On 6/10/19 8:44 PM, Tony Duell via cctalk wrote: For purely combinatorial PLDs, see my blog entry on the subject over at vcfed.org; I did the work to clone a few PALs some years ago and documented the process. FWIW, the setup to do this was a few TTL ICs connected to the parallel port of a PC. Nowadays, I'd probably do the same with an inexpensive MCU--the programmable nature of MCU pins lends a certain amount of flexibility to the process. Basically, you separate the inputs from the outputs and then run all combinations of the inputs, observing the outputs. If the tristate feature is used on outputs, there's a way to discover the difference between a tristated pin and a genuine input. There are a number of tools to perform reduction on the results, such as Logic Friday. After that, you're left with a bunch of logic equations that can be fed into a PAL/GAL assembler and programmed. --Chuck

On 2019-06-11 6:17 a.m., Paul Birkel via cctalk wrote: The process documented above is essentially the process I started last night, but in my case I used a GPIO in my HP 9000-332 to cycle through the inputs and record the output.? It is very handy to have general purpose parallel I/O for purposes like this.? In this case the number of possible states is reduced as two of the inputs are permanently tied high.? I am well on my well to developing logic equations to feed into palasm to generate a new JEDEC file which I can then burn into a GAL and test to see if it is correct. Paul.

On 6/11/19 10:54 AM, Tony Duell via cctalk wrote: I mentioned in my blog entry that I had a way of detecting this by driving the outputs through a series resistor and sensing the logic level of the pin. In other words, you've already labeled the pin as an output by exhaustively running through input combinations. What's needed is to differentiate a "driven" high level from a tristated one. By driving the outputs low through a series resistance you can determine if an output is high-impedance or not. I explained that badly, but it does work. --Chuck

On 6/11/19 11:40 AM, Paul Berger via cctalk wrote: That's nice; but in my case, I was sent a couple of PALs to clone without a hint of their origin. So the PAL was essentially a black box and the task was to come up with a box that behaved similarly. --Chuck

I realize that most are familiar with using the equations to create the PALs but I'm a circuit person more than an equation person for circuits. When I needed to create a PAL from a schematic, I first made a schematic of what the PAL was suppose to do, using the same basic model of logic that the PAL provided. Once I was done, I took the PAL map from the TI book and made red dots on each of the connections I needed. I'd then go back and add the don't care connections. You know, A*A!. I transposed them to a text file in the JEDEC format ( One needs to add on non-text character as I recall ) I then gave the file to my friend that had a PAL programmer and made the PAL. I made a simple ruler to convert dot locations to column offsets. While, Most seem to like the algebraic formulas, I find it harder to check than a schematic. I had one value that I wasn't sure about that require an experimental PAL blown, as the circuit didn't show if the bank select was 0 or 1 at reset. I've used this method on both the recreation of Jef Raskin's Swyft board and the PAL I needed for my 6532 to 6530 KIM-1 fix ( the KIM took a couple more PALs because as a human, I made some logic mistakes ). Copying a L type PAL is a little simpler but in the case of the Swyft board it was a R type PAL. Doing it from the schematic, and know what it was suppose to do was much simpler than trying every possible combination of clock and data from the pins of a PAL ( that I didn't have anyway ). Dwight ________________________________ From: cctalk <cctalk-bounces at classiccmp.org> on behalf of Paul Berger via cctalk <cctalk at classiccmp.org> Sent: Tuesday, June 11, 2019 5:54 AM To: cctalk at classiccmp.org Subject: Re: HP9816 PAL16L8 On 2019-06-11 1:19 a.m., Chuck Guzis via cctalk wrote: Old computer work too, I used a GPIO in a HP 9000-332. Years ago I also created general purpose I/O ports for a PC using 6821 chips since they are way more flexible that the Intel 8255, but the HP 9000 coupled with RMB makes for a great environment to bang out quick programs to do things like this. Paul.