Design Article
Replacing obsolete video game circuits with Xilinx CPLDs
In Choi, Retro Devices Technology
10/8/2008 12:24 PM EDT
Abstract
In this article, designer In Choi describes a project for his home business in which he replaces a defective part in a 1980s game system to show his full-time employer that they can adopt a methodology using Xilinx programmable devices to replace a range of parts that other semiconductor vendors are no longer producing.
Introduction
Parts obsolescence is nothing new, but it has attracted more attention recently, particularly in the semiconductor industry. Sometimes chip companies no longer produce parts for older products; others go out of business.
Although no single optimal solution exists, you can typically manage parts obsolescence using one of three methods:
- Find a form, fit, and functional substitute.
- Redesign the subsystem containing the obsolete part.
- Replace or redesign the entire system.
From a practical perspective, the last two methods are generally too expensive; it would probably be more cost-effective to replace the subsystem or the entire system.
Thus, the first method is the most practical, and we'll show how we at Retro Devices Technology did it using a Xilinx XC9536XL CPLD – specifically targeted for 5V transistor-to-transistor logic (TTL) gates and digital logic functions – to replace a 74LS32 microcircuit in the PCB of a 1980s video game system called Vectrex. For the sake of this project, we are assuming that the 74LS32 is obsolete.
Vectrex system overview
The Vectrex video game system was originally licensed and distributed by General Consumer Electric (GCE) in the early 1980s. It was a self-contained game unit much like the full-sized game machines in arcades but scaled down for home use. It was somewhat portable as long as you could plug it into an AC outlet.
The Vectrex system comprises an 8-bit 6809 microprocessor running at 1.5 MHz, an AY-3-8912 sound chip from General Instruments, a 9 x 11-inch monochrome monitor, and a game controller with four buttons, as shown in Fig 1. The Vectrex uses a vector graphics system as opposed to a pixel-based raster graphics system.
1. The Vectrex gaming system.
(Click this image to view a larger, more detailed version)
In this design, we used the XC9536XL CPLD to replace a part called the 74LS32 Quad-OR gate in the Vectrex video game system. The original designers used this part as decoding logic to perform read and write access to the RAM by the processor in the Vectrex (more information is available from the Vectrex website at www.vectrex.nl).
I chose to replace the 74LS32 with the XC9536XL CPLD to minimize development cost and to prove out my scalable design methodology simultaneously.
Design description
Our design requirements were to keep the physical size of the replacement part as small as possible, meet all of the electrical specifications, keep the power level at least equal to the obsolete part, and use a replacement part that is configurable to replace as many 5V TTL 74LS series gates and digital functions as possible.
I created the customized functions using VHDL, then synthesized the code into a configuration file and programmed the design into the CPLD. For the sake of this article, I'll refer to the replacement part as the retro logic device (RLD). Fig 2 shows the fully operational RLD device next to the 74LS32.
2. The RLD (left) and the original 74LS32 part it replaced (right).
(Click this image to view a larger, more detailed version)
The design footprint of the RLD is a little larger than the 74LS32 because of the PCB vendor's manufacturing limitations, but there was sufficient room in the original game circuit board to accommodate the RLD.
The electrical performance of the RLD was a non-issue, since the Xilinx XC9536XL CPLD already has TTL I/O compatibility. You can read the XC9536XL data sheet if you are not familiar with the specifications or how to properly power the I/O for TTL compatibility.
The power of the RLD is also comparable to many 74LS series functions. I achieved this power compliance by using a 3.3V low power voltage regulator on the RLD's PCB. I placed a bypass capacitor at the output of the voltage regulator to provide filtering for the CPLD.
From a signal integrity perspective, it is typically best to place the capacitor directly under the board and attach it to the CPLD's voltage and ground planes to provide the shortest (lowest inductance) path. However, the Vectrex circuit board didn't require a high level of performance.
I then placed a 6-pin JTAG connector on the RLD to configure the CPLD. Finally, I created a custom mating connector for the RLD for configuring the RLD. I attached this custom connector to the platform cable USB using flying leads.
Comments
Randy in NC
10/10/2008 3:04 PM EDT
Yeah, you could have done this 20 years ago with a PAL or a GAL.
Altera's Quartus (and MaxPlusII) have a full library of TTL parts available (including 7497's and 74167's for example).
Sign in to Reply
RDTLLC
10/10/2008 10:27 PM EDT
Clay,
I appreciate your response.
