Large power planes vs. power islands vs. slits for decoupling
Marc Battyani
following page, mentioned in another thread, I noticed that they have put
slits in the power plane around the FPGA:
http://www.nec-tokin.com/english/product/cap/proadlizer/test1.html
Is this a common practice? AFAIK larger planes have a higher capacitance,
but lower frequency resonances, than smaller power islands so does this kind
of slits around the FPGA avoid the lower frequency resonances while keeping
the plane capacitance?
Anybody has some feedback/papers/urls on this subject (large planes vs.
islands vs. slits)?
And BTW what do you think of these capacitors?
Thanks,
Marc
Tim
The NEC-Tokin pages you point to seems to have been written by someone
who was instructed to fill pages while conveying the minimum amount of
information.
As far as I can tell, these devices (4-terminal networks) are not simple
capacitors in the usual sense. I would be truly grateful if a c.a.f.
guru could post a few lines on how they work.
On the plane-splitting point, you can read app notes advocating this
practice and app notes claiming that plane splitting is a discredited
practice from an earlier era. I get the sense that the current (sic)
best practice is that ground plane splitting is rarely a good idea and a
degree of VCC plane splitting may help on a noisy board, but much better
would be to find out why VCC is noisy and fix it. Again, gurus will set
me straight... Also, see Symon's recent posts on puddles.
Tim
Austin
Other than their odd name, they are extremely low impedance t-lines.
As such, they are basically falt at 1200 uF from DC to daylinght.
So, if you put a power supply at one end, and isolate the ground (in
other words, that is what the slit is for) you transfer power to the
other end (hot and ground) with a .001 ohm t-line (looks like that very
low impedance at frequencies up to a few hundred MHz).
So, now you see a AC short, looking either way: from the power asupply
to the load, or from the load to the power supply.
The Playstation 3 uses 8 of them, they isolate each huge ASIC from the
other huge ASICs, and there are3 NO OTHER CAPS on the pcb....
Yes, that is right.
Whereas the Wii has > 300 little caps, PS3 has these 8 "magic" t-line
structures.
Do they work? Well, millions of PS3 aren't chunking away happily for no
reason at all.
We are looking at these seriously to reduce the bypassing requirements
down to the PS3 limit: a few of these, and NOTHING else (no other
bypass caps whatsoever).
So, it seems the NEC-TOKIN part is the first really new invention in
bypassing in many many long years of people who just like to ignore that
power distribution is a real issue, and one that needs some creativity.
My hat is off to the engineers who created this wonder.
Austin
Tim
>
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other
> bypass caps whatsoever).
How do you handle the multiple voltages? Are they expensive?
Jim Granville
Slits in the plane would make sense only if you were trying to keep
FPGA noise out of adjacent areas of the plane.
So, it will not help the FPGA, but it may help something else, which
could be far less tolerant of millivolts of noise, than the FPGA.
-jg
Jim Granville
> Marc,
>
> Other than their odd name, they are extremely low impedance t-lines.
>
> As such, they are basically falt at 1200 uF from DC to daylinght.
>
> So, if you put a power supply at one end, and isolate the ground (in
> other words, that is what the slit is for) you transfer power to the
> other end (hot and ground) with a .001 ohm t-line (looks like that very
> low impedance at frequencies up to a few hundred MHz).
>
> So, now you see a AC short, looking either way: from the power asupply
> to the load, or from the load to the power supply.
>
> The Playstation 3 uses 8 of them, they isolate each huge ASIC from the
> other huge ASICs, and there are3 NO OTHER CAPS on the pcb....
>
> Yes, that is right.
>
> Whereas the Wii has > 300 little caps, PS3 has these 8 "magic" t-line
> structures.
>
> Do they work? Well, millions of PS3 aren't chunking away happily for no
> reason at all.
>
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other
> bypass caps whatsoever).
To make best use of these, you may need to redo the power supply fanout
on the FPGAs.
Or, include them IN the package :)
I could, for example, see merit in allowing a BALL-Free area on the BGA,
designed to allow one of these to mount under the FPGA, on the opposite
side.
> So, it seems the NEC-TOKIN part is the first really new invention in
> bypassing in many many long years of people who just like to ignore that
> power distribution is a real issue, and one that needs some creativity.
>
> My hat is off to the engineers who created this wonder.
These are a SMD variant of T filter/feed thru networks, and a variant on
3 terminal SMD capacitors.
In all those, the power is passed THRU the decoupling element (so they
have a mOhm spec as well ), and it means there is no series L that kills
most trace+via+cap decouplers.
Such feed-thru decouplers, are also very good at keeping noise where it
belongs : their plots show very good localisation of the noise, but
also seem to show higher levels at 'die-central', than when using the
more distributed decoupling.
See my comment above, about using one of these underneath the BGA, on
the rear of the PCB.
I also see their (impressive) impedance plots, are the devices
themselves, and there is no mention of VIA patterns, or trace fanout limits.
-jg
Austin
> How do you handle the multiple voltages? Are they expensive?
