This repository centers around documenting modern ways of developing logic for and programming Atmel (Now Microchip) 5V GAL PLD and CPLD parts under recent Linux (Ubuntu 22.04) and Windows 10 22H2 versions.
(Modern/active equivalent of the PAL16V8 and GAL16V8 parts)
(Modern/active equivalent of the PAL22V10 and GAL22V10 parts)
(Active replacement for the EPM7032)
(Active replacement for the EPM7064)
(Active replacement for the EPM7128)
These parts are still active and highly worth considering wherever:
- 5V logic is a requirement, avoiding level shifting, low latency (7ns), instant-on & non-volatile, and situations needing Hi-Z / Open-collector states.
- Prototyping / Iteration (reprogrammable)
- Learning about logic: Through-hole / soldering-friendly is desired: All 16V8 or 22V10 parts are available in DIP packages; ATF150x parts are available in PLCC packages that can be placed in through-hole PLCC sockets. SMD packages are available for any of the parts.
- Replacing large quantities of various TTL/CMOS Logic Gates (74-series logic)
In short, these chips work very well wherever the above requirements are necessary, however, the software (and device programming) experience can be incredibly challenging owing to outdated and buggy software.
This repository aims to make it easier to work with these parts and hopefully keep them active for years to come.
This is a "Choose your own adventure novel". Covered here are many approaches and tradeoffs:
- Using the WinCUPL IDE (Erratic, unreliable)
- Using just the CUPL.EXE compiler from WinCUPL directly with some wrapper scripts here (5vcomp). Works in Windows/Linux. (recommended)
- Using Quartus (only for the CPLD. Works by first targeting a similar Altera CPLD and then using the POF2JED utility to convert.) Windows/Linux
- Making your own fusemap / .JED file with nothing more than a datasheet and text editor. Maybe need some graph paper...
- Experimental approaches with Yosys (Only for the CPLD parts. EDIF is fed into the Atmel fitter)
- Several Approaches to reverse-engineering a .JED file back into logic equations.
Scope: Expand here for why similar parts not covered
- The intention is primarily to make it easier to work with parts that are still active so they do not go NRND and eventually disappear from the market. While there are gems here for other related historical parts, this is not the focus of this repository.
- The ATF1500 is not covered because it is a more expensive part and does not support JTAG programming. It is fundamentally different from the ATF1502, ATF1504, and ATF1508
- The ATF750 and ATF2500 are also not covered for similar reasons. Other chips are almost certainly a better choice.
- We only consider true 5V parts (not merely parts with 5V tolerant inputs, of which there are many more).
- 3.3V parts cannot supply the minimum of 3.6V to drive the input of a 5V CMOS part high, so 5V tolerant parts are not enough in many cases. Even the parts covered here may not necessarily be capable of the required VoH as their output voltage drops off quickly under load. In these cases, pullup resistors can be considered.1
- Notably, however, driving 5V TTL inputs from a 3.3V part, on the other hand, is not a problem. A 5V TTL input has a threshold of 2V.
- 3.3V parts are not considered: There are simply better choices that are well documented. Also, the CPLD parts have VccIO inputs, so you can technically use them in 3.3V designs just as well.
- The following 5V IO capable parts probably should be covered, but they're already well supported, documented, modern tools, etc.
- The Greekpak devices
- The Cypress PSoC5LP (An ARM Cortex M3 with CPLD-like logic blocks), available in 68-pin QFN, 100-pin TQFP
- Any parts that are NRND or inactive are not covered, as we consider what can be reliably and sensibly purchased.
- Since all of the parts considered here are still in full production (as of 2023), they can be used in production designs.
- For the ATF150x CPLD parts specifically:
- The parts ending in 'BE' are not covered here as they seem to be very expensive. In principle these are interesting because they have multiple IO bank voltages.
- The parts ending in 'ASV' are 3.3V only. The 'AS' devices can be operated at 3.3V IO through the VccIO pins. The 44-pin devices to not show a VccIO pin, but it does seem to be the case that two pins are Vcc and two are VccIO.
- For applications where 5-Volt tolerant operation is acceptable, it might be worth considering the ispMACH4000 series. Parts such as the LC4032ZE are available is a somewhat soldering friendly TQFP, however there are likely similar challenges with software.
This repository isn't intended to be an introduction to digital logic, but a brief review and compare/contrast to similar things is provided here.
