|
The Memotech MTX Series |
|
Memotech Multi-Function Expansion System
MFX Engineering Changes
one ?)
Engineering Changes
January 2024
This page describes the engineering
changes made subsequent to the first shipments of MFX. (The
firmware page describes minor
enhancements to the firmware, the focus on this page is purely
hardware.)
The biggest issue that I experienced was
supply shortages and the variable quality of the FPGA modules
that we used in the design. (in hindsight, it would have been
better to use a device that was not so near the end of its
production life, but hey-ho!).
When MFX was released, the world was
still in the grip of the Covid pandemic and electronic parts
that were previously cheap and in plentiful supply suddenly
became scarce and correspondingly most expensive. It was hoped
that as the global situation improved, things would return to
something more like normality, but, even by early 2024, that
hadn't happened. The FPGA modules had increased significantly in
price, and although many Chinese sellers were advertising as
having them in stock, when orders were placed, they frequently
failed to deliver, and when they did, the quality was variable,
with many having faulty I/O, most likely due to poor quality
assembly, or possibly the use of factory reject FPGA chips.
Something needed to be done if MFX was going to continue without
a major redesign . . . . .
The FPGA module was so common and
(supposedly) available from so many different suppliers that I
thought that it must have been an Open Source project, at least,
originally, and as such, I should have been able to locate the
original design files with a view to having them manufactured
and assembled by my usual PCB supplier. Unfortunately, I was not
able to find the design for the original board anywhere on the
web, however, I did find a possible solution . . . .
Ralf Thelen, the designer of a number of hardware projects
for legacy
pinball machines, had used the same FPGA development board
for a number of his projects and had experienced the same
quality issues that I had. As he notes on his website, "Because
of this I made a 99% copy of the [Cyclone II Development] board
which can be fully assembled by
JLCPCB." Ralf described this and included links to his
repository containing the design files
on this page
of his website, lisy.dev.
The
original development board on the left and Ralf's on the
right - good isn't it! |
|
|
The most obvious difference is that the
JTAG port is
not fitted, but MFX doesn't need it - programming is done using
the
ASP port on the left. This made space to place all of the
components, including the oscillator and memory chip, on the
front side of the board which makes the cost of assembly
cheaper.
Obviously, this sounded like exactly what
I was looking for, but things got even better! Whist supplies of
the FPGA seem to be still available, at least in China, the
Cyclone II is obsolete and will surely become harder to source
before too much longer. I was really pleased when I saw that
Ralf had also done a version of the development board with the
same form factor for the Cyclone IV FPGA. This chip too is close
to "end of life", but should be available for longer than the
Cyclone II, with the added advantages that it is more powerful,
with additional internal memory and it is actually cheaper than
the Cyclone II.
It seemed that the Cyclone II version
would allow me to secure my own supply of development boards as
a like-for-like replacement for the originals and the Cyclone IV
version would also be a "drop-in" replacement with the added
benefit that the additional resources might offer scope for
further improvements to MFX.
Having made contact with Ralf, I have to
say how helpful and supportive he has been. At this point,
things are at an early stage, but I am actively pursuing this
option . . .
More details shortly . . .
|