WTD: info on AMD palce22v10

AMD spun-out its PLD lines as Vantis.

Before too long, Vantis was taken over by Lattice.

As far as I know, Lattice still offers these parts (or something very similar).

ISTR that the programming algorithm for PALCE devices was reasonably straightforward, so you could probably do something home-grown - maybe using some of your newly- found PALs in its logic :-) But in general, programmers for PLDs were expensive with good reason - getting the programming algorithms right is pretty tough. You may find it's better to junk the old parts and get some newer Lattice in-system=programmable (ISP) parts, which can be programmed much more simply and cheaply using a simple download cable that you can easily build from information in their data sheets.

--
Jonathan Bromley, Consultant

DOULOS - Developing Design Know-how
VHDL * Verilog * SystemC * Perl * Tcl/Tk * Verification * Project Services

Doulos Ltd. Church Hatch, 22 Market Place, Ringwood, Hampshire, BH24 1AW, UK
Tel: +44 (0)1425 471223                    mail: jonathan.bromley@doulos.com
Fax: +44 (0)1425 471573                           Web: http://www.doulos.com

The contents of this message may contain personal views which
are not the views of Doulos Ltd., unless specifically stated.

Is this algorithm documented somewhere ??

The data sheet I found does not mention any algorithm.

I have picked Atmel ATF1504 as a replacement device.

Thanks

Yeah, but can you get CPLDs in QFN packages??? Sometimes a "little" logic goes a long way. :)

Too bad we can't get any sort of FPGA in a physically small package. I could do a lot with a 100 LEs (or 50 slices) in a TQFP32/48 or a QFN32/48. The new tiny packages are really amazing.

I know I am not alone, but it seems we are not running with the herd.

Peter Alfke wrote:

All it takes is enough noise to convince the vendors to add the smaller packages - it's not a massive investment, mainly a mindset problem - 'We don't do that because no one buys them'. The SPLD/CPLD market is in 'follow mode', but the Microcontrollers and Logic have by now widely deployed QFN, and lattice have recently added QFN in 22V10, and TQFP48 in CPLD.

So the other, more sluggish, vendors will follow eventually.... Some will claim BGA is their small package solution, but they miss the point that cannot go onto single sided PCB.

Sweeping statements are dangerous.... We still use 16V8's for new designs, because of their price/size. Poor CPLD package offering is one problem slowing the replcaement of

22V10 : There are many sockets where your '300 year old' analogy has no more modern physical replacement. There are cost equivalents, but not physical equivalents, and I call that 'a little blinkered'

-jg

I too think that FPGA-s in small packages would be great things. I also think that this secrecy politics that programmable logic manufacturers use, has made FPGA very imaginationless component. Look at microcontrollers, how many different packages, architectures and so on. And hunderds of different manufacturers.

Returning to original question - i too are interested in programmer or programming algorithm for PALCE-s. GAL algorithms are available but these can't be used here.

regards, Raivo

That is a good point, that MCUs and FPGAs seem to be handled very differently in the market place. I have seen several startup FPGA companies look good and then fail. This even includes large companies like Motorola. They talked about an FPGA line using the "Pilkington" architecture and came very close to introduction before they shut it down. I never did hear anything about why they turned it off.

I expect that there is a bit more NRE and maintenance for FPGA products than there is for an MCU line. I can't really rationalize this other than to say that FPGA software seems to require constant upgrades while most MCU compilers are supported by a smaller team that mostly does minor upgrades and bug fixes. Am I off base on this? I would also say that they seem to make more changes to FPGAs when they introduce a new family than they typically do when they come out with a new MCU family member. Maybe that is it? MCUs normally have a new family member, while FPGAs get entirely new families.

Here is a partial list of major companies that introduced FPGAs, and then gave up:

Motorola (never got out of the starting gate, relied on external software...) Intel (sold it to Altera, who then canned it quietly) NSC (disappeared quietly) AMD (sold it to Lattice) ATT (sold it to Lattice) T.I. ( stopped being second source to Actel) Toshiba (never really made it) Cypress (?)

The moral of the story is that you survive as an FPGA manufacturer only when you are totally dedicated to that product line, which is true for Xilinx, Altera, Lattice, Actel, Quicklogic and some small start-ups. The Big Guys usually find it easier to make their money on other products.

Peter Alfke

Atmel .. still in it, but I still can't figure out why Dynachip .. got as far as producing silicon before dying Gatefield .. sold to was it Actel? Concurrent Logic .. became the Atmel/NSC architecture IBM .. got rights from the spoils of concurrent. Toyed with multi-chip module implementations, but never really made it out of the labs Not to mention numerous universities with academic architectures and the list goes on...

The fact of the matter is that devel> Ralph Malph wrote:

--

--Ray Andraka, P.E. President, the Andraka Consulting Group, Inc.

401/884-7930 Fax 401/884-7950 email snipped-for-privacy@andraka.com

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

Yes, and might yet hit critical mass - they service a niche with true one chip FPGAs (FLASH included)

Triscend anyone ? Both the Triscend, and Atmel FPSlic must struggle against the NIOS/MicroBlaze (etc) solutions.

that

Correct - Some stats were released last year (Altera?) showing they employed more SW engineers than HW designers.