Expand here for tutorials on Digital Logic
Ben Eater does a series of Videos on Digital Logic that are a really excellent introduction to some of the concepts here.Expand here for a description of how these parts compare to ladder logic on a PLC
- Each rung's output in ladder-logic can be thought of as a single macrocell.
- The inputs on a rung can be "normally open" or "normally closed" (active high or low), and can consist of any number of inputs (or even the state of another macrocell). The inputs defined on a single rung are basically equivalent to a single product-term belonging to a macrocell. There can be multiple product terms defined that activate a given macrocell.
Expand here for details on how all of these compare to FPGAs
Such parts are the spiritual predecessors of more modern FPGAs. Key differences between FPGAs and PLDs:- FPGAs are typically constructed from a large number of LUTs (Lookup tables). CPLDs use a sum-of-products structure.
- FPGAs typically expect to have their bitstream uploaded on powerup, requiring an external EEPROM. PLDs are typically non-volatile and instantly ready upon powerup.
- FPGAs usually support more standard means of programming, whereas many PLDs required specialized device programmers.
- There are likely exceptions to all of the above in some parts. These are not hard rules.
A high-level overview of what is required:
- Basic understanding of digital logic.
- The actual PLD/CPLD chip you'd like to work with from the usual suppliers (Mouser, Digikey, Octopart)
- A software workflow covered here. Highly recommended is using the 5vcomp script from here to call the CUPL.EXE compiler. This works on Linux and even Windows 10 22H2 x64!
- An EPROM/Device programmer if you wish to use the ATF16V8 or ATF22V10 parts.
- An EPROM or JTAG programmer for the ATF150x parts
for details on what it takes to program these parts in detail.
PLD/GAL - Programmable Logic Device. Small, generally DIP-package 5V programmable Logic.
CPLD - Complex Programmable Logic Device. Larger packages, many pins, much more complex.
5vcomp - A utility in this repository (batch file for Windows / shell script for linux) that is a wrapper around the CUPL.EXE compiler
Combinatorial Logic - Simple logic (AND, OR, NOT, gates, etc.) that does not use flip-flops / registers / clocks. Such logic could technically be implemented with an EPROM/Memory, where a series of inputs always maps to a known set of outputs.
Registered Logic - Logic that uses registers (flip-flops), and can thus hold state. On the GAL16V8 and GAL22V10, each macrocell can be configured as a D-Flip-Flop, and all flip-flops share the same clock pin. On the ATF150x, much more complex types of registered logic and clocking options are available.
Macrocell - Each output has a macrocell associated with it. These can often be configured as active high, active low, flip-flops, etc.
Product Term - Each macrocell has a number of product terms associated with it (typically around 5). A product term is essentially a giant AND gate with inputs to each pin on the device. Burning away fuses allows selecting which inputs are fed into this AND gate, ultimately selecting the conditions required for a product term to be activated. Product terms belonging to the same output macrocell are then combined into an OR gate before being fed into the macrocell. This means that there can be several combination of inputs that allow a given macrocell to be triggered. This architecture is called a Sum-of-Products logic array.
CUPL - A early (1983) programming language used to define the behavior of programmable digital logic. "Compiler for Universal Programmable Logic.", is essentially a predecessor to languages like Verilog/VHDL. CUPL.EXE is the compiler which is used to compile .PLD files written in CUPL, ultimately to be burned into programmable logic devices.
WinCUPL - A Windows front-end/IDE to the CUPL compiler and related programs. WinCUPL itself is best avoided but installing it is necessary to get the CUPL compiler and device libraries.
.dl File - A Library File is a binary file used by CUPL compiler that provides support for devices CUPL has the ability to compile logic for. This should not be confused with the Device/Primitive Libraries that are part of the fitter.
Netlist - A netlist is essentially an electrical schematic in a text file which defines connections. For the purposes here, it is an intermediary file format (Either EDIF or Berkeley PLA), which is used to describe the behavior of logic ultimately fed into the fitter.
.TT2 - The Berkeley PLA file format. An intermediary file which CUPL.EXE can generate that can be used by the Atmel fitters. Notably, one can use berkeley-abc to work with these files.
.EDF / .EDN - EDIF is another type of netlist format. The Atmel fitter can use this as both an input, as well as an output. Yosys is capable of generating this format, however, one will still need a techmap for this to work.
