On 6/7/2013 3:34 PM, Paul Schreiber wrote:
> exactly reproducing the Synthesis Technology look
>
>
For me, speaking as a manufacturer, I committed to this "look" not so
much in terms of the actual appearance but for manufacturing
efficiency. Once you choose the tooling for a production process, you
tend to stick with it, mainly because re-tooling is expensive in terms
of re-engineering all the hard points (jack to jack spacing, no. of pots
in a line, etc.) This is why we have standards like MU, 5U frac and
Euro: because they all adhere, more or less, to common design methodologies.
Recently, someone admonished me for not wanting to customize the
front panel for a module, saying something like "cmon, its ∗2013∗!"
Well, no matter what year it is, manufacturing efficiency is still the
top priority for someone who is making more than one or two of a given
module. Consider: it is fun to built the first one. The 100th of the
same module? Not so fun. So, you take whatever steps you can to make
the assembly of many of a given module as efficient and un-tedious as
possible. This is why I make jack boards and pot boards even if the
cost of the extra parts offsets the cost savings in man-hours: it still
∗saves time∗ and that is the one commodity I do not has as much of as I
once did.
The same is true for the two, soon to be four, module series I
currently sell. I made the VCO153 (Yamaha IG00153 VCO chip in
non-custom form with IG00158 waveshaper also in non-custom form
attached) in SMT because I have an SMT assembly line, and after the
soul-crushing drudgery of loading the feeders for the pick and place
machine, you can assemble 25 boards in about an hour. Once the paste
printer stencil is aligned and the feeders are loaded, and the reflow
oven's zones are all at operating temperature, the actual build process
takes about 2.5 minutes per board. Thus you try to get as many boards
through the line as you can before you have to reload for a different
product. (This is also why we try to design around a basis of common
SMT parts--we ∗hate∗ loading the deck feeders with parts reels--it takes
as much time as the assembly run!)
I took a different path with the GX1BPF: the boards are all
through-hole, and I hired an assembly house build them from my
construction notes and a provided assembled sample. That cost me $500
(for 25) but that is 50 hours of my time not spent assembling boards.
Especially the same board over and over. Why through-hole? The BPF is
the older design from my spree of module engineering back in 2003-4. I
built exactly three back then: Robert Rich and Lester Barnes have one
each, and I have the 3rd prototype. The VCO, while prototyped at that
time, was never fully tooled for production until late 2011. By that
time I had my SMT assembly line, and The VCO got the SMT approach. The
BPF will eventually--I've already made the board layout--but not until
these BPFs I have on hand sell out. Or down to less than 5 units.
Panels are another thing: good ones aren't cheap. The VCO153 uses a
Schaeffer/FPE black anodized, engraved with paint-fill panel that looks
very nice, but costs me $40. In quantity. The GX1BPF uses 3mm epoxy
spray-painted aluminium stock with a traditional silkscreen. It looks
decent, but not as good as the FPE or Paul's panels. But these were $17
each. You take what you can get when the getting is good, which is why
the modules sport different panel types. The point here is that
changing the tooling for panels where the non-refundable engineering
costs are per revision greatly affect the cost of the panel. The $125
NRE charge for the GX1BPF panel (for example) would need to be applied
again, and all of a sudden the $17/panel has become $22/panel.
In any case, while I am open to evolving the form factor of a module
family, it has to occur along lines that are friendly to the
manufacturer. There needs to exist a comfortable medium between "module
as art" and "module is manufacturable without driving the manufacturer
stark raving mad."
Even while I to stick to this design paradigm, I still try to be
accommodating. I offer cabling pads for direct-wired pots and jacks in
addition to my jack & pot board headers. In the case of the upcoming
MOTM-480 Mark II (CA3280s gone, V2164s now used) I will have
user-configurable jumpers and extra filter stage outputs to allow a full
split filter configuration without the need to cut traces or solder
directly to parts leads. The board will allow the use of a 3-pot
bracket as well as a 4-pot bracket, without the need for the steel
stud-mounted plate Paul uses. (I like that plate, I just don't like
having to get PEM bolts mounted into panels. ;)
This is pretty much the stance I work from, and while I try to evolve
things along their logical path, at the end of the day the devices still
need to be efficient to manufacture.
Crow
/∗∗/