[sdiy] My frustration as a technician..

Sat May 11 22:30:45 CEST 2013

Hi Martin,

On Sat, May 11, 2013 at 9:32 PM, Martin Klang <mars at pingdynasty.com> wrote:
>
> I use 0603 footprints for most SMT resistors and caps: 1.6 mm × 0.8 mm, on one side.
>
> Compare that to 7.5 or 10mm grid for through hole, on both sides -
> how could you fit the same components with dual footprints without completely changing the size and layout of the board?

Capacitors don't suddenly become bigger because they're through-hole.
I mean sure, the leads do have measurable size, but a tiny SMT
capacitor is still going to be tiny in DIP, and a big SIP one is going
to be big in SMT. I'm not sure how a capacitor that's 1.6x0.8mm would
suddenly become 10mm x 10mm just because it has through-hole leads,
unless you mean you'd change the capacitor type from say ceramic to
electrolytic. But that makes no sense. Besides, on packed through-hole
boards you usually put resistors and all other elements upright. So
the actual area of the board doesn't change as much as you suggest.

> And what do you do with ic's?

Well, ICs are a different thing. There are several situations one
might have here:

1. BGA digital IC
You're very unlikely to replace such an IC unless you're replacing it
with the same type. You're very, very unlikely to replace such an IC
with a subcircuit, but some people do that as well sometimes. We've
all seen old computers being upgraded by replacing the CPU with a
board that has several CPUs on it. The thing is, though, we're really
talking about analogue electronics here. The functionality of digital
electronics in BGA format is changed by modifying the code they run.
No reason to solder anything, usually, unless you have a fuse-blown IC
and need to rework it to place your own dev version on the board. If
you're placing such a circuit on a board, it would be nice if you'd:
i. place the digital section on a separate board, which could in any
case be fully replaced by something else (compare the new korg polysix
cpu boards)
ii. place enough room around the chip that it can be reworked with hot
air without too much hassle (adjacent chips unsoldering)

2. QFP or similar digital IC

The same really applies here as with BGAs, except those are easier to
rework without hot-air stations.

3. j-lead digital IC

Is it a hardware logic family, e.g. 74hc*? You'd be a nicer person if
you included a dip footprint around it. Is it something that only
comes in j-lead? Consider some of the earlier described things.

4. j-lead analog IC, BGA analog IC

Those come, but they're often accompanied by DIP versions. The most
likely situation is that people use SMT because it's much cheaper to
manufacture in SMT. It would be very nice if you included a dip
footprint around your part. If it's a part that doesn't come in a DIP
package, put a dip footprint around it anyways.

> I don't think you can ever arrive at an easy metric of 'x pence per hole, per board'.

No, I don't think you can.

> If you want to make a comparison, I suggest you take some existing smt layouts that you are interested in and re-do them with dual footprints in the different ways that you propose.
> Compare board fabbing costs, and add in the extra design time. Then you can get an idea of the real overhead.

That's definitely one way to find out first-hand. I'll try doing this
with any designs I'll be having made.

> Then take into consideration what the different layouts does to the performance of the board.
> Are the tracks longer or shorter, digital and analog sections better or worse isolated, are the decoupling caps further removed; what is the effect of parasitic elements?

weeeelll... track size, tight decoupling, etc is only going to matter
in high-speed digital. We're not necessarily talking about that here.

> If you were to do this I'd look forward to hearing the results.

Thanks!

Hi firemote,

On Sat, May 11, 2013 at 9:51 PM, firemote <firemote at gmail.com> wrote:
> Right, you're limiting the potential of the design, increasing the time it
> takes to design,

Wellll... given that you most likely have to make footprints for most
of your devices anyways (standard footprints usually suck), it's not
like you're doing that much more by placing a hole here and there
while making the footprint.

> making it more expensive ("cheap" tweaks add up with scale)

not sure what you mean here? Can you explain?

> and indebting yourself to supporting a tiny, niche community

Who said you need to support modded devices?

> that would
> actually mod the devices (let alone needing to test those expansion
> possibilities) well past the lifespan of the unit, and well beyond the
> unit's spec.

Ah, well, that's the point. There should be a way to change the unit's
spec at will.

> Whether someone ought to offer these or not, it's not easy, and
> it's not cheap when all factors are taken into account.

I still haven't seen a great point made in favour of this opinion.

> Myself, I find that devoting effort to this would make it more likely that
> the product I'd receive might have a defect than it would save any time
> trying to repair and replace components after it's well end-of-life.

Are you saying the manufacturer is more likely to ship defective
devices? Why would that be?

Or are you saying you, as the manufacturer, are more likely to receive
defective devices to service? Why would you service devices with a
broken warranty seal?

> I'd
> just buy boutique products that design for this rather than expect this
> functionality to be provided through commodity instruments.

Right, I see you're operating under a false premise. I specifically
stated I'm talking about specialist equipment which is used by people
who are actually smart enough to know how to modify it. I'm not
talking about having through-hole footprints on an iphone.

> On Sat, May 11, 2013 at 12:32 PM, Martin Klang <mars at pingdynasty.com> wrote:
>>
>>
>> I use 0603 footprints for most SMT resistors and caps: 1.6 mm × 0.8 mm, on
>> one side.
>>
>> Compare that to 7.5 or 10mm grid for through hole, on both sides -
>> how could you fit the same components with dual footprints without
>> completely changing the size and layout of the board?
>> And what do you do with ic's?
>>
>> I don't think you can ever arrive at an easy metric of 'x pence per hole,
>> per board'.
>>
>> If you want to make a comparison, I suggest you take some existing smt
>> layouts that you are interested in and re-do them with dual footprints in
>> the different ways that you propose.
>> Compare board fabbing costs, and add in the extra design time. Then you
>> can get an idea of the real overhead.
>>
>> Then take into consideration what the different layouts does to the
>> performance of the board.
>> Are the tracks longer or shorter, digital and analog sections better or
>> worse isolated, are the decoupling caps further removed; what is the effect
>> of parasitic elements?
>>
>> If you were to do this I'd look forward to hearing the results.
>>
>> best,
>>
>> /m
>>
>>
>> On 10 May 2013, at 23:01, cheater00 . wrote:
>>
>> > John,
>> >
>> > On Fri, May 10, 2013 at 11:33 PM, John Speth <jspeth at avnera.com> wrote:
>> >>> Rob, Dave, John, what do you guys say: does it really cost that much
>> >> more to
>> >>> put holes in your PCBs? Serious question.
>> >>
>> >> Yes, it really does.  Every little thing contributes to cost.
>> >> Generally,
>> >> the smaller the better (lower product weight, less material, lower
>> >> shipping cost, etc).  Pennies count in a world where customers demand
>> >> low
>> >> prices and capitalists demand high return on investment.
>> >>
>> >> If the maker of a product could see a higher profit in making products
>> >> with thru hole, it'd be on the street in no time.  It'll probably never
>> >> happen though.
>> >>
>> >> JJS
>> >
>> > Not really the answer I had been hoping for. Given 100 holes, how much
>> > does it cost to do:
>> >
>> > a) a hole, through-plated
>> > b) a hole, non-plated
>> > c) a pad which is a space for a hole
>> >
>> > at amounts of 100 PCBs, 10 000 PCBs, 100 000 PCBs?
>> >
>> > That's 10 000 holes, 1 000 000 holes, 10 000 000 holes.
>> >
>> > Cheers,
>> > D.
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at dropmix.xs4all.nl
>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>
>