Fitter - A fitter converts a netlist into the fusemap (.JED) file. It is specific to the device in question and provided by the device manufacturer (Atmel in this case). Fitters are needed only for the ATF150x CPLD devices, whereas the PLD devices can go straight from a .PLD to a .JED file. In more modern parlance the fitter is basically the place & route stage.
ATMEL.STD File - Part of the Atmel ATF150x fitter, the primitive/device library for PLA.2
APRIM.LIB File - Part of the Atmel ATF150x fitter, the primitive/device library for EDIF.2
.JED/JEDEC File - A fuse map intended to be "burned/programmed" into a logic device. A JEDEC file is ultimately a text file formatted specifically to the JESD3 standard. If you have a device programmer that has support for the exact device you are interested in programming, this file is all that is needed.
backpin.exe - Utility which can take the pin mappings that have been autoassigned by the fitter and backannotate them into the .PLD file.
.SVF File - Serial Vector Format. Generated from the .JED file, the .SVF can be used by any JTAG programmer (vendor-independent) to program a device that has a JTAG interface (So, the ATF150x CPLDs). The Windows ATMISP v6.7 or ATMISP v7 tools can be used to generate an .SVF from a .JED file, as well as the fuseconv.py utility by whitequark. Once you have an .SVF, you can use tools like OpenOCD or the experimental branches of Afterburner to program a CPLD with a JTAG interface.
CSIM.EXE - Part of WinCUPL. A tool for simulating the behavior of logic. This takes an .SI file (test vectors) and an .ABS file. Given these it produces an .SO file. Only provides functional simulation (so, logic states but not timing)
Wine - Wine is not an emulator. Allows running Windows programs under Linux.
Each of the subsections here represents a potential workflow to design logic equations for these parts. The majority of the focus will be on methods that avoid using the WinCUPL frontend/IDE directly (unreliable), but which do use the underlying CUPL.EXE command-line compiler which is fairly robust.
Finally, a word on preferred approach, given the options: Using the CUPL.EXE compiler via command line or avoiding CUPL altogether and using Quartus are probably the best ways, especially if you are interested in using Hi-Z states. Neither Yosys nor Digital seem to have robust support for 'inout' ports, Tristate/Hi-Z states, open collector outputs, etc. If these are important to you, it may be worth considering tools and workflows which are less experimental.
WinCUPL is basically an IDE, possibly before the term IDE came into existance. While logic for these parts can be written using WinCUPL itself, the experience may be fraught with difficulty as it is a quirky and often unstable Windows application (While it does run great under Wine, this doesn't really change things much). I've seen the editor itself crash just for looking at it sideways, and its copy-paste functionality behave in bewildering ways. It does however have value in the help files / documentation / examples. It should emphasized that the CUPL compiler itself is actually pretty solid/stable, and so the troubles of WinCUPL shouldn't be equated with CUPL itself. So, the recommended approach is to start here regardless and use it for documentation/examples/compiler/device-library and then simply avoid it for serious work by using the command line CUPL.EXE (perhaps through the 5vcomp helper scripts in this repository as in the next section).
You can Download WinCUPL from here.
To get it working under Linux with Wine, you'll need winetricks so you can install mfc40 and mfc42. On Ubuntu Linux, this would look something like:
dpkg --add-architecture i386
sudo apt-get install wine wine32:i386 winetricks playonlinux innoextract
WINEARCH=win32 WINEPREFIX=~/.wine wine wineboot
winetricks mfc40 mfc42
wine awincupl.exe
Furthermore, if you are intending on working with the ATF150x parts, you should probably grab the newer fitters out of the Atmel Prochip package. The utilities in this repository will refuse to work with the old fitters.
Since WinCUPL simply is a front-end / IDE on top of the CUPL.EXE compiler and related programs, one can write the desired logic in the CUPL language, save it in a .PLD file using their favorite editor and have CUPL.EXE compile it into a .JED file for programming into a PLD.
5vcomp is a simple wrapper around the CUPL compiler. This is probably the most solid approach assuming you are OK with using CUPL as a language. You should start with the WinCUPL approach as a prerequisite since it installs the CUPL compiler and has examples/help files. The workflows here simply make this easier/convenient by catching a lot of common issues and providing reasonable defaults to the compiler:
Assuming you're using CUPL either through WinCUPL or 5vcomp, this section has a general reference to the language.
