Searching for rad tolerant, non-volatile (once programmable) FPGA (or CPLD).

I'm looking for a rad-tolerant, non-volatile (preferably programmable
at once) FPGA or CPLD, for a new project (a satellite instrument).

After searching the Web, I've found out that Actel manufactures micro
antifuse FPGAs. These would be fine, but I would like to know if are
there any other alternatives besides Actel FPGAs.

The device should have around 2500 user gates or more.

Reliability is a concern to us, and availability of a similar or
equivalent, in System Programmable (ISP), part would be a plus.

Any single chip solution (preferably non-reprogrammable, to avoid
unintentional program changes, due to radiation) would be fine,
including CPLDs.

P.S: ASICs (Field Gate Arrays) would be OK, provided it cost less than
US$10,000.00 and takes less than a month or two to be fabricated. In
case we opt for an ASIC, only around 3-5 units would be produced.

If you are willing to help, you can e-mail me directly:
  maaf _att_ dea.inpe.br    (replace the _att_ by an @)

Best Regards,
Marcio
---------------------------------------------------
Marcio Afonso Arimura Fialho
Junior Electronic Engineer
DEA - Divis�o de Eletr�nica Aeroespacial
INPE - Instituto Nacional de Pesquisas Espaciais (Brazilian National
Institute for Space Research).

mitrusc1980-newsgroup@yahoo.com.br (Marcio A. A. Fialho) wrote:
> After searching the Web, I've found out that Actel manufactures micro
> antifuse FPGAs. These would be fine, but I would like to know if are
> there any other alternatives besides Actel FPGAs.

Since Xilinx stopped manufacturing such fpgas, I know no alternative
fpga supplier for Actel.
You could of course contact Xilinx to see if a reprogramable device is
suitable for you.

> The device should have around 2500 user gates or more.

What to you mean with "user gates"? 2input-NAND? 4 input LUT
When an RH1020 is to small you need an RT54SX32S. But be aware,
there's currently a relability problem when using the RT54SX-S
devices.

> Reliability is a concern to us, and availability of a similar or
> equivalent, in System Programmable (ISP), part would be a plus.

When using the 5V IO, you will have trouble finding a reprogrammable
equivalent.

> P.S: ASICs (Field Gate Arrays) would be OK, provided it cost less than
> US$10,000.00 and takes less than a month or two to be fabricated. In
> case we opt for an ASIC, only around 3-5 units would be produced.

I know no alternative when spending 30-50k$ for that number. AFAIK
there are Actel fpgas above 10k$ per device.

Marcio A. A. Fialho wrote:
> I'm looking for a rad-tolerant, non-volatile (preferably programmable
> at once) FPGA or CPLD, for a new project (a satellite instrument).
> 
> ...
> P.S: ASICs (Field Gate Arrays) would be OK, provided it cost less than
> US$10,000.00 and takes less than a month or two to be fabricated. In
> case we opt for an ASIC, only around 3-5 units would be produced.
> 

Hmmm... have you checked the prices of rad hard parts? You might be in 
for a rude shock, even with Actel FPGAs.

Thanks a lot Thomas. I'm glad you've answered this so fast.
I apologise for not posting an answer fast. That's because
I was very busy these days with other things to do.

Thomas Stanka (usenet_10@stanka-web.de) wrote:
> ...
> What to you mean with "user gates"? 2input-NAND? 4 input LUT
> When an RH1020 is to small you need an RT54SX32S. But be aware,
> there's currently a relability problem when using the RT54SX-S
> devices.
> 
I mean 2 input-NAND.

Does anyone knows if Actel still manufactures obsolete and
legacy rad-hard (or rad-tolerant) parts like RT1460A or
RT54SX16 ? I think these parts would suit or needs.
I will ask Actel about this.

What reliability problem is this related to the RT54SX-S devices?
Does this reliability problem affects RTSX-SU devices? 
Actel told me that RT54SX-X and RTSX-SU are based in the
same design. The difference is that RT54SX-X parts are
manufactured by MEC and RT54SX-SU are produced at UMC.

