PCB multiple designs surcharge

This is likely to be down to things like test setup costs. If PCBs are routed, it probably also increases rout cost. I suspect the above may refer to designs submitted as seperate files - if you have pre-panelised it, then as far as the PCB co is concerned it is one designed - I've submitted multi-design PCB panels to PCBcart with no problems.

On a sunny day (Sun, 07 Feb 2010 19:53:26 -0800) it happened John Larkin wrote in :

Only one?

ET would be the only 'reasonable' excuse for that kind of pricing.

Why? All boards are just a collection of separated nets if no components are placed. What is different if one group of disconnected nets is in one area of the board and another group in another area?

And if I design a board with galvanic isolation I have just such a layout, only difference is that you may see some silk print for optocouplers and DCDC converter 'connecting' the otherwise completely disconnected groups of nets. Would you accept a charge for 2 designs for this board?

No complaints. Ordered three small boards, received seven for no extra charge. It was a small, oblong shape that I guess they used to fill out the panel. Took a while to receive. Good quality. I'd use them again.

Actually there's one overhead circuit (upper left, voltage regs and such) and four candidate "pin driver" output circuits, one of which worked and got used in a couple of products. The other little bits are a lowpass filter (which we've used a couple of times), a couple of gaasfet testers (not used so far) and a couple of transmission line things for TDR/TDT testing. There's also one TDR test trace just for fun.

Oh, there are two dipswitch-based attenuators of different topology, one of which got used in a product for gain cals. We have since found a really tiny trimpot that's good to about 1 GHz.

John

On a sunny day (Mon, 08 Feb 2010 07:26:12 -0800) it happened John Larkin wrote in :

Nice, anything above 1GHz or so will probably remain magick to me :-) I simply do not have the equipment to measure that stuff. Have some parts here to try a diode sampler one day, just ordered hundreds of dollars more mondaine parts... The latest thing with ordering from conrad.nl is, that if you manage to enter all type numbers in the browser without the site flipping or browser crashing, then in the 'shopping card' they also include related things that you did *not* order... makes it hard to check, had the spreadsheet next to it. I think the idea is that you think 'oh I need that too'. Browser crashed just after I closed the secure connection. Why do all those sites have to have all that crap all over the screen?

To me, too. I haven't the nonlinear device models, the EM software tools, or the brute intelligence to predict how some of this stuff will behave. So a mixture of instinct, experience, and experiment will have to do.

I started with flea-market Tek gear, a 547 scope and a 1S2 TDR/sampler, about $120 total. Nowadays you can get an 11801 mainframe and a 20 GHz TDR/sampling head for around $1500.

ftp://jjlarkin.lmi.net/1S2.jpg

The 11801 is a beast, but it works beautifully. This one is displaying a 1 GHz square wave from the successful pin driver circuit on my multiple-circuit board, under the Mantis. The sampling head is on an extender cable so it can snug right up to the DUT.

ftp://jjlarkin.lmi.net/DSC01371.JPG

The test source is the SRS CG635 2GHz clock generator, a nice box with the usual SRS quirks. It has a clever but complex way of generating a very low jitter DDS-based clock.

I've done a couple of samplers, just for fun. The Tek S2 topology (SRD driving shorted transmission lines, 2-diode feedback sampler) works well and isn't hard to do.

ftp://jjlarkin.lmi.net/Sampler1.JPG

flipping or browser crashing,

Firefox under XP seems very solid to me. It's not elegant but it works.

John

more?

Vote with your feet; there are too many board houses around for you to have to put up with that crap.

JF

is

y by mail)

Good points. I suppose I was thinking if seperate files/failure notices were kept for the different sections - a small NRE/w'ever might be reasonable. But, would the OP expect to have the entire panel rejected if a failure were limited to a single circuit? Sems like a waste if the other circuits can still be singulated and used.

I dunno.

ned

this

ply by mail)

In my case, there are 3 different functional units, 4 versions (using different chips) of 1 functional unit and 2 to 5 copies of each version. That way, we can control the inventory of different modules with regular runs. Yes, they are fully independent, so one failure will not affect other modules. I am not sure how PCBcart interprets multiple designs. But I have encountered several PCB fab houses/ dealers that want to charge multiple setup costs. I usually just negotiate it down or walk away.

On a sunny day (Mon, 08 Feb 2010 08:59:49 -0800) it happened John Larkin wrote in :

Nice, at least they had decent knobs beack then :-) Knobs have evolved faster then us humans, maybe a few generations onward people will have little forked fingers to toggle those tiny levers :-)

It is a nice big screen, but the edges of the waveform are rounded, not a real square wave. < 10GHz bandwidth?

Yes, you did show that before, and I did take a real good look at it. Something like that I will probably try.

flipping or browser crashing,

Yea, but firefox needs some gtk libs in Linux that I do not have, and did not want to compile on my system last time I tried. So I use Opera, it is way more stable then my last firefox.