An overview of how things work which shows how one can go from a CUPL .PLD into the .JED files needed to program a device. Note that the Atmel Fitter (place and route) stage is only used when working with the ATF150x CPLD parts. For most simpler parts, CUPL is capable of generating a .JED directly.
(diagram from built-in WinCUPL help)
Expand here for details of the command line flags for the CUPL.EXE compiler
Run CUPL using the following command line format:cupl [-flags] [library] [device] source
where
-flags is the following set of compiler options:
-j JEDEC download format
-h ASCII-HEX download format
-i HL download format
-n use input filename for output file
-a create absolute file
-l create listing file
-e create expanded macro definition file
-x create expanded product-terms in documentation file
-f create fuse plot/chip diagram in documentation file
-p create PDIF database interchange format file
-b create Berkeley PLA format file
-c create PALASM format file
-d deactivate unused OR terms
-r disable product term merging
-g program security fuse
-o treat all state machines as “one-hot”
-u use specified library for compilation
-s perform logic simulation after compilation
-w perform simulation with waveform output (MS-DOS only)
-m0 no minimization
-m1 quick minimization (default)
-m2 Quine McCluskey
-m3 Presto
-m4 Expresso
-q MIcrosoft format for error messages
-zq QuickLogic’s QDIF file
-kb Optimize product term usage for pin or pinnode variables. This overrides the DEMORGAN statement if it appears in the source file
-kd DeMorganize all pin and pinnode variables. This overrides the DEMORGAN statement if it appears in the source file
-ks Force product term sharing during minimization. This is also referred to as group reduction
-kx Do not expand XOR to AND-OR equations. This is used for device independent designs or designs targeted for fitter-supported devices where the fitter supports XOR gates
A word of warning is that the Device: section at the top of a .PLD file is more than just the part number you are interested in programming -- it is actually a device mnemonic which selects different macrocell configuration modes.
So, if you're having trouble getting a flip-flop to work, it might be because you have selected the mnemonic for "simple mode".
As an example, the compiler can be set to four different modes for the ATF16V8 (similar considerations apply to the 22V10 parts, etc):
Registered - G16V8MS
Complex - G16V8MA
Simple - G16V8AS
Auto - G16V8
Expand Here for a list of mnemonic prefixes
EP Erasable Programmable Logic Device (EPLD)
G Generic Array Logic (GAL)
F Field Programmable Logic Array (FPLA)
F Field Programmable Gate Array (FPGA)
F Field Programmable Logic Sequencer (FPLS)
F Field Programmable Sequence Generator (FPSG)
P Programmable Logic Array (PAL)
P Programmable Logic Device (PLD)
P Programmable Electrically Erasable Logic (PEEL)
PLD Pseudo Logical Device
RA Bipolar Programmable Read Only Memory (PROM)
While CUPL is self-sufficient to generate the final .JED files for most simple devices, the ATF150x CPLD parts rely on a fitter executable (essentially place and route) that was supplied by Atmel. These fitters work with the CUPL compiler, but the fitters can in theory be made to work with anything that can supply a netlist in the correct EDIF or PLA TT2 format to them. Yosys, Berkeley-ABC, and SpyDrNet might all be able to do this in different ways.
Expand for command line options for the latest known version of the ATF1502.EXE fitter. Output is similar for ATF1504 and ATF1508 devices.
Atmel ATF1502 Fitter Version 1918 (3-21-07)
Copyright 1999,2000 Atmel Corporation
Usage: FIT1502.EXE [-i] input_file[.tt2] {options}
Options:
-help
-o output_file_name (for *.tt3 and *.jed)
-device package_type (PLCC44/TQFP44)
-tech tech_name (ATF1502AS/ATF1502ASV/ATF1502BE)
-module module_name
-preassign TRY|keep|ignore (pin preassignment options)
-silent (no message on screen)
-h2 (advanced help option)
-has (advanced help option for AS)
-hbe (advanced help option for BE)
Advanced help options:
Atmel ATF1502 Fitter Version 1918 (3-21-07)
Copyright 1999,2000 Atmel Corporation
-strategy c [command file name]
-strategy ifmt (input file format) [TT | edif]
-strategy lib (library file name for edif input)
-strategy open_collector = [ OFF | on | = pin_name1 pin_name2...]
