What to do with those old XC3S1200E chips (o;

[davorin]

Good afternoon (o;

Someone got a good idea what to do with those?

My employer ordered them a looong time ago, though they didn't made it into any project...
Well at least I can still use ISE 14.7 for them..though not sure what the limits are by means of what max. size parallel RGB TFT I can drive and so on...
though it would be a good excersize for PCB design and soldering in my pizza smt oven (o;

[RoGeorge]

Not sure how reliable a DIY oven can solder a 300 balls BGA stored for years.

That's a few thousands worth of chips.  Try to sell them.

[davorin]

Well already tried to sell them...no luck...not sure sure anyone wants to design something with an obsolete and unsupported IDE.

OTOH no costs for me frying those buggers (o;
Guess I'll do a simple board based on the butterfly one 3e design....at least a m68k core fits easily for some retro feeling ;-)

Or digging up my old pacman/galaxian files...

[asmi]

Not sure how reliable a DIY oven can solder a 300 balls BGA stored for years.

There should not be any problems soldering them. Just bake them before reflow, and you are good to go.

[davorin]

Bake?

For sure I would need leaded solderpaste....

[asmi]

Bake?

For sure I would need leaded solderpaste....

Judging by the marking, these chips have lead-free balls, so I would recommend either to use lead-free solder paste, or reball them with leaded balls before reflow.

[RoGeorge]

The Xilinx ISE IDE is still active, can be downloaded from Xilinx.  Finding licenses for Xilinx is not a problem either.



Old chips tend to absorb moist with time, so when put in the oven, the steam formed inside the chip enclosing material will expand, and thus might crack or pop the chip, similar with pop-corn.   

To avoid that, old chips are "baked", meaning slow heat before trying to solder them.

Google for the exact procedure of baking temperature and duration details.

[davorin]

Ah okay....good to know about this baking procedure....

Well I've installed ISE 14.7 just recently...runs fine on Debian 11.1..except fpga-editor requires old XMotif libs and some tweaking ;-)

[james_s]

Those are nice parts, I still have some smaller Spartan3 boards that I use. I wonder if there's already an existing board design they would fit? That would make a really nice dev board. The problem with a part like this is that it is well beyond the typical DIY level.

[asmi]

The problem with a part like this is that it is well beyond the typical DIY level.

I strongly disagree. This part is small enough to solder it even with just a hot air gun and some tacky flux, not to mention that I lot of hobbyists nowadays have reflow ovens (either homemade, or some cheap stuff from the East).

[james_s]

There's more to it than just soldering, laying out a part like that requires quite a complex PCB, and I think that while there certainly are hobbyists who are soldering large BGA packages, I think it is far from typical.

[MadTux]

Or take a digilent/similar dev board as inspiration, discard the peripheries and instead route the pins to external (2/3/4 row) pin connectors, where you could plug in you own design/circuitry
Similar to these boards, but perhaps with a little bit more pin fanout, even has schematics ;-)
https://pldkit.com/xilinx/xmf3
Some DRAM on board perhaps would be a nice option, too.

Or perhaps ask the guys at pldkit, if they want these 40 FPGAs and you get like 15 assembled boards in return?


BTW, how did you get Xilinx FPGA editor working, I'm stuck with that Wind/U crap:

Code: [Select]

Wind/U Error (193): X-Resource: DefaultGUIFontSpec (-*-helvetica-medium-r-normal-*-14-*) does not fully specify a font set for this locale
Cannot register service: RPC: Unable to receive; errno = Connection refused
unable to register (registryProg, registryVers, tcp)
Wind/U Error (248): Failed to connect to the registry on server arch64-fpga2

Warning!!: XKEYSYMDB environment variable is set to a wrong location
Cannot register service: RPC: Unable to receive; errno = Connection refused
What I did to proceed to Wind/U error:
/opt/Xilinx/14.7/ISE_DS/ISE/bin/lin64/_fpga_editor: error while loading shared libraries: libXm.so.3:
sudo pacman -S openmotif
cd /usr/lib
sudo ln -s libXm.so.4 libXm.so.3

/opt/Xilinx/14.7/ISE_DS/ISE/bin/lin64/_fpga_editor: error while loading shared libraries: libstdc++.so.5:
sudo pacman -S libstdc++5

Wind/U error:
sudo pacman -S xorg-fonts-100dpi xorg-fonts-75dpi
doesn't fix it.... :-/

[Someone]

Not sure how reliable a DIY oven can solder a 300 balls BGA stored for years.

Not just age, check out the dust caked on them!