Tim,
Use more than one, one for each supply rail.
As for cost, it cn't be that bad: if PS3 needs 8 to replace 300 0402
0.1uF caps?
Austin
Austin
I think what is "new" is that these are true four terminal (two
port)devices.
You think of them with S11, S22, and S21, S12 parameters.
The isolation goes both ways, and from either port, it looks like a dead
AC short (which is better than a capacitor, which only looks like a dead
short at one frequency where it is resonant).
They are actually built like a long t-line, with huge C per unit length,
and a very tiny L per unit length, which leads to their extraordinary
behavior. The careful ratio of LC causes it to act, and appeaqr to be
flat over the wide range of frequencies in terms of it Zo.
Austin
colin
Ausin
Does this mean that xilinx are actively charactersing these devices
for your FPGAs and that we can get hold of the results sometime soon?
Colin
Symon
"Austin" <aus...@xilinx.com> wrote in message
news:esev91$s...@cnn.xsj.xilinx.com...
> Marc,
>
> Other than their odd name, they are extremely low impedance t-lines.
>
> As such, they are basically falt at 1200 uF from DC to daylinght.
>
Unfortunately, once you mount them, they no longer work to infinity (or
beyond). As ever, the design is limited by the physical layout of the parts
and connections, not the devices.
>
> So, if you put a power supply at one end, and isolate the ground (in other
> words, that is what the slit is for) you transfer power to the other end
> (hot and ground) with a .001 ohm t-line (looks like that very low
> impedance at frequencies up to a few hundred MHz).
>
I think this paragraph is wrong, Austin, there are no slits in the ground
plane in the designs shown at:-
http://www.nec-tokin.com:80/english/product/cap/proadlizer/test1.html.
The slots are only in the power planes. Ground isn't isolated, that would be
very bad indeed. You've got to get signals in and out as well as power.
But hold on a minute, I see on this site they show the thing used with a
slot in the ground plane.
http://www.chemi-con.co.jp/english/support_e/proadlizer_e.html
Ouch! What happens to all the signals going to and from the device?
I see here:-
http://www.nec-tokin.com/english/product/pdf_dl/proadlizer_e.pdf
they still mention the ground plane slot, claiming the ground plane slot
offers "more optimal performance" [sic]. My bullshit sense is tingling! I'd
suggest this is not true for a real system with signals traversing this
ground plane slot. It makes me wonder if they really know for what this
device should be used.
>
> So, now you see a AC short, looking either way: from the power asupply to
> the load, or from the load to the power supply.
>
> The Playstation 3 uses 8 of them, they isolate each huge ASIC from the
> other huge ASICs, and there are3 NO OTHER CAPS on the pcb....
>
> Yes, that is right.
>
> Whereas the Wii has > 300 little caps, PS3 has these 8 "magic" t-line
> structures.
>
> Do they work? Well, millions of PS3 aren't chunking away happily for no
> reason at all.
>
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other bypass
> caps whatsoever).
>
> So, it seems the NEC-TOKIN part is the first really new invention in
> bypassing in many many long years of people who just like to ignore that
> power distribution is a real issue, and one that needs some creativity.
>
> My hat is off to the engineers who created this wonder.
>
> Austin
>
OK, what these devices appear to be good at is isolating the power supply
noise from the ASICs/FPGA and stopping it spreading over the board. They
also provide bypassing for the ASICs/FPGA. It's kinda like a series ferrite
and a parallel capacitance at the ASIC/FPGA end. The 'new' bit is that it's
all integrated onto a nice low ESL package. Sadly, the end user then has to
attach this to his circuit which will affect this impedance. The blurb on
the site shows how they localise EMC problems, reducing emissions, more than
what great bypass
capacitors they make.
I see these parts got a brief mention on SI-list back in Jan '04, so I'd
suggest they're not a 'miracle breakthrough' or we'd have heard more about
them already. I would tend to go for the X2Y stuff we discussed last week on
CAF.
http://www.googlefight.com/index.php?lang=en_GB&word1=proadlizer&word2=x2y+component
Or, perhaps slightly more seriously, :-
http://www.x2y.com/bypass/measure/comparative_device_only.pdf
Anyway, whatever else, it's nice to see people doing designs with tiny local
power planes for the FPGA, and not believing all the claptrap about plane
capacitance and resonances to help PDS for big BGA packages.
Cheers, Syms.
Symon
"Austin" <aus...@xilinx.com> wrote in message
news:esev91$s...@cnn.xsj.xilinx.com...
> Marc,
>
> Other than their odd name, they are extremely low impedance t-lines.
>
> As such, they are basically falt at 1200 uF from DC to daylinght.
>
Unfortunately, once you mount them, they no longer work to infinity (or
beyond). As ever, the design is limited by the physical layout of the parts
and connections, not the devices.
>
> So, if you put a power supply at one end, and isolate the ground (in other
> words, that is what the slit is for) you transfer power to the other end
> (hot and ground) with a .001 ohm t-line (looks like that very low
> impedance at frequencies up to a few hundred MHz).