A patent typically calls out prior art on the first page, and my patent does call out any prior art as well as the CPLD limitation you indicated. The patent also alludes to the fact that it can be done in an ASIC. The ASIC would address the electrical concerns you pointed out. Unfortunately I did not have millions of dollars to spin an ASIC, so my initial attempt was with COTS components. By the way, the CPLD could be replaced with an FPGA with Schmitt Trigger I/O, but It cannot address 15V HO I/O.
All legal issues were resolved between my Patent Attorney and USPTO. The patent was issue on Oct. 2, 2007; this IS NOT patent pending but an official US Patent. I might add that the concept originated back in 1994, but due to unforseen situations it took me this long to get to this point.
Finally, the patent was designed to cover both commercial and military applications. My main target was for the HIGH REL, RAD HARD military market. Any devices that did not pass MIL screening and testing would be down graded for commercial applications to offset manufacturing and production costs. The patent does not re-invent the wheel. It shows a process and methodology to address parts obsolescence in an elegantly simple and unified way and in an economical way. Especially when a military 54LS32 costs about 100 times the commercial part.
Regards,
In
Sign in to Reply
RDTLLC
10/10/2008 10:29 PM EDT
Randy,
I am in the military market. From my experience, obsolescence in the military market is not so trivial.
Do a little research on GIDEP, Rochester Electronics, Lansdale Electronics, L-3 ES, and GEM program from Sarnoff. All these companies have yearly budgets that run into millions of dollars. They all cater to military parts obsolescence. These companies all have ASIC fabrication facilities just to recreate older IC technologies.
For every unique function, each manufacturer has to have a unique wafer. The by products are wasted silicon and extensive management and tracking of specific device functions.
My method only requires one wafer; each identical and can be re-configured. That alone could reduce manufacturing costs considerably.
If you have a simple answer to this problem, you could file a patent and put these guys out of business. By the way, there are more politics in the military world.
Regards,
In
Sign in to Reply
Vectrex
10/11/2008 1:51 PM EDT
Cool project. Would love to see a Vectrex on a chip. Or at least as close as possible give the display constraints.
Sign in to Reply
RDTLLC
10/12/2008 10:53 PM EDT
Vectrexer,
I agree, it is cool. But I am being biased.
Most of my time is focused on developing the hardware. I wouldn't doubt someone already has done it.
Regards,
In
Sign in to Reply
clay_cowgill
10/20/2008 2:09 AM EDT
Hi In,
(Hopefully this doesn't double-post-- had problems the first time.)
Getting a patent issued and having a defensible one are two different issues. The patent office has a LOT of examiners with a lot of pressure to plow through applications. Getting a patent issued and having a strong, valid patent are different things. (Patent examiners and patent lawyers tend to get paid the same for a 'weak' patent or a 'strong' one... YMMV.)
In the case of your patent, there's just plenty of prior art out there.
Signetics 1983 databook has an example of replacing a BCD driver IC with one of their PLS series parts. Pick up an old copy of "Digital Design Using ABEL" by Pellerin & Holley (published 1994). Chapter 6: "Basic TTL Logic Functions" goes on to describe how to use ABEL with PLDs to replace 74241's, 245's, 373's, 274's, 166's, 164's, 154's, 259's...
Hitting closer to home-- we were shipping a TQFP packaged XC9536 with a 6-pin programming header on a little DIP module as a programmable replacement for either of two chips used on Pacman arcade game boards. The module was the same for either device, just a different fusemap to select between them. Essentially the exact same item as your patent, only in a larger (28 pin) DIP form factor. Those were shipping commercially in 2000.
Having the patent is cool and all-- I just wouldn't recommend trying to sue any 'infringers' based on the strength of the patent...
-Clay
Sign in to Reply
RDTLLC
10/20/2008 5:29 PM EDT
Hi Clay,
I am not a patent attorney so it is ill advised for me to comment on the validity of what is patentable or the patent process. It is what it is, and it is part of the business. The subject matter would be best served between the client and the patent attorney. This forum would be best served if the discussion is limited to the technical process and methodology of the article.
Having said that, I would be more than happy to talk to you personally about the specifics of the product and how it could enhance to marketability of your products. Please contact me via email through my website.
Regards,
In
Sign in to Reply
WildMan
11/6/2008 10:54 AM EST
Wow, that is cool! I still have my Vectrex system running.
Sign in to Reply