RoHS

I've read accounts indicating that RoHS compliance has been known to cause 
issues in high-speed designs.  Being that FPGA's are usually at the core of 
a lot of these designs, I thought I ask here if anyone has had first-hand 
experience with these problems.

RoHS compliance is great, but not at the expense of reliability or 
manufacturability.  If RoHS compliance isn't desirable due to these or other 
issues, what approach have you or your company taken?

Thanks,

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
eCinema Systems, Inc.

To send private email:
x@y
where
x = "martineu"
y = "pacbell.net"

Well, I dont have any account of FPGA problems, but you have to be ready for 
trouble when it comes to fabrication, and pcb design.  PCB material has to 
change to meet the new standards (RoHS is not just about lead) New materials 
= New impedances... High speed designs might depend on certain track 
impedances, and if you try and port an old design to the new process - they 
won't match!
Then there's the problem of finding _ALL_ components on your design in a 
RoHS version... and finding RoHS solder, and then putting it all together.
A soldered joint that has been made with a RoHS process looks different to a 
traditional joint... so good joints might be flagged as bad etc etc.
Anyways, I don't have FPGA specific comments, but obviously the FPGA is 
going to be on a PCB with other devices. HTH
Ben

"Martin" <0_0_0_0_@pacbell.net> wrote in message 
news:eOief.20170$dO2.7437@newssvr29.news.prodigy.net...
> I've read accounts indicating that RoHS compliance has been known to cause 
> issues in high-speed designs.  Being that FPGA's are usually at the core 
> of a lot of these designs, I thought I ask here if anyone has had 
> first-hand experience with these problems.
>
> RoHS compliance is great, but not at the expense of reliability or 
> manufacturability.  If RoHS compliance isn't desirable due to these or 
> other issues, what approach have you or your company taken?
>
> Thanks,
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Martin Euredjian
> eCinema Systems, Inc.
>
> To send private email:
> x@y
> where
> x = "martineu"
> y = "pacbell.net"
>
>
> 

Martin wrote:

> I've read accounts indicating that RoHS compliance has been known to cause 
> issues in high-speed designs.  Being that FPGA's are usually at the core of 
> a lot of these designs, I thought I ask here if anyone has had first-hand 
> experience with these problems.
> 
> RoHS compliance is great, but not at the expense of reliability or 
> manufacturability.  If RoHS compliance isn't desirable due to these or other 
> issues, what approach have you or your company taken?

There is no way around RoHS compliance, at least
over here. There are some exception for certain
industries. Unless you belong to them,
You just have to make it work.
The exception are being constantly reevaluated.
So if your competition can make it work without
lead, you'll be out soomer or later.

Rene
-- 
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

To expand on Rene's comment, there's no way around RoHS compliance
**anywhere**. Molex, for instance, has obsoleted all it's high density
parts (to my knowledge) in favour of new RoHS compliant parts. They are
probably doing this across the board, as they don't want to keep
different stocks on hand with the logistics nightmare involved.

As to high speed problems:

Different materials will undoubtedly cause impedance vectors to change.
In addition, the loss tangents across materials (very important in
extremely high frequency designs) will vary. Whether they will be
better or worse is unknown, but after all the research into existing
materials, my take is they will be worse.

IC packaging will have to change. Most existing plastic packages have
absolute maximum temperature ratings of about 220C, but the RoHS reflow
profile is well above this (about 260C).
See this Xilinx app note:
http://direct.xilinx.com/bvdocs/appnotes/xapp427.pdf

Changing the materials to make a device compatible with those reflow
characteristics is bound to have an effect somewhere, although I don't
know the specifics for Xilinx.

In addition to that, the solder joint of a tiny part (such as a
coupling cap) has parasitics that are (somewhat) material dependent.
Change the materials, change the parasitics. This too is an issue for
high speed designs.

Note - when I mention extremely high frequencies, I am referring to bit
times (on the wire pair) of 1nSec or less (PCI express, InfiniBand and
Fibre channel come to mind).

So will it affect us in high speed designs? Undoubtedly.

The issue is we probably won't know the effects until there are large
numbers of boards already made.

Cheers

PeteS

PeteS wrote:
[snippage]
> In addition to that, the solder joint of a tiny part (such as a
> coupling cap) has parasitics that are (somewhat) material dependent.
> Change the materials, change the parasitics. This too is an issue for
> high speed designs.

The thermal resistance will also be different.  (I don't know whether
it will be better or worse, but Murphy says it will be worse.)  A
significant amount of heat leaves the package through the balls, and
this may matter for high speed designs.

Regards,
Allan

..A
significant amount of heat leaves the package through the balls, and
this may matter for high speed designs.
...

Very true (not so much if it's a flipchip ;)

It will most definitely change the thermal characteristics. That being
the case, we'll either get poorer guaranteed performance, or poorer
temperature ranges. I'd take the termperature range and cool it (of
course, that adds a whole new dimension).

Cheers

PeteS

PeteS wrote:

> To expand on Rene's comment, there's no way around RoHS compliance
> **anywhere**.

Long term, yes.  But, I'm not sure that I agree in terms of a mid-2006 hard 
deadline (which is a ridiculous idea IMHO).

As a small manufacturer, if I have to choose between a potential support and 
reliability nightmare from a shift to RoHS and simply not offering affected 
product in the EU, I'd go for option #2.  New designs can be executed for 
RoHS compliance from the start, but few small to medium manufacturers are 
going to go through the trouble, expense and liability of redoing whole 
product lines unless a non-trivial portion of their business is in the EU. 
While the EU is an important market, the potential downside of a hasty 
transition is significantly more severe than the alternative.