>
I think this paragraph is wrong, Austin, there are no slits in the ground
plane in the designs shown at:-
http://www.nec-tokin.com:80/english/product/cap/proadlizer/test1.html.
The slots are only in the power planes. Ground isn't isolated, that would be
very bad indeed. You've got to get signals in and out as well as power.
But hold on a minute, I see on this site they show the thing used with a
slot in the ground plane.
http://www.chemi-con.co.jp/english/support_e/proadlizer_e.html
Ouch! What happens to all the signals going to and from the device?
I see here:-
http://www.nec-tokin.com/english/product/pdf_dl/proadlizer_e.pdf
they still mention the ground plane slot, claiming the ground plane slot
offers "more optimal performance" [sic]. My bullshit sense is tingling! I'd
suggest this is not true for a real system with signals traversing this
ground plane slot. It makes me wonder if they really know for what this
device should be used.
>
> So, now you see a AC short, looking either way: from the power asupply to
> the load, or from the load to the power supply.
>
> The Playstation 3 uses 8 of them, they isolate each huge ASIC from the
> other huge ASICs, and there are3 NO OTHER CAPS on the pcb....
>
> Yes, that is right.
>
> Whereas the Wii has > 300 little caps, PS3 has these 8 "magic" t-line
> structures.
>
> Do they work? Well, millions of PS3 aren't chunking away happily for no
> reason at all.
>
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other bypass
> caps whatsoever).
>
> So, it seems the NEC-TOKIN part is the first really new invention in
> bypassing in many many long years of people who just like to ignore that
> power distribution is a real issue, and one that needs some creativity.
>
> My hat is off to the engineers who created this wonder.
>
> Austin
>
OK, what these devices appear to be good at is isolating the power supply
noise from the ASICs/FPGA and stopping it spreading over the board. They
also provide bypassing for the ASICs/FPGA. It's kinda like a series ferrite
and a parallel capacitance at the ASIC/FPGA end. The 'new' bit is that it's
all integrated onto a nice low ESL package. The blurb on the site shows how
Symon
"Austin" <aus...@xilinx.com> wrote in message
news:esev91$s...@cnn.xsj.xilinx.com...
> Marc,
>
> Other than their odd name, they are extremely low impedance t-lines.
>
> As such, they are basically falt at 1200 uF from DC to daylinght.
>
Unfortunately, once you mount them, they no longer work to infinity (or
beyond). As ever, the design is limited by the physical layout of the parts
and connections, not the devices.
>
> So, if you put a power supply at one end, and isolate the ground (in other
> words, that is what the slit is for) you transfer power to the other end
> (hot and ground) with a .001 ohm t-line (looks like that very low
> impedance at frequencies up to a few hundred MHz).
>
I think this paragraph is wrong, Austin, there are no slits in the ground
plane in the designs shown at:-
http://www.nec-tokin.com:80/english/product/cap/proadlizer/test1.html.
The slots are only in the power planes. Ground isn't isolated, that would be
very bad indeed. You've got to get signals in and out as well as power.
But hold on a minute, I see on this site they show the thing used with a
slot in the ground plane.
http://www.chemi-con.co.jp/english/support_e/proadlizer_e.html
Ouch! What happens to all the signals going to and from the device?
I see here:-
http://www.nec-tokin.com/english/product/pdf_dl/proadlizer_e.pdf
they still mention the ground plane slot, claiming the ground plane slot
offers "more optimal performance" [sic]. My bullshit sense is tingling! I'd
suggest this is not true for a real system with signals traversing this
ground plane slot. It makes me wonder if they really know for what this
device should be used.
>
> So, now you see a AC short, looking either way: from the power asupply to
> the load, or from the load to the power supply.
>
> The Playstation 3 uses 8 of them, they isolate each huge ASIC from the
> other huge ASICs, and there are3 NO OTHER CAPS on the pcb....
>
> Yes, that is right.
>
> Whereas the Wii has > 300 little caps, PS3 has these 8 "magic" t-line
> structures.
>
> Do they work? Well, millions of PS3 aren't chunking away happily for no
> reason at all.
>
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other bypass
> caps whatsoever).
>
> So, it seems the NEC-TOKIN part is the first really new invention in
> bypassing in many many long years of people who just like to ignore that
> power distribution is a real issue, and one that needs some creativity.
>
> My hat is off to the engineers who created this wonder.
>
> Austin
>
OK, what these devices appear to be good at is isolating the power supply
Symon
"Austin" <aus...@xilinx.com> wrote in message
news:esev91$s...@cnn.xsj.xilinx.com...
> Marc,
>
> Other than their odd name, they are extremely low impedance t-lines.
>
> As such, they are basically falt at 1200 uF from DC to daylinght.
>
Unfortunately, once you mount them, they no longer work to infinity (or
beyond). As ever, the design is limited by the physical layout of the parts
and connections, not the devices.
