On Mon, 14 Mar 2005 20:51:10 +0000, Jules Richardson Luckly I have experience with both ADSP2181 and a CPLD/FPGA. I built my own little board to read out the internal memory from an ADSP2181, designed signal generation board with ADSP2181, and wrote thousands of lines of ADSP2181 code. On the other hand, I built a QBUS MSCP SCSI with Xilinx CPLD and now working on an Altera CPLD. I didn't read much of floppy controller but in my preliminary opinion, CPLD/FPGA is a better fit. If you use a 2181 here, you utilize only the (synchronous) serial port which can be easily built with a CPLD. A CPLD on parallel port has another advantage, that you might be able to eliminate a microcontroller which I guess is needed for a USB sollution. I don't have knowledge of USB so please correct me if I am wrong. vax, 9000

Philip wrote: There must be one heck of a markup in the UK, then. Here in the US, you can buy a single 36-macrocell Xilinx XC9536XL for US$1.62, and a single 72-macrocell XC9572XL for US$3.09. Maybe you should order locally. These are easily programmed using zero-cost development software you can download from the Xilinx web site, and a simple parallel port cable you can build yourself easily. Eric

Ben wrote: Anybody that can't find $25 worth of stuff they want to buy from Digikey isn't very serious about hardware hacking. Why would you want to spend $100 plus $10/part for Atmel stuff, when you can the software for Xilinx CPLDs for free, build your own cable for under $10, and buy the chips for $1 to $3? Xilinx has a CPLD design kit for $50 that includes a board with two CPLDs, a JTAG programming cable, and the software and documentation on CD. http://www.xilinx.com/xlnx/xil_prodcat_product.jsp?title=design_kit Eric

The problem isn't finding $25 of stuff that I can use, it's finding another $25 when I've found I've forgotten to buy a 74LS00 or something like that.... Sure I can use the extra stuff I have to order later, but... -tony

der Mouse wrote: All free CAD software I have seen is windows based and getting more GUI all the time. With atmel software at least the software is small... most of the other stuff is much larger. While CUPL is crummy it does the job and fairly clear in just what is compiled. Ben alias woodelf

The software may be 'free', but the machine to run it on, and the OS to run it under, most certainly aren't. And I do have an objection to trusting my design to a piece of binary-only software that I have no easy way of checking if it's doing the right thing (it's very difficult to be sure a CPLD or FPGA is doing the right thing under _all_ conditions). I would mind if they'd properly document the chips so I could write my one CAD software if I wanted to. But they don't. AFAIK there is no 100% documented CPLD or FPGA available (100% documented meaning you can go programmer). Having hand-wired TTL running at over 50MHz and ECL at at least twice that, I'll stick to ways I can debug and that I know work. I've got enough TTL to last me for quite a time... -tony

Agreed. I didn't realise there was a linux version out now. I've only ever seen Windows versions... For the <nth> time, I have never been offered a Pentium machine. In fact I've never been offered a PC more powerful than a 286. On the other hand, all 4 of my PERQs, most of my PDP11s, most of my HP desktops, etc were free... No, but there's a big different. The only real use for a PDP11 C compiler is to program PDP11s. If you don't have a PDP11, you probably don't want a C compiler for it. This is not the case for FPGA development tools (you could want to program an FPGA without owning any specific type of computer). And another difference was that if I want to program in C, I have a large choice of possible machines. I could use DECUS C on a PDP11. I could use GCC on just about anything. I could even try writing my own compiler (the C standard is available, albeit most certainly not free, the programming specs for many computers are available). But if I want to use these FPGAs, I have to use the manufacturer's software on their choice of computers. Having done battle with older Xilinx chips (and hated every minute of it! [1]), I found the editor to be absolutely essential to undo the damage done by the rest of the software. If that's not included, then IMHO the package is worthless for serious work. [1] This was about 10 years ago, on a high-spec PC for the time. A compilation cycle took _overnight_. Which meant that if I made a small change to the design I couldn't test if until the next morning. I could have hand-wired the darn design quicker. But it's a darn sight easier if you can probe 'internal' signals... You're saying there's a CPLD that I can program without using a single piece of proprietary software or a proprietary programmmer (In other words, I can start from my schematic/logic equations, turn it into the programming data, by hand if need be, and then get that into the chip without having to use anything undocuemtned)? Can you please tell me some familes/companies? XC6200 (there were going to be at least 2 chips the XC6216 and the XC6264). I got the data sheets and connected the UK distributors. By the time they thought they could get me some 'sample quantity' chips, the darn things had been discontinued. That's why I never used them/ Sure. I am not saying I want to write the tools myself, but if the specs were available, I am sure some clever programmers would start writing open-sourcce versions (as has happened with so many other things!). -tony

Compensating effects... FPGA's getting more complicated at the same rate the (development system) processing capabilities are increasing??? Wonder how long with would take to gen the code for a 1986 FPGA on todays machine, or todays FPGA on a 1986 box........???

