Recommedations for GAL IC and Programmer

[martyn_hill]

Hi everyone

After some time trying in vain to program a cheap GAL (ATF16v8CZ) as some address decoding and glue logic for my latest QL expansion project, I thought I'd tap into the expertise of the forum to see what recommendations others had for choice of GAL IC and compatible Programmer.

Turns out the ATF IC I had chosen is not easily programmable without a more expensive programmer - I myself purchased a cheap 'MiniPro TL866A' so-called Universal Programmer that worked well enough for EPROM programming, but singularly failed to change a single bit (sic) of the ATF GAL IC I had purchased (from RS).

If it wasn't already obvious, I need an economical solution - like many here, my electronics efforts just a tad above simple tinkering at the weekend.

Any ideas?

[Paul]

I use same programmer.
Lattice GAL 16V8D work fine.
I buy them here:
http://stores.ebay.de/G-C-Supermarket-H ... 1090683909
HTH
Paul

[1024MAK]

As you have discovered, the various different makes of GAL are different. Both in the programming details and sometimes in the operation of the device.

Also GAL chips should be avoided if possible for new designs, as the main suppliers no longer make them.

In another forum, a member has not had very good results from a type claiming to be a Lattice 7.5nS type. The delay was in the order of 17nS and the outputs were very noisy. He did not say which eBay supplier he obtained them from.

Mark

[Paul]

Also GAL chips should be avoided if possible for new designs, as the main suppliers no longer make them.
Mark

Please take my excuses but I cannt agree here.
If hardware like gals are not recommended for new designs this is correct if you would like to produce millions of them. Talking about QL extensions (or other retro computer) number of about 50 are more likely to be found. So supplies are not really a problem.
Care should be taken if you like to try "new" logic parts in old computers.
Most new devices are not 5V tolerant. So you can't use them.
Even old parts like cplds that are still 5V tolerant are often too quick for our retro computers.
E.g. The QL design is very noisy. Where old parts like 200ns EPROMs and gals will work ok, new cpld's with fast flash ROMs see two or memory accesses where old hardware didn't even realise these short spikes.
So please don't use fast logic in old computers.
Kind regards
Paul

[prime]

Personaly I have a Wellon VP-590, was £370 so not budget, but does pretty much every chip I've needed to. I have a friend who has the VP-290 which is a more reasonable £98 and does do the Atmel PLDs.

Personally for most things I have switched to using Xilinx CPLDs as they are cheaper, more flexible and are programmable in circuit.

Cheers.

Phill.

[martyn_hill]

Thank you Paul, Mark and 'prime'!

I have read several warnings about using the older GALS - it does make sense.

In my case, DIP/through-hole packaging meant not having to 're-tool' (I use the term loosely - I only have vero-board, wire-wrap and soldering iron available!) or learn anything too complex for the task at hand, so it is reassuring to know that GALS could still offer an effective solution in this context.

Those modern SMD CPLDs looked to take me too far from my comfort zone!

I shall have a go with the Lattice variety and continue to use my cheapo TL866A programmer based on Pauls' experience (@Paul - I did scan the image on this forum of your latest memory card to see what GAL you used, but couldn't make-out the text!).

I''ll update this thread with my findings (once I get hold of a couple of Lattice 16v8's).

Meanwhile (sabotaging my own thread...) a couple of questions over typical GAL outputs...

a) In the ATF GAL documentation (haven't downloaded the Lattice version yet), there was conflicting data on output current - specifically SOURCING, which we know we need to drive (high) DSMCl when decoding our address-space and disable the 8301. At one point it suggests old-style TTL output drive (24mA SINK, 4mA SOURCE), yet in other places indicating symmetric CMOS type outputs (+-25mA).

Should we expect GALS to be more CMOS than TTL in output drive? Perhaps I should just read the Lattice manual!

b) I've seen some JEDEC-ready equations for older QL harwdare and note that its typical to use the OE/Tri-state extension on output pins to drive lines such as DTACKl and DSMCl - rather than, say a BAT42 diode (see Nasta's excellent tutorial in Illiont's previous RAM upgrade posting).

I also couldn't find anything other than a GAL and the SRAM chip in Paul's latest upgrade photo (i.el. no extra diode or similar), so assumed that this approach had been taken...

My question: if you enable OE/Tri-state for any one output pin, do you then need to prepare product-terms for ALL the output pins?

Also, is it always necessary to allocate an output pin for the OE (intermediate) result?

Thanks again for your replies, everyone!

[Paul]

I used Lattice16V8D for the ram expansion and gal equations from tcat:
As you see only DSMCL and DTACKL are tri stated.
And no diodes.

Code: Select all

GAL16V8
Glue logic for 512x8 SRAM 
Sinclair QL RAM Expansion
(c)2014 Sinclair QL forum

# control bits set by assembler
# SYN = 1
# AC0 = 1
# AC1(n) = 1,1,1,1,1,1,1,1

#1   2    3      4      5    6      7     8   9   10
 1   NC   A19    A18    NC   /DS    NC    NC  NC  GND
 0   NC   /RAMCE VALID  NC   /DTACK DSMCL NC  NC  VCC
#11  12   13     14     15   16     17    18  19  20

  # input 1 is wired to VCC
  # input 0 is wired to GND
  # NC - signal not connected
  # / - signal active on low level
  # / NOT, + OR, * AND operators

  # enables SRAM on VALID memory range and data strobe DS
  VALID = A19 * /A18 + /A19 * A18
  RAMCE = VALID * DS

  # start of our cycle DSMCL goes high, else tristate high impedance 
  IF( VALID ) 
  DSMCL = 1

  # end of our cycle DTACK goes high, else tristate high impedance 
  IF( VALID * DS ) 
  DTACK = 1

# END

[martyn_hill]

Thank you Paul!

That answers my questions

Out of interest, I don't see any inversion of RWl - how did you generate the OEl for reading from the SRAM (or did you tie that to Vcc and rely on WEl having precedence over OEl?)

Also, I see no tri-state -like statements in tcat's equations - perhaps you only showed an extract?

Anyways - now to purchase some Lattice 16v8s to play with - wish me luck!

[Dave]

I have been using 16V8 and 20V8 with no problems using a generic programmer. As long as they're not faster than approximately 60-70ns they seem to reliably work. Faster than that and the noise of the transitions is too much. One exception to this was a bunk of remarked parts from an ebay supplier in china that were fake.