>
> So, if you put a power supply at one end, and isolate the ground (in other
> words, that is what the slit is for) you transfer power to the other end
> (hot and ground) with a .001 ohm t-line (looks like that very low
> impedance at frequencies up to a few hundred MHz).
>
I think this paragraph is wrong, Austin, there are no slits in the ground
plane in the designs shown at:-
http://www.nec-tokin.com:80/english/product/cap/proadlizer/test1.html.
The slots are only in the power planes. Ground isn't isolated, that would be
very bad indeed. You've got to get signals in and out as well as power.
But hold on a minute, I see on this site they show the thing used with a
slot in the ground plane.
http://www.chemi-con.co.jp/english/support_e/proadlizer_e.html
Ouch! What happens to all the signals going to and from the device?
I see here:-
http://www.nec-tokin.com/english/product/pdf_dl/proadlizer_e.pdf
they still mention the ground plane slot, claiming the ground plane slot
offers "more optimal performance" [sic]. My bullshit sense is tingling! I'd
suggest this is not true for a real system with signals traversing this
ground plane slot. It makes me wonder if they really know for what this
device should be used.
>
> So, now you see a AC short, looking either way: from the power asupply to
> the load, or from the load to the power supply.
>
> The Playstation 3 uses 8 of them, they isolate each huge ASIC from the
> other huge ASICs, and there are3 NO OTHER CAPS on the pcb....
>
> Yes, that is right.
>
> Whereas the Wii has > 300 little caps, PS3 has these 8 "magic" t-line
> structures.
>
> Do they work? Well, millions of PS3 aren't chunking away happily for no
> reason at all.
>
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other bypass
> caps whatsoever).
>
> So, it seems the NEC-TOKIN part is the first really new invention in
> bypassing in many many long years of people who just like to ignore that
> power distribution is a real issue, and one that needs some creativity.
>
> My hat is off to the engineers who created this wonder.
>
> Austin
>
OK, what these devices appear to be good at is isolating the power supply
Symon
"Austin" <aus...@xilinx.com> wrote in message
news:esev91$s...@cnn.xsj.xilinx.com...
> Marc,
>
> Other than their odd name, they are extremely low impedance t-lines.
>
> As such, they are basically falt at 1200 uF from DC to daylinght.
>
Unfortunately, once you mount them, they no longer work to infinity (or
beyond). As ever, the design is limited by the physical layout of the parts
and connections, not the devices.
>
> So, if you put a power supply at one end, and isolate the ground (in other
> words, that is what the slit is for) you transfer power to the other end
> (hot and ground) with a .001 ohm t-line (looks like that very low
> impedance at frequencies up to a few hundred MHz).
>
I think this paragraph is wrong, Austin, there are no slits in the ground
plane in the designs shown at:-
http://www.nec-tokin.com:80/english/product/cap/proadlizer/test1.html.
The slots are only in the power planes. Ground isn't isolated, that would be
very bad indeed. You've got to get signals in and out as well as power.
But hold on a minute, I see on this site they show the thing used with a
slot in the ground plane.
http://www.chemi-con.co.jp/english/support_e/proadlizer_e.html
Ouch! What happens to all the signals going to and from the device?
I see here:-
http://www.nec-tokin.com/english/product/pdf_dl/proadlizer_e.pdf
they still mention the ground plane slot, claiming the ground plane slot
offers "more optimal performance" [sic]. My bullshit sense is tingling! I'd
suggest this is not true for a real system with signals traversing this
ground plane slot. It makes me wonder if they really know for what this
device should be used.
>
> So, now you see a AC short, looking either way: from the power asupply to
> the load, or from the load to the power supply.
>
> The Playstation 3 uses 8 of them, they isolate each huge ASIC from the
> other huge ASICs, and there are3 NO OTHER CAPS on the pcb....
>
> Yes, that is right.
>
> Whereas the Wii has > 300 little caps, PS3 has these 8 "magic" t-line
> structures.
>
> Do they work? Well, millions of PS3 aren't chunking away happily for no
> reason at all.
>
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other bypass
> caps whatsoever).
>
> So, it seems the NEC-TOKIN part is the first really new invention in
> bypassing in many many long years of people who just like to ignore that
> power distribution is a real issue, and one that needs some creativity.
>
> My hat is off to the engineers who created this wonder.
>
> Austin
>
OK, what these devices appear to be good at is isolating the power supply
Symon
news:eses5g$sfo$1$8302...@news.demon.co.uk...
>
> On the plane-splitting point, you can read app notes advocating this
> practice and app notes claiming that plane splitting is a discredited
> practice from an earlier era. I get the sense that the current (sic) best
> practice is that ground plane splitting is rarely a good idea and a degree
> of VCC plane splitting may help on a noisy board, but much better would be
> to find out why VCC is noisy and fix it. Again, gurus will set me
> straight... Also, see Symon's recent posts on puddles.