> > ... availability of a similar or
> > equivalent, in System Programmable (ISP), part would be a plus.
> 
> When using the 5V IO, you will have trouble finding a reprogrammable
> equivalent.
> 

3.3V FPGA would be fine.

> > P.S: ASICs (Field Gate Arrays) would be OK, provided it cost less than
> > US$10,000.00 and takes less than a month or two to be fabricated. In
> > case we opt for an ASIC, only around 3-5 units would be produced.
> 
> I know no alternative when spending 30-50k$ for that number. AFAIK
> there are Actel fpgas above 10k$ per device.

It seems that an ASIC is indeed very expensive. Anyone has a value for the
cheapest rad-tolerant ASICs (around 1k 2input NAND-gates) ?

Best Regards,
Marcio.

"Marcio A. A. Fialho" wrote:
> 
> Does anyone knows if Actel still manufactures obsolete and
> legacy rad-hard (or rad-tolerant) parts like RT1460A or
> RT54SX16 ? I think these parts would suit or needs.
> I will ask Actel about this.

I know of an engineer at NASA who works with rad-hard parts.  He has
posted here in the past, but I have not heard from him lately.  From an
old post his email address is 

rk <stellare@NOSPAMPLEASE.erols.com>

Obviously remove the NOSPAMPLEASE

His name is Rick Katz, IIRC.  I was going to do a little work for him
once, but our schedules did not work out.  He seems like a very nice guy
and I am sure he would be happy to adivse you or point you to some
sources.  If he is swamped when you contact him, don't let that put you
off.  Just try again in a couple of weeks or so.  

-- 

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX

mitrusc1980-newsgroup@yahoo.com.br (Marcio A. A. Fialho) wrote
> Does anyone knows if Actel still manufactures obsolete and
> legacy rad-hard (or rad-tolerant) parts like RT1460A or
> RT54SX16 ? I think these parts would suit or needs.
> I will ask Actel about this.

At least there should be devices deliverabel. Never tried to get one
of them, but RH1020 are still in stock (RT1020 are out of order).

> Thomas Stanka (usenet_10@stanka-web.de) wrote:
> > ...
> > What to you mean with "user gates"? 2input-NAND? 4 input LUT
> > When an RH1020 is to small you need an RT54SX32S. But be aware,
> > there's currently a relability problem when using the RT54SX-S
> > devices.
> >  
> What reliability problem is this related to the RT54SX-S devices?

There's a problem due to the possibility of having weak fuses which
might increase the delay of a path in your design. This antifuse might
be stable for up to 2100h before showing this delay.
Try starting with klabs.org to get more details, as this problem is
very new, there are a lot of "uncertain" informations.

> Does this reliability problem affects RTSX-SU devices? 

No, only the availability of UMC-Parts *g*.

> Actel told me that RT54SX-X and RTSX-SU are based in the
> same design. The difference is that RT54SX-X parts are
> manufactured by MEC and RT54SX-SU are produced at UMC.

Yes thats right. The basic difference is that UMC-Parts use an other
fuse design. A second point is, that the inrush-current issue due to
power cycling is solved in UMC-parts.


> > > P.S: ASICs (Field Gate Arrays) would be OK, provided it cost less than
> > > US$10,000.00 and takes less than a month or two to be fabricated. In
> > > case we opt for an ASIC, only around 3-5 units would be produced.
> > 
> > I know no alternative when spending 30-50k$ for that number. AFAIK
> > there are Actel fpgas above 10k$ per device.
> 
> It seems that an ASIC is indeed very expensive. Anyone has a value for the
> cheapest rad-tolerant ASICs (around 1k 2input NAND-gates) ?