Logistically, we've seen billion-dollar companies who supply us with 
components tell us that they will not have RoHS compliant parts available 
until "sometime in 1Q06".  Clearly, migrating a design to RoHS parts in 
haste --with barely a few months of testing before shipping-- would be a 
horrendously bad idea.

My prediction is that the EU will have to soften the rules a bit because a 
full transition to 100% compliance by all entities that move product into 
the region is quite literally impossible.  Few legislated hard deadlines, 
such as this one, ever execute exactly as written.  Logistically, it is 
impossible for the EU to even approach the orders-of-magnitude in potential 
regulatory work that this will create.  The whole thing, if executed by the 
letter of the law, could be quite damaging to business in general. 
Something's got to give.  Or not.  Who knows?

It'll be interesting to watch.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
eCinema Systems, Inc.

To send private email:
x@y
where
x = "martineu"
y = "pacbell.net"

On Tue, 15 Nov 2005 22:30:43 GMT, "Martin" <0_0_0_0_@pacbell.net> wrote:

>PeteS wrote:
>
>> To expand on Rene's comment, there's no way around RoHS compliance
>> **anywhere**.
>
>Long term, yes.  But, I'm not sure that I agree in terms of a mid-2006 hard 
>deadline (which is a ridiculous idea IMHO).
>
>As a small manufacturer, if I have to choose between a potential support and 
>reliability nightmare from a shift to RoHS and simply not offering affected 
>product in the EU, I'd go for option #2.  New designs can be executed for 
>RoHS compliance from the start, but few small to medium manufacturers are 
>going to go through the trouble, expense and liability of redoing whole 
>product lines unless a non-trivial portion of their business is in the EU. 
>While the EU is an important market, the potential downside of a hasty 
>transition is significantly more severe than the alternative.
>
>Logistically, we've seen billion-dollar companies who supply us with 
>components tell us that they will not have RoHS compliant parts available 
>until "sometime in 1Q06".  Clearly, migrating a design to RoHS parts in 
>haste --with barely a few months of testing before shipping-- would be a 
>horrendously bad idea.
>
>My prediction is that the EU will have to soften the rules a bit because a 
>full transition to 100% compliance by all entities that move product into 
>the region is quite literally impossible.  Few legislated hard deadlines, 
>such as this one, ever execute exactly as written.  Logistically, it is 
>impossible for the EU to even approach the orders-of-magnitude in potential 
>regulatory work that this will create.  The whole thing, if executed by the 
>letter of the law, could be quite damaging to business in general. 
>Something's got to give.  Or not.  Who knows?

Remember there is quite a lot of stuff, especially the sort of things that big FPGAs go into, that
is currently outside the scope of Rohs - this has not been well covered in the press. 
Included categories  consumer electronics, household appliance,  IT/telecom, toys or tools,  so most
test/lab type equipment is not covered.  
There are also currently  specific exemptions for network infrastructure, and servers

Source : http://www.dti.gov.uk/sustainability/weee/RoHS_Regs_Draft_Guidance.pdf

> Remember there is quite a lot of stuff, especially the sort of things that 
> big FPGAs go into, that is currently outside the scope of Rohs - this has 
> not been well covered in the press.

It's worst than this.  The press has little to do with it.  The standard 
itself is nebulous and inconclusive.  Some aspects of it are almost comical 
(no formal way to ascertain compliance...but you can go to jail and be fined 
up to UK$5,000 for a violation).

The real problem is that it has setup a "pass the buck" scenario, as we say 
on this side of the pond.  Resellers ask distributors, who ask manufacturers 
who ask component manufacturers, for RoHS compliance guarantees.  The lack 
of clarity and apparent intransigence of the system makes resellers and 
distributors fearful of anything other than "yes, absolutely, it is RoHS 
compliant".  There is no amount of technical maneuvering that can convince 
someone in fear of a stiff fine and imprisonment that anything other than 
total compliance is safe.

Again, it'll be interesting to watch and see what happens come the middle of 
next year.

To be on-topic.  It took our very large distributor almost a month to come 
up with an RoHS-compliant part number to replace a Virtex 2 we've been using 
in a design for a couple of years.  It surprised me that it took that much 
work.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
eCinema Systems, Inc.

To send private email:
x@y
where
x = "martineu"
y = "pacbell.net" 

Martin wrote:
<snip>
> Again, it'll be interesting to watch and see what happens come the middle of 
> next year.
> 
> To be on-topic.  It took our very large distributor almost a month to come 
> up with an RoHS-compliant part number to replace a Virtex 2 we've been using 
> in a design for a couple of years.  It surprised me that it took that much 
> work.

and there are more escape clauses appearing, as they realise the 
silliness of some of this :

eg from the last link
"..and any lower melting temperature solder required to be used with 
high melting temperature solder to complete a viable electrical connection."

Seems now 'viable connection' and reliable operation are being 
recognized - someone must have pointed out that lead-free may reduce 
safety !

Perhaps if you are unable to _prove_ you have a reliable/viable 
electrical connection ( and that includes thermal cycling stresses and 
fatigue failures, over time ) then these escape clauses now apply ?

  Amazing to see light bulbs exempt - given their short life spans, and
direct waste flows, shouldn't they be first in the line ?

-jg