>
> Tim
Hi Tim,
Separate ground planes are very rarely a good idea. AFA I can tell, they
seem to have arisen because mixed signal parts, like ADCs, have separate
analog and digital ground pins. The reason for this is because the package
they are in has some impedance in the connection from the die to the circuit
board. You don't want noisy digital ground currents travelling down the same
connections as the analog ground current as you'll get noise injected into
your analog circuitry. However, once the ground signals get out of the
device and into your negligible-impedance ground plane, there's no problem.
All the ground pins can be joined together. Here's a link:-
http://www.analog.com/analog_root/static/raq/moreInfo/raq_groundingClean.html
Here's a link about signals traversing slots in ground planes (there's tons
of this stuff on t'internet) :-
http://www.hottconsultants.com/techtips/tips-slots.html
The same thing applies when signals travel from one reference (say 'analog
ground') to another (say 'digital ground').
VCC is different because it's not generally used as a reference and modern
devices have many different supplies. It's impractical to have a plane for
every single one. You can achieve good performance by having small local
planes adequately bypassed. The small local planes pool the capacitors
together to achieve the characteristics required. It's also fairly easy to
isolate ICs' supplies from each other this way.
Whenever thinking about this stuff, it's important to remember that the
vias, connections to the package, and package impedance must ALL be taken
into account. It's no help that a Proadlizer has 1pH of inductance if the
vias have 2 orders of magnitude more impedance.
HTH, Syms.
p.s. Apologies for my multi-posts in this thread. My usenet provider told me
it wasn't posting when I guess it really was.
Marc Battyani
"Austin" <aus...@xilinx.com> wrote
>
> Other than their odd name, they are extremely low impedance t-lines.
>
> As such, they are basically falt at 1200 uF from DC to daylinght.
>
> So, if you put a power supply at one end, and isolate the ground (in other
> words, that is what the slit is for) you transfer power to the other end
> (hot and ground) with a .001 ohm t-line (looks like that very low
> impedance at frequencies up to a few hundred MHz).
>
> So, now you see a AC short, looking either way: from the power asupply to
> the load, or from the load to the power supply.
OK for the power supply but a ground island looks like a rather bad thing
for the signal lines. So unless I missed something, I will keep my precious
ground planes with no slits or islands. ;-)
[...]
> We are looking at these seriously to reduce the bypassing requirements
> down to the PS3 limit: a few of these, and NOTHING else (no other bypass
> caps whatsoever).
>
> So, it seems the NEC-TOKIN part is the first really new invention in
> bypassing in many many long years of people who just like to ignore that
> power distribution is a real issue, and one that needs some creativity.
>
> My hat is off to the engineers who created this wonder.
Yes, this is really cool. I'm trying to get some but they seem rather hard
to find.
The only drawbacks I see with them is their size and the huge number of vias
you need to put to link to the planes with a low inductance that can block
the signals routing.
Marc
Marc Battyani
"Symon" <symon_...@hotmail.com> wrote
> Separate ground planes are very rarely a good idea. AFA I can tell, they
> seem to have arisen because mixed signal parts, like ADCs, have separate
> analog and digital ground pins. The reason for this is because the package
> they are in has some impedance in the connection from the die to the
> circuit board. You don't want noisy digital ground currents travelling
> down the same connections as the analog ground current as you'll get noise
> injected into your analog circuitry. However, once the ground signals get
> out of the device and into your negligible-impedance ground plane, there's
> no problem. All the ground pins can be joined together. Here's a link:-
> http://www.analog.com/analog_root/static/raq/moreInfo/raq_groundingClean.html
Very interesting.
> Here's a link about signals traversing slots in ground planes (there's
> tons of this stuff on t'internet) :-
> http://www.hottconsultants.com/techtips/tips-slots.html
> The same thing applies when signals travel from one reference (say 'analog
> ground') to another (say 'digital ground').
>
> VCC is different because it's not generally used as a reference and modern
> devices have many different supplies. It's impractical to have a plane for
> every single one. You can achieve good performance by having small local
> planes adequately bypassed. The small local planes pool the capacitors
> together to achieve the characteristics required. It's also fairly easy to
> isolate ICs' supplies from each other this way.
In fact this was my original question: power islands vs. large power planes.
When I look at the diffrent app notes, there seem to be a consensus on the
use of large planes rather than islands. (excepted this new proadlizer
thing). So is there some apps notes/ paper comparing these different ways to
do a PDS.
> Whenever thinking about this stuff, it's important to remember that the
> vias, connections to the package, and package impedance must ALL be taken
> into account. It's no help that a Proadlizer has 1pH of inductance if the
> vias have 2 orders of magnitude more impedance.
If you look at boards using them, it seems that they are used with a great
number of vias to keep the inductance low (cf photo 2):
http://www.nec.co.jp/techrep/en/journal/g06/n05/t060514.pdf
This can potentially block a large number of routes though.
Marc
Symon
news:Sr6dnXrgTOe...@giganews.com...