[jmelson]

Well already tried to sell them...no luck...not sure sure anyone wants to design something with an obsolete and unsupported IDE.

I am still making boards with Spartan 3A and 3AN chips, but the smaller end of the product range, and TQFP144 package.
Jon

[Bassman59]

I suppose that you could use them for a personal project ... of course, you need a good idea first. And there's the rub, as the bard said.

[asmi]

There's more to it than just soldering, laying out a part like that requires quite a complex PCB, and I think that while there certainly are hobbyists who are soldering large BGA packages, I think it is far from typical.

if you don't want some kind of high-speed stuff (DDRx, PCIE, whatever), you don't need a complex PCB. FT256 can be fully routed out on a cheap-ass 4 layer process offered by JLCPCB. These boards are within the reach of just about any hobbyist, unless by a "typical" hobbyist you mean somebody stuck in the past millennium with through-hole parts and other obsolete crap like that.

[james_s]

if you don't want some kind of high-speed stuff (DDRx, PCIE, whatever), you don't need a complex PCB. FT256 can be fully routed out on a cheap-ass 4 layer process offered by JLCPCB. These boards are within the reach of just about any hobbyist, unless by a "typical" hobbyist you mean somebody stuck in the past millennium with through-hole parts and other obsolete crap like that.

So probably 80% of the hobbyists out there? Vast numbers of people still work mostly with through hole parts, or with big chunky SMT stuff. A lot of them are older folks who simply lack the visual acuity or manual dexterity to work with tiny parts.

I mostly abandoned through-hole years ago and have built plenty of things around 0.5mm pitch ICs, but 4 layer boards only recently became affordable and I have not yet tried having one made personally. I have little doubt that I could succeed in doing so, but I have not yet had an opportunity to try and thinking of the hobbyists I know, I don't think any of them have built anything with a BGA part. I think you are exhibiting a bit of the Dunning-Kruger effect and grossly overestimating the average skill level of electronics hobbyists. There are of course some that would have no problem working with these parts, but those who could design, lay out and assemble a suitable PCB are probably in the 1 percentile. A handful more could probably succeed in assembling a PCB that someone else designed with a reasonable yield if given access to suitable tools. It would be interesting to see some actual data on this.

[asmi]

So probably 80% of the hobbyists out there? Vast numbers of people still work mostly with through hole parts

I doubt that, and suspect that you are projecting here. But I would love to see some real numbers.

I mostly abandoned through-hole years ago and have built plenty of things around 0.5mm pitch ICs, but 4 layer boards only recently became affordable and I have not yet tried having one made personally. I have little doubt that I could succeed in doing so, but I have not yet had an opportunity to try and thinking of the hobbyists I know, I don't think any of them have built anything with a BGA part. I think you are exhibiting a bit of the Dunning-Kruger effect and grossly overestimating the average skill level of electronics hobbyists. There are of course some that would have no problem working with these parts, but those who could design, lay out and assemble a suitable PCB are probably in the 1 percentile. A handful more could probably succeed in assembling a PCB that someone else designed with a reasonable yield if given access to suitable tools. It would be interesting to see some actual data on this.

I was just a hobbyist mere 7 years ago working with TH parts only. About a year into the hobby I assembled my very first SMT board, ~a year later I built my first FPGA board (ice40 ultra in QFN48 package), few months later I designed and built my first FPGA board with Xilinx 7 series FPGA and BGA parts (there were 4 of them on that board - main FPGA, 2 64Mb HyperRAM chips and a QSPI flash chip), about half a year since that I made my first FPGA board with DDR3L device on a board and some 5Gbps differential transmission lines.

All of the above does not require massive financial investments nowadays (current chipnageddon notwithstanding, but that will hopefully pass soon-ish), but rather commitment, dedication and willingness to learn. I needed to commit some more significant financial resources because PCB tech wasn't there at the hobbyist's price point, but now it's there and is super cheap. So I fully expect that there are quite a bit of people who can follow the path similar to mine. They are the target audience for my beginner's Spartan-7 devboard which I posted on Github for anyone to grab and modify for their needs.

4 layer boards are actually easier to route than 2 layer ones because you don't have to worry about routing power and ground connections - you just drop a via and that's it! This allows for a more dense layout too, meaning you can place your components closer, which helps with miniaturization (and saving on PCB cost).

[james_s]

So probably 80% of the hobbyists out there? Vast numbers of people still work mostly with through hole parts

I doubt that, and suspect that you are projecting here. But I would love to see some real numbers.