Well the price depends on many factors like quantity, flow and so on.
The cheapest device suitable for long term missions should be the
RH1020 with ~2000 NAND2-gates. The RT54SX32S device in B-Flow should
be quite cheaper, in E-Flow a bit more expensive than the RH1020.

bye Thomas

mitrusc1980-newsgroup@yahoo.com.br (Marcio A. A. Fialho) wrote in message news:<394351c1.0411251428.49ebc1ee@posting.google.com>...
> I'm looking for a rad-tolerant, non-volatile (preferably programmable
> at once) FPGA or CPLD, for a new project (a satellite instrument).
> 
> After searching the Web, I've found out that Actel manufactures micro
> antifuse FPGAs. These would be fine, but I would like to know if are
> there any other alternatives besides Actel FPGAs.
> 
> The device should have around 2500 user gates or more.
> 
> Reliability is a concern to us, and availability of a similar or
> equivalent, in System Programmable (ISP), part would be a plus.
> 
> Any single chip solution (preferably non-reprogrammable, to avoid
> unintentional program changes, due to radiation) would be fine,
> including CPLDs.
> 
> P.S: ASICs (Field Gate Arrays) would be OK, provided it cost less than
> US$10,000.00 and takes less than a month or two to be fabricated. In
> case we opt for an ASIC, only around 3-5 units would be produced.
> 
> If you are willing to help, you can e-mail me directly:
>   maaf _att_ dea.inpe.br    (replace the _att_ by an @)
> 
> Best Regards,
> Marcio
> ---------------------------------------------------
> Marcio Afonso Arimura Fialho
> Junior Electronic Engineer
> DEA - Divis&#4294967295;o de Eletr&#4294967295;nica Aeroespacial
> INPE - Instituto Nacional de Pesquisas Espaciais (Brazilian National
> Institute for Space Research).

I knew a Nasa engineer who also designed satellite instruments, he was
personally horified at the idea of FPGAs rushing into space, but I am
sure I heard that the mars missions had them anyway.

Search google groups for past posts on this.

Perhaps they get reconfigured regular to limit soft changes.

If it wasn't for the rad hard issue and your schedule I'd suggest
using something like the MOSIS or EUROPRACTICE shuttle services and do
a small ASIC. The cost sharing on an old technology device would give
you low cost in a few months and the performance would certainly match
FPGA. Ask them about RAD as well.

regards
johnjakson_usa_com

Jon,

Yes, the Mars Mission Landers' uses scrubbing once every 12 hours.  No 
redundancy. Turns out SEU's on Mars are not a big deal.  Not as bad as 
even here on earth!  12 V1000's per rover.  Controls wheels and some arm 
functions (or so I'm told).

The lander FPGA that fired the squibs for the parachute was a different 
matter:  fully TMR AND scrubbed continuously (reprograms while 
operating).  It also had three processors that were voting on the firing 
-- and if they failed to instruct, the FPGA would take over (or so I'm 
told).

Applications in space have outgrown the fuse technology FPGAs (they are 
just too small), and Xilinx FPGAs are being applied like crazy to space 
missions.

And our FPGAs have no 'power on surge', no funny long term fuse failure 
issues, and have well behaved IO's on power up and down (they stay 
tristate).


I am told that the next Mars lander walks around faster than a person 
does due to increased compute power from our newest FPGAs (rather than 
crawling about at a snail's pace).  That is how much more power there is 
from Virtex 1000 .22 micron to Virtex II Pro 130 nanometer 2VP100 
(roughly ten times the logic, at four times the speed, not to mention 
the 405PPC, etc.).

Our Mil/Aero/Automotive Group handles all the work for this area.

Did you know that under the hood of a car is a more hostile environment 
than low earth orbit? (totally random interjection)

Fully triple modular redundant designs are now trivial to do with the 
Xilinx XTMR tool.  Select those modules that you wish to triplicate, and 
after your design is done, tested, and verified, XTMR it, and do a final 
verification.

http://www.xilinx.com/products/milaero/tmr/index.htm

The remaining issue is single event functional interrupts (SEFI) which 
are upsets that reset the entire device.  The cross section for that 
single point failure is very tiny, but for space applications (and 
others) all single points of failure must be accounted for (even if they 
are a once in a million year event).

All devices (ASIC, FPGA, uC) will have some SEFI cross section, as there 
is at least one node somewhere that when hit will upset the whole design 
(like the reset line).