>
> "Symon" <symon_...@hotmail.com> wrote
>
>>
>> VCC is different because it's not generally used as a reference and
>> modern devices have many different supplies. It's impractical to have a
>> plane for every single one. You can achieve good performance by having
>> small local planes adequately bypassed. The small local planes pool the
>> capacitors together to achieve the characteristics required. It's also
>> fairly easy to isolate ICs' supplies from each other this way.
>
> In fact this was my original question: power islands vs. large power
> planes. When I look at the diffrent app notes, there seem to be a
> consensus on the use of large planes rather than islands. (excepted this
> new proadlizer thing). So is there some apps notes/ paper comparing these
> different ways to do a PDS.
>
Yeah, there does seem to be a lot of folks designing with large planes. They
do work well, but I think they're an expensive solution to the problem. (An
FPGA has Vccint 1.2V, maybe 2 Vccos, 3.3V and 2.5V, a Vccaux, and maybe MGT
power supplies. Where do you draw the line?)
Now that FPGA circuits are moving into the sub-ns rise time region, I think
designers would do well to look at the solutions used by microwave engineers
over the past decades, rather than try and take the traditional digital
designs into this frequency region. IME, microwave designers eschew power
planes in favour of more ground planes! Of course, on the other hand, they
don't have to worry about 1000 ball BGA packages.
>
>
> If you look at boards using them, it seems that they are used with a great
> number of vias to keep the inductance low (cf photo 2):
> http://www.nec.co.jp/techrep/en/journal/g06/n05/t060514.pdf
>
> This can potentially block a large number of routes though.
>
> Marc
>
>
current to the BGA through vias and balls. I think that the X2Y caps on the
backside of the FPGA are still a good solution.
Cheers, Syms.
Symon
>
>
> OK for the power supply but a ground island looks like a rather bad thing
> for the signal lines. So unless I missed something, I will keep my
> precious ground planes with no slits or islands. ;-)
>
Syms.
Austin Lesea
Yes.
Austin
Austin Lesea
I agree that a slit in the ground plane is going to be an impedance
discontinuity for signals that cross the split.
For that reason, I would probably have any signals that cross the split
have a continuous ground underneath them. That little bit of ground
that now connects the "isolated" ground plane to the rest of the ground
plane is also a DC short across the NEC-Tokin device, but from an AC
traveling wave point of view, it is an open, as the power and ground
currents return to the power supply, and are unlikely to return over a
small bridge between two planes.
I admit that designing this way requires thinking of the DC conditions,
then the AC conditions, on both the power, and the IO. It means you
have all kinds of opportunities to make mistakes, and have worse
behavior, too. Far simpler to stay with one good ground, and isolate
only the power planes (or better yet, don't even isolate the power
planes). But, does "simple" work?
Using the two port by just shorting the two ports together, and putting
it on the far side of the pcb right under the FPGA also is a use model
that might make sense. The issue there is how to connect it to the
power and ground planes with as low an inductance (impedance) as possible.
I could not see the internal layers of the PS3 pcb, but I suspect they
spent a lot of time thinking about this. For them, not only is their
performance to worry about, but also EMI/RFI requirements for the many
regions they wish to sell into (FCC Part 15, etc.).
Most engineers would rather just forget the power until the very end of
their design and layout, and never have to really analyze the power
distribution system (PDS). The PSD for the user's guide for V5 takes a
"here is the BOM" approach: just use these caps, place them like this,
and you are done. The solution offered may be overkill for some
designs, but will generally be adequate for 99% of the applications.
Those that take PSD design much more seriously, will do their own
engineering, and provide their own solution.
Austin
Jim Granville
> But hold on a minute, I see on this site they show the thing used with a
> slot in the ground plane.
> http://www.chemi-con.co.jp/english/support_e/proadlizer_e.html
> Ouch! What happens to all the signals going to and from the device?
> I see here:-
> http://www.nec-tokin.com/english/product/pdf_dl/proadlizer_e.pdf
> they still mention the ground plane slot, claiming the ground plane slot
> offers "more optimal performance" [sic]. My bullshit sense is tingling! I'd
> suggest this is not true for a real system with signals traversing this
> ground plane slot. It makes me wonder if they really know for what this
> device should be used.
I think they focus solely on the Powersupply, and forget about
things like ground bounce ( that does not appear in S21 )
- so from an attenuation viewpoint, dual splits would work,
and could maybe help if that section of the FPGA was internal only, and
had no Pin-drive signals [rare, but not impossible]
I was pondering going even further, by taking a device along these
lines, and assembling into a cavity in the PCB (object is to remove the
vias), but I think that would have problems with getting solder paste
into the cavity ?
Of course, the ideal is to have Xilinx put this into the BGA carrier .
-jg
Symon
news:45ec...@clear.net.nz...
>
> Of course, the ideal is to have Xilinx put this into the BGA carrier .
>
Right.
I wonder if some of those interposer materials might offer a solution in the
future? Some sort of bypass circuit in between the BGA and the PCB somehow?
Dunno, Syms.
