Forums

Migration from Spartan-2E to Spartan-3E

Started by Stefan Tillich September 26, 2006
Hi,

I'm investigating the possibility to replace an Spartan-2E (xc2s400e) 
with an Spartan-3E (xc3s1200e) device on a PCB.

The core voltage drops from 1.8V to 1.2V. Has anyone experience 
regarding the footprint differences (easy/hard/impossible to adapt) and 
any other differences which must be taken into account?

Regards,

Stefan
Stefan Tillich wrote:
> Hi, > > I'm investigating the possibility to replace an Spartan-2E (xc2s400e) > with an Spartan-3E (xc3s1200e) device on a PCB. > > The core voltage drops from 1.8V to 1.2V. Has anyone experience > regarding the footprint differences (easy/hard/impossible to adapt) and > any other differences which must be taken into account? > > Regards, > > Stefan
The footprint is different; beyond that, all parts can be routed. As noted with the S3 vs S3E thread 2 days ago, there are a couple "gotchas" that you need to be aware of. There are only 4 IO banks, not 8. While not necessarily a "problem" your existing design might dedicate eighths of the chip to a specific VCCIO rather than even quarters. Also, many of the S3E IOBs are input-only to save on die size; while this is fine for many, the loss of flexibility is felt. I wouldn't expect the redesign to be a problem, otherwise. I think S2 had tristate buffers but not S2E; either way, S3 and S3E don't have them. Memory blocks are now larger and DLLs are more functional as DCMs, both have "compatibility modes" for older designs. There's no longer a power-on surge. You also have new, cheaper programming ROM options. I hope the conversion is fun.
Stefan Tillich <stefanti@NOSPAMgmx.at> wrote:

>Hi, > >I'm investigating the possibility to replace an Spartan-2E (xc2s400e) >with an Spartan-3E (xc3s1200e) device on a PCB. > >The core voltage drops from 1.8V to 1.2V. Has anyone experience >regarding the footprint differences (easy/hard/impossible to adapt) and >any other differences which must be taken into account?
The blockrams in the S3 are bigger and S3 does not have tristate busses. -- Reply to nico@nctdevpuntnl (punt=.) Bedrijven en winkels vindt U op www.adresboekje.nl
>I think S2 had tristate buffers but not S2E; either way, S3 and S3E >don't have them. Memory blocks are now larger and DLLs are more >functional as DCMs, both have "compatibility modes" for older designs. >There's no longer a power-on surge. You also have new, cheaper >programming ROM options.
Are you sure the power-on surge is gone ..?, so far what I have seen this issue is on Spartan-3/3E aswell.
pbdelete@spamnuke.ludd.luthdelete.se.invalid wrote:
>> I think S2 had tristate buffers but not S2E; either way, S3 and S3E >> don't have them. Memory blocks are now larger and DLLs are more >> functional as DCMs, both have "compatibility modes" for older designs. >> There's no longer a power-on surge. You also have new, cheaper >> programming ROM options. > > Are you sure the power-on surge is gone ..?, so far what I have seen this > issue is on Spartan-3/3E aswell.
There are power on requirements, indeed. Simply not the 4x operating current (wild estimate, unsubstantiated) of previous parts. I haven't taken a current probe to my well-behaved boards, however. The fundamental problem with the older parts was addressed in the silicon. - John_H
John_H <newsgroup@johnhandwork.com> wrote:
>pbdelete@spamnuke.ludd.luthdelete.se.invalid wrote: >>> I think S2 had tristate buffers but not S2E; either way, S3 and S3E >>> don't have them. Memory blocks are now larger and DLLs are more >>> functional as DCMs, both have "compatibility modes" for older designs. >>> There's no longer a power-on surge. You also have new, cheaper >>> programming ROM options. >> >> Are you sure the power-on surge is gone ..?, so far what I have seen this >> issue is on Spartan-3/3E aswell.
>There are power on requirements, indeed. Simply not the 4x operating >current (wild estimate, unsubstantiated) of previous parts. I haven't >taken a current probe to my well-behaved boards, however. The >fundamental problem with the older parts was addressed in the silicon.
What's the power-on requirements for Spartan-3/3E 200-400k chips then..? I read a some power supply design documents. Seemed like a mess.. ;)
pbdelete@spamnuke.ludd.luthdelete.se.invalid wrote:
> John_H <newsgroup@johnhandwork.com> wrote: >> pbdelete@spamnuke.ludd.luthdelete.se.invalid wrote: >>>> I think S2 had tristate buffers but not S2E; either way, S3 and S3E >>>> don't have them. Memory blocks are now larger and DLLs are more >>>> functional as DCMs, both have "compatibility modes" for older designs. >>>> There's no longer a power-on surge. You also have new, cheaper >>>> programming ROM options. >>> Are you sure the power-on surge is gone ..?, so far what I have seen this >>> issue is on Spartan-3/3E aswell. > >> There are power on requirements, indeed. Simply not the 4x operating >> current (wild estimate, unsubstantiated) of previous parts. I haven't >> taken a current probe to my well-behaved boards, however. The >> fundamental problem with the older parts was addressed in the silicon. > > What's the power-on requirements for Spartan-3/3E 200-400k chips then..? > I read a some power supply design documents. Seemed like a mess.. ;)
There aren't Iccpo values published in the data sheets any more because the power up requirements are captured within the quiescent maximum values on the data sheet. Check out the TechXclusives article Austin Lesea wrote "back in the day" http://tinyurl.com/pfaef or the longurl version: http://www.xilinx.com/xlnx/xweb/xil_tx_display.jsp?iLanguageID=1&category=&sGlobalNavPick=&sSecondaryNavPick=&multPartNum=1&sTechX_ID=al_power While there may not be a large amount of "data" in this article, it addresses the issue that was and where it is now. If you need more specifics, I'm sure you can get it directly from Xilinx, perhaps even Austin. If you maintain the correct supply ramp rates with the Spartan3/L/E/A devices (touched on in that article and in some other Answers Database notes) you should have no surge issues.
Stefan Tillich wrote:
> Hi, > > I'm investigating the possibility to replace an Spartan-2E (xc2s400e) > with an Spartan-3E (xc3s1200e) device on a PCB. > > The core voltage drops from 1.8V to 1.2V. Has anyone experience > regarding the footprint differences (easy/hard/impossible to adapt) and > any other differences which must be taken into account? > > Regards, > > Stefan
Read the FPGA configuration section CLOSELY. The JTAG pins and the dedicated configuration pins are powered from VCCAUX (2.5V). The other non-dedicated configuration pins are powered from Bank2 VCCO pins. I know of two designs this was over looked (1 was mine !!!) and it did cause problems that needed a board re-layout. My design was drawing MUCH more current than it should have and I tracked it down to this. I wish Xilinx would put out a "Gottcha FAQ" that have the top changes from the previous family that will point out problems when moving to the new family.
Stefan Tillich wrote:
> Hi, > > I'm investigating the possibility to replace an Spartan-2E (xc2s400e) > with an Spartan-3E (xc3s1200e) device on a PCB. > > The core voltage drops from 1.8V to 1.2V. Has anyone experience > regarding the footprint differences (easy/hard/impossible to adapt) and > any other differences which must be taken into account? > > Regards, > > Stefan
Read the FPGA configuration section CLOSELY. The JTAG pins and the dedicated configuration pins are powered from VCCAUX (2.5V). The other non-dedicated configuration pins are powered from Bank2 VCCO pins. I know of two designs this was over looked (1 was mine !!!) and it did cause problems that needed a board re-layout. My design was drawing MUCH more current than it should have and I tracked it down to this. I wish Xilinx would put out a "Gottcha FAQ" that have the top changes from the previous family that will point out problems when moving to the new family.