The "legitimate" part of such a surcharge would be because PCB fab houses often determine their prices based on "average" boards with a certain hole density, a certain part density/number of parts, a certain board size, etc., and by combining designs yourself they view it as "gaming the system" because you're purposely creating a board that's likely well outside of "average" but you're still getting the "based on average" pricing and hence reducing their profits.

Note that the cheaper the board house (usually indicating lower margins), the more likely they are to start nickel and diming you for this sort of thing.

As other have mentioned, it's also possible they do it "just because they can," looking to increase their profits with no real increase in their own costs. How much it's this vs. the "legitimate" reason above is anyone's guess for a given fab...

It does make sense to examine different board houses when you're caught in this situation, of course -- obviously it's better for everyone if a board house, through a different pricing model or better efficiency, can make money with your particular design while still giving you a price that you find acceptable.

---Joel

I'm not sure I follow your reasoning here. :<

Assume I have two designs (or N designs -- where N is largest number that can coexist on a panel). How is it any different than having

- the greatest hole/part density

- the least hole/part density if both of these designs would have been "based on average pricing"?

I can see if the resulting panel had to be laid out by the vendor (i.e., he can't step-and-repeat a *single* design). I can see if the vendor had to panelize M different shaped boards. I can see if the processing options for each board were different.

But, if all of the boards were the same size/shape/process; and, if I panelized them AS IF they were a single board M times larger than the individual boards (i.e., pretend it is one board that is just cut in M-1 spots), then I really don't see the justification for this.

E.g., if I place the M artworks side by side with line on the silkscreen layer that says "cut here" and have *this* "single board" panelized -- with it very obviously intended to be cut after-the-fact on each of those silkscreened lines, what claim would the vendor have to increase the price?

Now, I turn to the vendor and say: "I am going to cut these boards on these lines after I receive them from you. What will *you* charge me to make those cuts for me?" then I can see *some* vendors asking for "a bit more" for this "service". But, knowing that you have other options if his "adder" is too high:

- cut them myself

- find another house that will "cut the boards" cheaper (gratis) and give him the whole order.

Note that this does add another processing step -- he has to cut the panels into these "multiboards" if he wants to drill them stacked (depends on quantity and how tight your tolerances are) -- then cut them *again*.

Again, if there is something truly special about each individual board...

I would argue that they might be more likely to look at this as a "freebie" that they can give away for relatively low cost to win/keep your business. (?)

Hi D,

It's based on intent: The "average" customer they *want* (who's submitting one design on the board) might submit a greatest hole/part density board one day, but they're expecting he'll submit a least hole/part density board the other, or some combination so that, over time, it averages out. The customer who's combining designs himself, they figure, is likely to continually do so and therefore *always* have boards that are above average in hole/part density.

I would agree that if a board house is charging you based on a very detailed breakout of the design (number of holes, number of different hole sizes, sq. in. of board, sq. in of copper, aspect ratio of board, etc.), there's no longer any good reason to charge for multiple designs, but some board houses don't have the technology to do this automatically and therefore offer more "buffet" pricing -- particularly for prototype boards. (Advanced Circuits does this -- you pay the same for a prototype run of 3 PCBs that are 1"x2" as you do for 3 PCBs that are 4"x6"... and trust me, I feel a bit pinched in that my boards do tend to almost always run towards "tiny!")

At the end of the day, it is just a slightly different business model that may or may not be the most acceptable to any given customer. With restaurants, buffets have a market (PCB houses where up to a certain size and number of holes, it's all the same price, they don't care at all how many designs are on the board), traditional restaurants have a market (PCB houses that have a very detailed algorithm for coming up with a board price, charging "bit by bit"), and then there's the "hybrid" buffet (often found on cruise ships!) where all the food is free but the drinks aren't as a means of loosely coupling how much you pay to how much you eat (PCB houses that charge per design).

If the multiple designs are just a "step and repeat" (no different drills or copper usage) or relatively close to it and you used a reasonable shape, I would agree, and it's probably a successful tool. Many businesses seem to find it more successful to discount an artificially inflated price than to just offer a cheaper price in the first place (jewelry comes to mind...).

---Joel

g one

day,

her,

who's

y.

led

sq.

There are no significant differences in holes per area or hole sizes, since they are just different versions with different IC packages. Anyway, even if they charge extra for hole density, it is still much less than what they charge for multiple designs (a very subjective measurement).

ses

re

Yes if within reasonable limit. How can you justify 30% more for 4 designs and 150% more for 8 designs? Why are design #5 to #8 so much more expensive than design #1 to #4?

" as

that

t may

nts,

re on

very

"),

all

much

or

to

I would have to agree that those upcharges appear to be unjustifiable based on any real costs and are just an attempt to grab extra profit (...that they no doubt assume they're "losing" because you won't be coming back 8 times...).

Not a firm I'd want to do business with... :-)

We just got a batch of boards with a bad via, the same on all the boards. Our gerbers look fine and we paid for bare-board testing.

John

From?