembly? This again reminds us that we have to create a test procedure for the fab before we start manufacturing in quantities. If you're a community member who could contribute to the test procedure - we would greatly appreciate that (mention on the web-site, t-shirt, even free hardware).
I am afraid that I didn't understood what do you mean.
I'm willing to try a simple semi-automated visual testing - make pictures of all boards and then subtract them from a reference one. If performed well, we should clearly see all errors on the resulting picture. The only question is how hard is it to "perform well".
Do you mean to design some tooling and firmware/software for testing?
I think initially (while we're at low quantities) it should be a set of procedures for manual or semi-automatic testing. Like sending commands to UmTRX and checking it's responses, testing Tx and Rx with the LMS loopback or a physical loopback cable. These may need only slight modifications of the UHD code.
A visual inspection, e.g. using a photos of boards, as I described, might help as well.
Later we will need to improve this procedures to be more automatic and this will probably include special tooling, etc. But I think this is only feasible when we move to higher quantity production or at least when we finalize our hardware configuration. Right now we need something to make sure the next batch will be tested.
<snip>
A visual inspection, e.g. using a photos of boards, as I described, might help as well.
Does the assembly house for the UmTRx have access to an AOI (automated optical inspection) setup? This is what you want, and is essentially what Alexander described previously: you work with the assembly house to get a "good" board built, then they train their AOI setup to inspect each board that follows to check for assembly issues that don't match what was done in the good build. This process isn't perfect of course, but we've found that it cuts down on assembly issues considerably.
This shouldn't add any significant cost to the build, and honestly, I don't know if I'd even consider using an assembly house that didn't have this capability. It is a major time saver in the end.
I think we should asking fab to use UV inspection system because of BGA usage. Seems first batch was assembled without any inspection because to small q-ty. On other hand, we have only two broken boards and I am not sure why.
Best regards, Andrey Sviyazov.
2012/11/12 John Orlando john@epiq-solutions.com
<snip> > > A visual inspection, e.g. using a photos of boards, as I described, > might help as well.
Does the assembly house for the UmTRx have access to an AOI (automated optical inspection) setup? This is what you want, and is essentially what Alexander described previously: you work with the assembly house to get a "good" board built, then they train their AOI setup to inspect each board that follows to check for assembly issues that don't match what was done in the good build. This process isn't perfect of course, but we've found that it cuts down on assembly issues considerably.
This shouldn't add any significant cost to the build, and honestly, I don't know if I'd even consider using an assembly house that didn't have this capability. It is a major time saver in the end.
-- John Orlando CEO/System Architect Epiq Solutions http://www.epiqsolutions.com
<snip>
A visual inspection, e.g. using a photos of boards, as I described, might help as well.
Does the assembly house for the UmTRx have access to an AOI (automated optical inspection) setup? This is what you want, and is essentially what Alexander described previously: you work with the assembly house to get a "good" board built, then the assembly house trains their AOI setup to inspect each board that follows to check for assembly issues that don't match what was done in the good build. This process isn't perfect of course, but we've found that it cuts down on assembly issues considerably.
This shouldn't add any significant cost to the build, and honestly, I don't know if I'd even consider using an assembly house that didn't have this capability. It is a major time saver in the end.
-
Alexander Chemeris -
Andrew Karpenkov -
Andrey Sviyazov -
John Orlando -
John Orlando