Jim Granville
Hmm, yes, the largest one is close to die-sized, at ~17x15mm,
(a monster, and also 2.5mm thick), but for that you get wideband
low-milliohms impedances, and 300uF/1000uF,
so imagine a couple of BGA layers made of this stuff....
- probably worth more to customers than putting the boot-prom into the
package, measured in practical PCB area cost, and EMC/RFI gains....
Interesting this was driven by the games industry, which also gives us
the Cell CPU..
In the short term, I think one of these/rail on the opposite side of the
PCB, would be a usefull step, that only needs the Chip & FPGA vendors to
talk to each other.
Of course, Xilinx's volumes are far less exciting than PS3.
It's a little hard to decode the exact footprint/keepouts of these.
-jg
sw...@x2y.com
> Some in-line stuff about Proadlizer, my other comments at the end.
>
> "Austin" <aus...@xilinx.com> wrote in message
>
> news:esev91$s...@cnn.xsj.xilinx.com...
>
> > Marc,
>
> > Other than their odd name, they are extremely low impedance t-lines.
>
> > As such, they are basically falt at 1200 uF from DC to daylinght.
>
> Unfortunately, once you mount them, they no longer work to infinity (or
> beyond). As ever, the design is limited by the physical layout of the parts
> and connections, not the devices.
>
>
>
> > So, if you put a power supply at one end, and isolate the ground (in other
> > words, that is what the slit is for) you transfer power to the other end
> > (hot and ground) with a .001 ohm t-line (looks like that very low
> > impedance at frequencies up to a few hundred MHz).
>
> I think this paragraph is wrong, Austin, there are no slits in the ground
> The slots are only in the power planes. Ground isn't isolated, that would be
> very bad indeed. You've got to get signals in and out as well as power.
>
> But hold on a minute, I see on this site they show the thing used with a
> Ouch! What happens to all the signals going to and from the device?
> CAF.http://www.googlefight.com/index.php?lang=en_GB&word1=proadlizer&word...
> Or, perhaps slightly more seriously, :-http://www.x2y.com/bypass/measure/comparative_device_only.pdf
>
> Anyway, whatever else, it's nice to see people doing designs with tiny local
> power planes for the FPGA, and not believing all the claptrap about plane
> capacitance and resonances to help PDS for big BGA packages.
>
>
> - Show quoted text -
All, Austin's description of the NEC Proadlizer is fairly accurate.
It is a transmission line filter. The really wilde S21 insertion loss
curves occur when the device interrupts both Vcc and Vss ( gnd ). But
it is still quite impressive interrupting just Vcc. When we decouple
power to a device or a node, we are concerned with two issues: S21
insertion loss which measures the ability of a device to isolate
noise, and Z22 which is the impedance that the filter presents to the
load. While Proadlizers with plane slits are killer at S21, they are
quite pedestrian at Z22 exhibiting about 200pH mounted with a ton of
vias. If you were to try and use these for Virtex 4 or 5 in the 672
pin or smaller packages, or Altera parts prior to Stratix III that do
not have internal bypass caps on an I/O rail you would set yourself up
for a world of hurt. Using a Proadlizer with larger V4 or V5 or
Altera Stratix III where the chips do have substantial bypass caps in
the package can work to isolate the local bypass from the plane.
Why do we want to bypass large devices from the planes? Because by
doing so PROPERLY, we can: reduce EMI propagation to the board edges,
raise the SRF of the power cavity to put it well above the cross-over
frequency of power distribution low pass in the package, and isolate
big hungry parts from smaller parts and each other. In the PS3
application, those mighty ASICs have a lot of bypass under the lid.
>From a system design perspective this is cheaper than trying to make
the PCB do all the work. The PCB just becomes a low frequency power
distribution network. Since Sony isn't asking the PCB to distribute
high frequency, using the Proadlizers to isolate noise is an effective
way to limit EMI propagation. That allows meeting FCC with thicker
dielectric in the power planes of the PCB.
The last I checked, Proadlizers were in the dollars range / part. But
the capacitor industry is very competitive and this may have improved.
X2Y's ( I consult for X2Y ) can also be used to effect high frequency
isolation by feeding Vcc through the G1/G2 connection and grounding
the A and B connections. We have seen EMI improvements of 50dB with
carefully designed etch and a pair of 1206 size X2Ys.
On the load side of things ( your chip ) you need to pay attention to
the local impedance versus frequency, Z22. Problems occur if the
impedance is too high. At audio frequencies this is usually a
resonance problem between the voltage regulator and the bulk bypass
network. Above 1MHz it is either from too much inductance and/or a
resonance.
Regards,
Steve.
Symon
news:1173137413.4...@n33g2000cwc.googlegroups.com...
> In the PS3
> application, those mighty ASICs have a lot of bypass under the lid.
>>From a system design perspective this is cheaper than trying to make
> the PCB do all the work. The PCB just becomes a low frequency power
> distribution network. Since Sony isn't asking the PCB to distribute
> high frequency, using the Proadlizers to isolate noise is an effective
> way to limit EMI propagation. That allows meeting FCC with thicker
> dielectric in the power planes of the PCB.