I mostly abandoned through-hole years ago and have built plenty of things around 0.5mm pitch ICs, but 4 layer boards only recently became affordable and I have not yet tried having one made personally. I have little doubt that I could succeed in doing so, but I have not yet had an opportunity to try and thinking of the hobbyists I know, I don't think any of them have built anything with a BGA part. I think you are exhibiting a bit of the Dunning-Kruger effect and grossly overestimating the average skill level of electronics hobbyists. There are of course some that would have no problem working with these parts, but those who could design, lay out and assemble a suitable PCB are probably in the 1 percentile. A handful more could probably succeed in assembling a PCB that someone else designed with a reasonable yield if given access to suitable tools. It would be interesting to see some actual data on this.

I was just a hobbyist mere 7 years ago working with TH parts only. About a year into the hobby I assembled my very first SMT board, ~a year later I built my first FPGA board (ice40 ultra in QFN48 package), few months later I designed and built my first FPGA board with Xilinx 7 series FPGA and BGA parts (there were 4 of them on that board - main FPGA, 2 64Mb HyperRAM chips and a QSPI flash chip), about half a year since that I made my first FPGA board with DDR3L device on a board and some 5Gbps differential transmission lines.

All of the above does not require massive financial investments nowadays (current chipnageddon notwithstanding, but that will hopefully pass soon-ish), but rather commitment, dedication and willingness to learn. I needed to commit some more significant financial resources because PCB tech wasn't there at the hobbyist's price point, but now it's there and is super cheap. So I fully expect that there are quite a bit of people who can follow the path similar to mine. They are the target audience for my beginner's Spartan-7 devboard which I posted on Github for anyone to grab and modify for their needs.

4 layer boards are actually easier to route than 2 layer ones because you don't have to worry about routing power and ground connections - you just drop a via and that's it! This allows for a more dense layout too, meaning you can place your components closer, which helps with miniaturization (and saving on PCB cost).

How am I projecting? I switched mostly to surface mount at least a decade ago, I'm not sure you understand what "projecting" means.

Look I'm just explaining what I've seen, I personally know exactly zero hobbyists who have built anything with BGA parts, I know they exist but I don't know any and judging by what I've seen posted on this forum I think the number is very small. Remember I'm talking about *hobbyists* here, not professional engineers. Just look at what is available from companies that sell to hobbyists, vast quantities of premade modules that get connected together with wires. Scour the net looking for DIY projects, I can't think of any off hand that I've seen that use BGA parts though I know they must exist somewhere. Most kits are still through-hole, maybe try posting a survey on this forum and see how people respond? Clearly you are an exceptionally skilled hobbyist, that's great, but recognize that you are the rare exception and not the rule, this is obvious just from a cursory glance around the internet.

[MadTux]

It's also the question about prototyping. Chance of designing a BGA board for some application and have it working on first iteration is rather low, this is why I rather put BGA/TQFPs on breakout adapters and wire them on breadboard/deadbug style or use classic THT components.
If everything works, you either design a PCB for it and use SMD/SOIC parts instead of the THT at the breadboard (reduces size and makes soldering more easy) or simply use the finished wired prototype with BGA breakout parts if you only need one working device....

[NorthGuy]

Well already tried to sell them...no luck...not sure sure anyone wants to design something with an obsolete and unsupported IDE.

With all the shortages out there you still cannot sell them?

[james_s]

Probably not a lot of companies producing products want to buy a small batch from some random person. This is the sort of thing a chip broker might want though. Once had someone give me quite a lot of money for a few reels of oddball parts I struggled to find any use for.

[davorin]

Good day (o;

Was busy reading datasheets and app notes lately....and looked at some old schematics (o;
And also the schmeatics symbols is done now in KiCAD....as the one from octopart throwed errors...

If I understand the section right about JTAG programming....I can set the programming mode to always be SPI serial flash mode but still can use JTAG for reprogramming the FPGAs SRAM or even the attached SPI flash...right?

From DS212:

JTAG port that is always available any time the FPGA is powered and regardless of the mode pin settings.

[james_s]

There are multiple ways of programming one but generally yes you use JTAG both for loading the bitstream directly during development and for programming the configuration EEPROM that is hooked up externally. You'll need a Xilinx platform USB cable, the Chinese clones work fine.

[davorin]

Excellent...makes layout easier ;-)

A USB cable clone is already on the way...but just out if curiosity..would a Digilent JTAG-HS2 cable also work for flashing the SPI flash....

I assume iMpact uses JTAG boundary scan to toggle the flash lines accordingly....and not via a preloaded design as it is used with newer FPGAs...