Ferranti ULAs

[dominicbeesley]

Hello,

I've been playing around with CPLDs and FPGAs recently and am currently (trying to) making a 65816 board for a BBC Micro. I'd like to use 1980's technology but of course I can't pick up the phone to Ferranti and ask for them to make me a ULA. What I'd like to do is emulate that in either a CPLD or FPGA but I have no idea of what the capabilities of the Ferranti ULAs was and there seems to be very little information about the internals on the net. Does anyone have any information?

Thanks

Dom

Even a weeny FPGA will knock the socks off a BBC ULA, and the dev. kits are quite cheap.

[dsergeant]

I worked at Ferranti many years ago and in one of my projects designed a ULA. It is all individual transistors and other components and you design the mask to interconnect them to produce the circuit you want. I am not sure you could easily do the same function in a modern FPGA, they are totally different things.

It was great fun at the time, drawing the mask on large transparent sheets with red and green strips and took ages. As merlinmaxwell says, a modern FPGA can do infinitely more in a much smaller chip...

[dominicbeesley]

Thanks lads,

I know that even the smallest FPGA could exceed the number of possible gates on a ULA. It's more that I'd like to keep my design to something that _could_ have been realised in a ULA - I know the design principles would be quite different but I'd be interested in how many transistors/gates/interconnects were present in the various Ferranti ULA's to get an idea of the possible complexity.

I think I might buy this book http://www.zxdesign.info/book/ out of interest...

Currently I have kind of got a design working. It's erratic and I'm not sure if its due to my misunderstanding the chip timings, decoupling, too long connections to the motherboard or a wiring mistake...it will run a simple test program for between 1 and 100 seconds before going crazy.

I've just received a 500MS/s logic analyzer from China which I'm hoping will help but it seems to pick up a lot of noise which might indicate the problem!

[acollins22]

I saw a talk by Chris Smith a couple of years ago. He certainly seemed to know his stuff. I see on his site that he's built FPGA stuff. When I get chance I'll work me way through.


Cheers,

Andy.

[GrahamN]

I can recommend the Spectrum ULA book - I bought a copy a couple of months ago, and it seems very comprehensive (though I'm only about 1/3 way through at the moment).

I've also bought a FPGA dev kit, but probably won't start looking at it seriously until late this year or possibly even next year the way things are going! I intend to try and build a circuit with standard 74LS chips first, then see if I can emulate that with the FPGA. The main idea is to try and emulate a Sinclair ULA as a plug-in replacement - something that seems to have been started by the author of the book, but for some reason never actually completed - presumably because it was actually easier to emulate the whole computer with the FPGA.

[dominicbeesley]

I'm looking forward to it arriving. I've still spent a bit of time looking and not found any information on the "capacity" of the Ferranti ULAs so I'll be interested to see what the book says. It looks like the xc9500 CPLDs I've been using are probably less capable than the bigger ULAs...but thats a guess!

Good luck with the FPGA stuff Graham. I've managed to get a BBC micro working in an DE0 nano. I did start with someone else's project - though I've made a lot of changes. Once you get your head around the various components of VHDL/Verilog work (I stuck to VHDL) it quickly makes sense.

What is a struggle are the absolutely dog awful development tools - they all seem to be based on the same set of ancient, badly thought out, buggy tools with a veneer of counter-intuitive tcl script and a java user interface slapped on which is full of excellent features such as schematic editors and the like that don't actually work!

If you stick to the basics and get your head around the timing thing (you DO need to worry about timing analysis on anything complex) then just have a lot of patience. I've run into so many problems that I thought were down to me just to turn out to be a bug or undocumented feature in the software.

When you get things working it is satisfying though and building complex systems relatively simple. There are lots of VHDL cores on the net too which can save you time or act as a tutorial.

I'm meant to be learning about Sharepoint for work, I have 3 500 page books to avoid reading so I'll no doubt read the Spectrum ULA book pretty sharpish!

D

I find Alteras free tool not too bad (worth getting used to to save thousands of pounds). All these devices have loads of pins and they all need defining, a bit of a pain but I can't think of a better way, I use a combination of the 'logic circuit' design tool and VHDL/AHDL, works for me.

[woodchips]

Just come across a Feranti short form catalogue from 1978 with a little info on the ULAs.

Each cell had 3 NPN transistors, 2 of load resistors and 3 of input resistors. Resistor value changed depending on the array speed variant. Each I/O pad had 2 of pull up resistors and an uncommitted NPN transistor.