-strategy JTAG = [ off | ON ]
-strategy pd1 [ OFF | on ] (power down 1)
-strategy pd2 [ OFF | on ] (power down 2)
-strategy TDI_pullup = [ OFF | on ]
-strategy TMS_pullup = [ OFF | on ]
-strategy DEBUG = [ on | OFF ]
-strategy output_fast [on | OFF | = pin_name1 pin_name2...]
-strategy pin_keep [ off | = pin_name1 pin_name2...]
-strategy ues [value ] (2 ASCII characters)
-strategy security [ OFF | on ]
-strategy tPD = [ 5 | 7 ]
-strategy voltage_level_A [ 1.8 | 2.5 | 3.3]
-strategy voltage_level_B [ 1.8 | 2.5 | 3.3]
-strategy fast_inlatch [ OFF | on | = pin_name1 pin_name2...]
-strategy schmitt_trigger [ OFF | = pin_name1 pin_name2...]
-strategy pull_up [ OFF | = pin_name1 pin_name2...]
-strategy unused_To_PinKeeper [ off | ON ]
-strategy pull_up_unused [ OFF | on]
-strategy unused_To_Ground [ OFF | on]
-strategy pull_down [ OFF | = dedicated_pin1 dedicated_pin2...]
-strategy Latch_Synthesis [ON | off ]
-strategy Optimize [ON | off]
-strategy Cascade_Logic [ON | off |= pin_name1 ..pin_nameN]
-strategy Foldback_Logic [ON | off |= node_name1 ..node_nameN]
-strategy Soft_Buffer [on | OFF |= node_name1 ..node_nameN]
-strategy XOR_Synthesis [on | OFF |= pin_name1 ..pin_nameN]
-strategy Push_Gate [on | OFF]
-strategy Verilog_sim [sdf | Verilog | OFF]
-strategy Vhdl_sim [sdf | vhdl | OFF]
-strategy Out_Edif [on | OFF]
-strategy Global_Fold [node_name1 ..node_nameN]
-strategy Global_OE [node_name1 ..node_nameN]
-strategy OE_node [node_Number1..node_NumberN]
-strategy logic_doubling [on | OFF]
-strategy twoclock [clockname]
-strategy pinfile
- https://github.com/willie68/WCPLD
- https://github.com/Manawyrm/PAL-GAL-CI
- Recently, two different extensions for VS Code for CUPL have been written:
- https://marketplace.visualstudio.com/items?itemName=tlgkccampbell.code-cupl
- This one handles just syntax highlighting for CUPL .PLD files
- https://marketplace.visualstudio.com/items?itemName=VaynerSystems.VS-Cupl
- This is an entire workflow, which has a bit more functionality beyond just syntax highlighting.
- https://marketplace.visualstudio.com/items?itemName=tlgkccampbell.code-cupl
One can literally create a fusemap by hand for a PLD.
- See this blog post by Frank DeCaire, where he documents his journey of doing so.
While not the easiest approach, just as one can write G-Code in notepad or Assembly code in a hex editor, manually creating a fusemap is technically possible. This assumes that you have a datasheet for your device which has a description of the fusemap and the details of how the macrocells work. With this in hand, one could write a JEDEC file with the desired functionality and a text editor. This would be non-trivial and error-prone (if double-negatives confuse you, this is even more exciting), but it demonstrates that such a thing could be done, at least with the older PLDs (16V8, 22V10), and even with the ATF750 (some datasheets actually had the fusemap for this part).
It is worth noting that the fusemap for the ATF150x parts has been recently documented in prjbureau. Given the complexity of these devices over PLDs, writing a fusemap by hand for these parts would probably be a bad idea.
These will only be covered very briefly:
- ABEL: "Advanced Boolean Expression Language" was created in 1983 by Data I/O Corporation.
- PALASM: Introduced by Monolithic Memories, Inc. (MMI) in the 1980's
- A modern version of this is called GALASM which is a continuation of something called GALer. This might be worth considering if you are happy with just PLDs.
- Atmel-Synario: A 1999-era ABEL-HDL and Schematic PLD/CPLD design tool.
- Viewlogic's Workview/ProPLD: 90's-era software.
- Atmel-ProPLD: (possibly Atmel's bundled version of the above?)
Atmel Prochip is not free, however, you can download it from here, and may be able to request a trial license from Microchip. This workflow supports Verilog/VHDL, which is great if one wants to move away from CUPL entirely and can afford to purchase a license.