Contact one of our Mil/Aero/Automotive FAE experts to get all the details.

Austin

"Austin Lesea" <austin@xilinx.com> wrote in message
news:coq35v$af14@cliff.xsj.xilinx.com...
> Did you know that under the hood of a car is a more hostile environment
> than low earth orbit? (totally random interjection)
>
Depends if the engine's running or not and whether you're wearing a
spacesuit or not.

Actel has more space experience than any other FPGA.  If you want non-volatile,
and the features fit your application, then Actel is the way to go.  Rich Katz
is a
good resource.  Also, you'd do well to review the proceedings from the MAPLD
conferences.  That conference is heavy on space application of FPGAs.

For John Jackson: there are lots of FPGAs in space now.  Actel has been used in
space applications for over a decade.  Xilinx FPGAs are also being used in
space applications, including the Mars Rover.  I've done a few designs slated
for
low earth orbit.  Xilinx has a line of radiation tolerant devices that have
additional
process steps to guard against SEU induced latch-up.  Depending on the
application,
there are various degrees of fault tolerance/correction that can be applied.  It
all
depends on how big a problem it is if the particular circuit is upset.  There
are
many advantages to using FPGAs in space apps, especially reconfigurable ones
like the Xilinx parts.

Thomas Stanka wrote:

> mitrusc1980-newsgroup@yahoo.com.br (Marcio A. A. Fialho) wrote
> > Does anyone knows if Actel still manufactures obsolete and
> > legacy rad-hard (or rad-tolerant) parts like RT1460A or
> > RT54SX16 ? I think these parts would suit or needs.
> > I will ask Actel about this.
>
> At least there should be devices deliverabel. Never tried to get one
> of them, but RH1020 are still in stock (RT1020 are out of order).
>
> > Thomas Stanka (usenet_10@stanka-web.de) wrote:
> > > ...
> > > What to you mean with "user gates"? 2input-NAND? 4 input LUT
> > > When an RH1020 is to small you need an RT54SX32S. But be aware,
> > > there's currently a relability problem when using the RT54SX-S
> > > devices.
> > >
> > What reliability problem is this related to the RT54SX-S devices?
>
> There's a problem due to the possibility of having weak fuses which
> might increase the delay of a path in your design. This antifuse might
> be stable for up to 2100h before showing this delay.
> Try starting with klabs.org to get more details, as this problem is
> very new, there are a lot of "uncertain" informations.
>
> > Does this reliability problem affects RTSX-SU devices?
>
> No, only the availability of UMC-Parts *g*.
>
> > Actel told me that RT54SX-X and RTSX-SU are based in the
> > same design. The difference is that RT54SX-X parts are
> > manufactured by MEC and RT54SX-SU are produced at UMC.
>
> Yes thats right. The basic difference is that UMC-Parts use an other
> fuse design. A second point is, that the inrush-current issue due to
> power cycling is solved in UMC-parts.
>
> > > > P.S: ASICs (Field Gate Arrays) would be OK, provided it cost less than
> > > > US$10,000.00 and takes less than a month or two to be fabricated. In
> > > > case we opt for an ASIC, only around 3-5 units would be produced.
> > >
> > > I know no alternative when spending 30-50k$ for that number. AFAIK
> > > there are Actel fpgas above 10k$ per device.
> >
> > It seems that an ASIC is indeed very expensive. Anyone has a value for the
> > cheapest rad-tolerant ASICs (around 1k 2input NAND-gates) ?
>
> Well the price depends on many factors like quantity, flow and so on.
> The cheapest device suitable for long term missions should be the
> RH1020 with ~2000 NAND2-gates. The RT54SX32S device in B-Flow should
> be quite cheaper, in E-Flow a bit more expensive than the RH1020.
>
> bye Thomas

--
--Ray Andraka, P.E.
President, the Andraka Consulting Group, Inc.
401/884-7930     Fax 401/884-7950
email ray@andraka.com
http://www.andraka.com

 "They that give up essential liberty to obtain a little
  temporary safety deserve neither liberty nor safety."
                                          -Benjamin Franklin, 1759