Tony wrote: [...] For a Spartan 3 FPGA (XC3S400), in less than five minutes I can compile a design that includes a 32-bit RISC processor core, a VGA controller, a PS/2 mouse controller, and an SDRAM interface. If I wanted to build the equivalent out of SSI/MSI chips, it would take me weeks if not months. Depends on the change. If it's simple, the times might be comparable. For a complex change, doing it on the FPGA is much faster. For instance, suppose I decide that I want to change the load and store byte instructions to be little-endian rather than big-endian. With an HDL and FPGA, that's a simple change to somewhere between one and four lines of code, depending on the design style. With a hand-wired design it's at least 32 signals to be rewired, assuming that the relevant data path isn't used for anything else. If the data path is shared, it might require the addition of more buffers and the control logic to drive them. IF I want to do things like experiment with adding extra pipeline stages, the FPGA is an even bigger win. Eric

It was thus said that the Great Tony Duell once stated: Just an observation, but this is reminding me so much of the arguments I've read about Assembly vs. C ... -spc (Or now, about C++ vs. Perl ... )

On Tue, 15 Mar 2005 16:30:39 -0800 (PST) "Eric Smith" <eric at brouhaha.com> wrote: I would suspect that while there might be hundreds of thousands of man hours, that there just can't possible be hundreds of thousands of man *years* involved. Hundreds of thousands of man years is just frightening to think about. Maybe something like the Apollo mission.

On Mon, 2005-03-14 at 23:31 +0000, Philip Pemberton wrote: You know way more about this side of it than me. ?10 - ?20 seems like a lot of money though, but as others said maybe us brits are being ripped off and buying overseas is a lot cheaper. Heh. Actually, soldering SMT stuff isn't too bad given the right iron and tip. So I'm game for anything there. Is that bidirectional? Just a random thought, but the parallel port interface is going to be an 8 bit bus and a handful of control wires presumably (R/W, 'start', 'clock' etc.) (clock for clocking data between buffer and parallel; board should have its own onboard clock for data between buffer and floppy) Anyway, my random thought was that an 8 bit bus and a few control lines is probably right for interfacing to a USB chip too. So design a parallel port version with space for a header on the board; if people want a USB version they just plug a second board into the first that taps into the parallel bus and control lines. Likely no more expensive than designing a USB version from the ground up, but you get a parallel port version for free (and at leass cost than a USB version because you don't need the plug-in board) Pretty much. See my other email to the list :) See above :) As Dwight says it's a bit harder for hard-sectored disks. I'd be tempted to say that v1 is soft-sectored only though as it'd be quicker to get something working as proof of concept; but v2 could follow on quickly with some extra logic to handle hard sectored media. But it can be done. And archival rather than decoding's the primary goal anyway, so people can mess around with that once the basic hardware's going. I'll definitely try hacking some code together to do decoding when it gets to that stage; it'd be an interesting challenge. Long way off (need working hardware first!) but if the catweasel *drivers* are open source, it might make sense to have a software layer that mimics the catweasel API and allows us to use catweasel drivers reading / writing there; our software layer handles stuff like sync sep (which I gather on the catweasel is done in the hardware) Go for it. If you happen to be local (I'm near Cambridge) I've got access to many different types of drives and media. But yep, starting with a single 3.5" drive (and assuming it's the *only* drive connected to the hardware) would seem a good place to start. cheers Jules

On 15/03/2005, at 10:31 AM, Philip Pemberton wrote: Well my preferred laptop (ie the one I spent my money on, rather than my work one) doesn't have a parallel port. Of course, I can always use a Linux box instead. Additionally, aren't there issues with "modern" Windows operating systems and parallel ports? Huw Davies | e-mail: Huw.Davies at kerberos.davies.net.au Melbourne | "If soccer was meant to be played in the Australia | air, the sky would be painted green"