Prochip should be downloaded regardless because there are newer fitters for the ATF150x devices that can be extracted from this installation, and these fitters are required in every other approach mentioned here. The newer versions of the fitters should mention version 1918 (3-21-07) when invoked from a command line. (The fitters that come with WinCUPL are old and should be replaced with the ones from this package).
In essence, Prochip is the Atmel fitter bundled with:
- "Precision® RTL Synthesis" from Mentor Graphics for Verilog/VHDL synthesis.
- (optionally) "ModelSim®" from Mentor Graphics for Functional/Timing Simulation
- (optionally) "Protel Design Explorer 99SE" for Schematic/CUPL design entry.
- It turns out that the Altera (Now Intel) Quartus II 13.0sp1 Web Edition can be used to produce a .POF file targeting various CPLD chips made by Altera in the MAX EPM3K/EPM7K series, which can be converted to target an ATF150x CPLD.
- Intel® Quartus® II Web Edition Design Software Version 13.0sp1 for Windows
- Intel® Quartus® II Web Edition Design Software Version 13.0sp1 for Linux
- When installing, you only need "MAX II/V, MAX3000/7000" under device support. Unchecking the other devices can save ~2GB.
- You may have trouble running it as libpng12.so.0 may be required. See here.
- Quartus frequently crashes and can even crash your whole desktop session. Fix that by replacing the bundled libtbb.
- You can move the .desktop shortcut into ~/.local/share/applications/Quartus II 13.0sp1 (64-bit) Web Edition.desktop
- I found I had to set Terminal=true for it to work.
- The resulting .POF file can be converted using a utility called POF2JED from Atmel (Now Microchip). This is further detailed in this application note.
- Important!: Newer versions of Quartus will not work. v13.0sp1 is the last version that had support for the MAX EPM3K/EPM7K chips. Support for these chips has been removed from newer versions of Quartus. You MUST use the old version.
"Digital is an easy-to-use digital logic designer and circuit simulator designed for educational purposes."
This is an interesting option as one can create a schematic and have a .JED file generated for a GAL16V8 or GAL22V10. If one provides the fitters to Digital, it can produce .JED files for the ATF150x series as well. Note that this is more of an educational tool for learning about logic. You may have trouble if you expect fullly featured support of these devices (Tri-state pins, Bi-directional IO, etc.)
- https://github.com/hneemann/Digital If this appeals to you, you might be interested in similar software (though no support for the Atmel parts):
- Logisim
- Logisim Evolution
In theory, one can use Yosys Open SYnthesis Suite (Yosys) with the help of the Atmel Fitters a specific CPLD and a techmap to produce .JED files. This is a bit more experimental, but some have managed to make this work. This allows an almost entirely open-source workflow using Verilog, and probably Icestudio if one prefers schematic capture as well. A good place to start would be using the OSS CAD Suite to get the big parts of the suite set up. After that, there are two approaches to making this work:
- https://github.com/whitequark/prjbureau
- prjbureau demonstrates going from RTLIL to a .JED file
- https://github.com/hoglet67/atf15xx_yosys/
- This example goes from plain old verilog into a .JED file by implementing a techmap.
- https://github.com/michaelhunsberger/JsonToCupl/
- This is an example of how to use Yosys to generate CUPL code.
- Potentially interesting as one could use this to generate a .PLD even for the simpiler 16V8 or 22V10 devices
- http://forum.6502.org/viewtopic.php?f=10&t=7601
Finally, since yosys is extremely complex, a section on understanding the basics is in order especially from the context of these devices. For the moment, however, others have written guides for different parts:
This was a ~1999/2000 era circuit board design tool made by Altium that worked in Windows and which had support for the SPLD and CPLD parts mentioned here. It is mentioned here for completeness sake, but the author has no direct experience with it. It is said to have supported CUPL and Schematic entry for development of logic but not Verilog nor VHDL.
Berkeley ABC can be made to read and write verilog and the PLA format used by the Atmel Fitters. If this works, it could potentially eliminate the need to use the CUPL language altogether and instead have a path from verilog to the Atmel CPLDs without the need for expensive software. This remains to be tested.
SpyDrNet is capable of generating an EDIF netlist which could in theory be fed into the Atmel Fitters.
There are a few choices on how a PLD/CPLD part can be programmed depending on whether it supports JTAG. If using JTAG, be mindful of making sure you are using a programmer with the correct voltage levels and not to unintentionally programatically disable the JTAG interface.