-jg

I don't know that it was the software that killed this product. They were using NeoCad which was pretty successful (at least technically) for the other vendors they supported.

This line was second sourced by Cypress. I can only assume that Altera did not want to cooperate with Cypress and did not feel the need for second sources.

PLDs only AFAIK. Still being made. It was sold off not because it was failling, but because it was successful and AMD needed cash to keep the CPU business running.

ATT never wanted to be a significant player. They developed the ORCA line to get telecoms to migrate to their ASICs. Can't say how well that worked.

I don't remember then selling FPGAs or similar, they did CPLDs. TI did a complete makeover and decided that they were all about DSP in telecoms. They sold off everything that was not a key technology and bought a lot of stuff that was.

I never even heard of them.

Still there. I can't say how much market share they have, but they are selling product still.

You forgot Philips who's product line you bought (or at least part of it). But being solely a PLD vendor does not guaranty success. There are any number of PLD makers that have gone by the wayside, several of which were bought by Xilinx, some by Altera and others got swallowed up by the remaining players.

Heck, even the FPGA software companies are not immune to being swallowed up. The current Xilinx tools are just the latest version of NeoCad software I belive.

I think there is something else that is required to make it in the FPGA/PLD world. But I think the biggest factor is just that it is a limited market and unless you are getting 10%+ market share, it is not worth the commitment. So a shakeout to four or five players was inevitable.

Xilinx's purchase of Neocad was the nail in the coffin for Motorola's FPGA, but they were going nowhere fast before that.

AMD also had an FPGA, which they spun out under Vantis. IIRC, lattice has the rights to that now.

TI was a second source for Actel in the late 80's, early 90's.

--

--Ray Andraka, P.E. President, the Andraka Consulting Group, Inc.

401/884-7930 Fax 401/884-7950 email snipped-for-privacy@andraka.com

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

There is one untypical case in the history of FPGAs that no one mentioned so far. ATT sold aproximately ten years ago Xilinx compatible chips! Theyr datasheet said that these are pin-to-pin replacements. Those were marked with ATT prefix and were clones of XC3000 series chips. Afterwards Lucent Technologyes made also those chips. Does anyone know more about the story?

I have also heard that NEC was one of those who tryed to make FPGAs...

Raivo

I thought Peter mentioned ATT. ATT semi became Lucent who by then was making a new chip more like the 4000 family with Sync RAM in the LUT. Seems they bought rights to the various patents on the 4000 and decided to not do the clone thing, but rather to branch out and differentiate themselves. Their new product became known as ORCA and spawned three or so generations all of which was sold to Lattice. I used some of the 2C and 3C parts on a board I built. But, once again, it was the lack of good software or support that made me drop them. Their chip editor tool really sucked and the rest was not so good. So they went with NeoCad for their official tool. When NeoCad was bought by Xilinx all the current customers had rights to the source code and had to do their own maintenance and upgrades just like Xilinx was doing. I guess Xilinx figured this "leveled" the playing field. In reality, it leveled the competition.

One really good point that was lost when NeoCad was bought, was the fact that you could use one tool and target mulitple vendor's chips. Now we have HDL which allows the same thing if you buy each back end tool.

Xilinx selected ATT as a legitimate second source for XC3000 ( such things seemed to be important in those days), with the hope of speeding up the XC4000 project. When this did not work out and relationships soured, ATT came out with ORCA in competition with XC4000. All this later ended up with Lattice.

Another lesson: FPGAs do not transplant well from one manufacturer to the other ( see Altera-to-Cypress, Actel-to-T.I., Xilinx-to-MMI, before it got swallowed by AMD, Xilinx-to-ATT. The judgement is still out on AMD-to-Lattice, and ATT-to-Lattice )

Lots of blood >

As much as i know ATT never made their own software for Xilinx compatible chips. Why? Do they considered it too costly or got only rights to deal with silicon, not the soft?

It seems to me that in this case software problems were reasons why ATT droped this product line. As soon as Xilinx ceased to support 3000 series, ATT chips also became useless.

Interesting to know what happened first: ATT deciding that 3000 is too oldfashioned to make profit of it or Xilinx droping non-A series support and so making ATT chips useless?

Raivo

I don't think Xilinx "dropped" the XC3000 family, they still sell them today, AFAIK. They don't provide design software unless you really need it. I believe someone posted here recently that they needed this software and they got help to reach the right person at Xilinx for this.

But you may be partly right about them dropping support for the 3000 to cut ATT out of *new* Xilinx business. But then with *every* new generation, all the FPGA companies want you to design in the latest parts rather than the old ones. They know the competition also has new parts and the old ones won't get design wins... mostly.

In one case, Altera has a part that is still fully supported for new designs and has full 5 volt tolerance without having high startup current issues. I don't think Xilinx has any current parts like that. So I guess an old part could get a design win over even the latest parts since they no longer have 5 volt tolerance (they almost don't have 3 volt tolerance).