>
>
> Steve.
>
Hi Steve,
Firstly, thanks for your post. I lurk on the SI-list (posting is a pain with
all those 'out-of-office' replies) which is how I found your stuff on X2Y.
The paragraph of yours I've quoted above explains to me what's going on
inside a PS3. For a while I was worrying that I'd missed a big trick with my
FPGA board designs. If FPGA manufacturers follow this lead of putting even
more bypassing on the BGA carrier, and I guess that soon they will have no
choice, then the PCB power distribution requirements will become a little
less rigorous.
Thanks again, Symon.
Marc Battyani
<sw...@x2y.com> wrote
Interesting. Here is a proadlizer test showing this on pages 16-17:
http://home.att.net/~istvan.novak/papers/DC05_TF7.pdf
It looks like the proadlizer is not that good in decoupling mode though
200pH is still pretty good for 1200湩 and that a few low ESL caps are
probably needed anyway.
Marc
Tim
> ...............If FPGA manufacturers follow this lead of putting even
> more bypassing on the BGA carrier, and I guess that soon they will have no
> choice, then the PCB power distribution requirements will become a little
> less rigorous.
Which cap geometries are used for in-package decoupling?
Marc Battyani
According to the following paper (see page 34), they use (used?) 8 pins IDC
caps:
http://home.att.net/~istvan.novak/papers/DC05East_TFMP2_v1.pdf
X2Y and 10 pins IDC caps are much better now so maybe they changed.
Marc
sw...@x2y.com
SI-List sanity.
Intel has been doing what you mention for a long time. The cut-off
frequency between the PCB and package is down in the low MHz, and
falling. For all of those other chips, the keys to good decoupling
are: low inductance, low inductance, and low inductance. That is why
I like X2Y's so much and consult for X2Y. Six vias with one X2Y will
get you 100pH at the surface + 4-5pH / mil for plunge down to the
power cavity. Only IDC caps come close with similar numbers for eight
vias. Resonance management is a matter of managing phase. That too
means getting low inductance.
I have long been fascinated with the Proadlizers but have never found
a situation where I felt they were the best answer. They are (or
were ) pricey, big and need a lot of vias. I have always found I
could synthesize a cheaper solution with cheaper capacitors, sometimes
a little creative etch, and in the very rare instance a little iron.
The FPGA guys have an interesting set of trade-offs to resolve. They
tend to use power pins for signal return path as well as power
distribution. They also don't get to choose return path of things
like DDR2 memory, which is Vss for data and Vccio for address /
control. As edge rates continue to push upwards, this gets trickier
to manage.
Regards,
Steve.
On Mar 6, 3:20 am, "Symon" <symon_bre...@hotmail.com> wrote:
> <s...@x2y.com> wrote in message
Martin Thompson
> Symon, you are welcome. An out of office junk filter is mandatory for
> SI-List sanity.
>
I use gmane.org to lurk on SI-List - it allows you to use a proper
newsreader, which makes life a lot nicer...
--
martin.j...@trw.com
TRW Conekt - Consultancy in Engineering, Knowledge and Technology
http://www.conekt.net/electronics.html
glen herrmannsfeldt
> I agree that a slit in the ground plane is going to be an impedance
> discontinuity for signals that cross the split.
> For that reason, I would probably have any signals that cross the split
> have a continuous ground underneath them. That little bit of ground
> that now connects the "isolated" ground plane to the rest of the ground
> plane is also a DC short across the NEC-Tokin device, but from an AC
> traveling wave point of view, it is an open, as the power and ground
> currents return to the power supply, and are unlikely to return over a
> small bridge between two planes.
My first thought on your previous post on split power planes, and
considering them as transmission lines, is that you can distribute
the effect of the clock transitions back to the power supply.
If you have multiple FPGAs on a board, all clocked at the same
time, but with different length low impedance transmission lines
to the power supply, (bypass capacitors), you distribute the effect
of those capacitors. Then again, you should be able to
get a similar effect with different clock delays.
If you have a split ground plane it would seem that the signal
lines should go around, and not cross the split.
-- glen
KJ
> If you have a split ground plane it would seem that the signal
> lines should go around, and not cross the split.
>
split. The return current for that signal will follow it's copper
which is in the ground or power plane so it will be forced around the
split. This separates the signal current from it's return current
which causes a nice big radiating loop....bad for EMI if the edge rate
on that signal is single digit nanoseconds.
Kevin Jennings
glen herrmannsfeldt
>>If you have a split ground plane it would seem that the signal
>>lines should go around, and not cross the split.
KJ wrote:
> The signal will go where the copper is which is over the ground plane
> split. The return current for that signal will follow it's copper
> which is in the ground or power plane so it will be forced around the
> split.
What I meant was the the copper should follow the ground plane,
such that the signal does go that way. It says signal lines,
so I don't see how you would read it any other way.
-- glen