Finally, if one is able to read a .JED out of a device, this can sometimes be reversed back into equations, provided the security fuse on the device has not been set. Given a .JED file, the following approaches can be taken to arrive at the equations:
- By hand, comparing the .JED file to the fusemap / macrocells in the datasheet. See this blog post by Frank DeCaire.
JED2EQN.EXE- A DOS utility floating around on the internet.jedutil- MAME can be compiled with a utility called jedutil which does something similar. Sometimes it is broken out into a seperate package "mame-tools" Finally, brute force can be used on a PLD that is strictly combonatorial can be read out as though it is an EPROM by stepping through all combinations of possible inputs. In this approach, security fuses does not matter because one is not trying to read out the fusemap directly. Once state/registers are involved, this becomes much more challenging.
CSIM.EXE can be fed test vectors and be used to simulate the behavior of a particular chip, or even a virtual device. The following things are required to do this successfully:
- You must create an .SI file containing the desired test vectors.
- Provide an "Absolute file". The .ABS file is generated by CUPL.EXE when the -a flag is passed to it. This generates a binary file based on the logic equations from the source CUPL .PLD file. CSIM will then generate a .SO output file, and optionally append test vectors to an existing .JED file for testing purposes.
Expand here for command-line options to CSIM.EXE
csim [-flags] [library] [device] source
where
-flags is the following set of simulator options:
-l create listing file.
-j append test vectors to JEDEC file.
-n use source filename for JEDEC file.
-v display simulation results to terminal.
-u use specified library for simulation.
library is the library name and path name if the -u flag is being used to specify a
library other than the default library.
device must be the same device mnemonic as was used in the CUPL compilation.
Specifying the device is optional; if a device is not specified, CSIM uses the device
CUPL compiled (contained in the .ABS file).
source is the user-created ASCII test specification file (filename.SI). The
extension .SI is assumed for the source file and may be omitted when giving the
CSIM command.
Creating a .SI file:
- An .SI file should have the same header information as the original .PLD source file. If not, this will generate warnings.
- Comments begin with a /* and end with a */
- An .SI file can have the following keywords/statements: ORDER, BASE, and VECTORS
- The ORDER keyword is used to list the variable / inputs and outputs to be used in the simulation table, and to define how they are displayed. Typically, the variable names are the same as those in the corresponding CUPL logic description file.
- The BASE keyword specifies a number base. Hexadecimal is the default if unspecified.
- The VECTORS keyboard specifies a list of test vectors (signals that are applied and expected outputs).
- If you simply want to see what will happen on the outputs rather than setting a pre-determined expected value, set the outputs to *
Expand for a list of valid Test Values used in a test vector
0 Drive input LO (0 volts) (negate active-HI input)
1 Drive input HI (+5 volts) (assert active-HI input)
C Drive (clock) input LO, HI, LO
K Drive (clock) input HI, LO, HI
L Test output LO (0 volts) (active-HI output negated)
H Test output HI (+5 volts) (active-HI output asserted)
Z Test output for high impedance
X Input HI or LO, output HI or LO Note: Not all device programmers treat X on inputs the same; some put it to 0, some allow input to be pulled to 1, and some leave it at the previous value.
N Output not tested
P Preload internal registers (value is applied to !Q output)
* Outputs only -simulator determines test value and substitutes in vector
' ' Enclose input values to be expanded to a specified BASE (octal, decimal, or hex). Valid values are 0-F and X.
“ ” Enclose output values to be expanded to a specified BASE (octal, decimal, or hex.) Valid values are 0-F, H, L, Z, and X.
- Engineering Practices for PIC Microcontrollers and the ATMEL (ISBN
This repository is merely a bunch of tips, tricks, helper scripts and documentation. The real work comes from:
- Whitequark for putting together Prjbureau, which documents the fusemap for these devices, provides an ability to go from a .JED file to an .SVF, documentation and more.
- Yosys
- hoglet67 for putting together atf15xx_yosys, which shows a workflow using yosys and provides a techmap.
- Countless other tips, tools, contributions and from all over the web and plenty of trial-and-error working around the quirks of WinCUPL and the fitters themselves.
Footnotes
-
https://www.eevblog.com/forum/beginners/5v-pal-or-gal-with-cmos-outputs/ ↩
-
From the Readme.txt of fit5_0.zip (an older version of the Atmel Fitters) ↩ ↩2
