Arghhh

Is the mouser distant from you? That's often a disaster. I prefer to keep the layout-er in the next cube, and stay close.

Best next thing is to have a copy of his/her software and shoot the files back and forth often. I usually do the initial placement and the critical routing on, say, one channel, and stay close to the layout as it progresses.

Some layout people place parts based on the rubberbands. That's usually a disaster. They have to place based on the schematic.

Layer-layer crosstalk, trace widths, mechanical, thermal, pinouts... too many ways to mess up.

I've always found women to be the best layout people. They seem to listen better.

John

I've worked with folks who have their layout done remotely, and it can work very well. It can be a disaster, too, but it _can_ work.

Possibly because their manliness isn't threatened by "having to do what the @#$% circuit designer said".

Male or female mice?

Just had a slightly similar issue with a customer--an engineer who was on his way out the door ditched the 0603 metal film resistors in favour of 0402 thick films without telling anyone...>30 dB noisier down in the

There was plenty of room for the 0603s, but nooooo, he had to be creative.

:(

What a maroon. There may be enough pad space to bodge in 0603s for the demo--I sure hope so.

Cheers

Phil Hobbs

No 0402 metal films out there?

That's the kind of lesson that one needs to learn by getting burned -- it's certainly not something that you're going to learn in school, and if you do it'll be one sentence in the middle of a 3rd- or 4th-year circuits class, and you won't even write it down in your notes because it won't be on the test.

At times like these, it's best to remind yourself that you get paid by the hour (you _do_ demand payment by the hour to fix customer's screw ups, yes?), and the more they screw things up the more guaranteed income you get.

Not above 100k, whereas we need 499k-750k.

Not this guy. Both I and the main technical honcho there were very very clear about using metal film only.

Unless they disappear because they miss their milestones. Besides, these guys are smart, have great technology, and work really hard, so they deserve to succeed. (Besides, I'm constitutionally unable not to root for my customers. Waste and failure offend me deeply, especially when it's so unnecessary.)

Current plan to make the demo milestone is to put 0603s on edge, diagonally across the pads.

Growl.

Cheers

Phil Hobbs

Get some of these!

John

Thanks. Those only go up to 100k, unfortunately, whereas there's a photodiode nonlinearity issue that we have to work around, hence the low photocurrents and attendant high resistances. There are the RG1608 and RGH1608 Susumus in 0603, which go up to 300k or so (1M theoretically, but nobody stocks them). Mouser has some 400k-ish ones.

We might be able to fit 0805s on there if we really hold our breath.

Cheers

Phil Hobbs

Saved by an ace technician--he figured out how to get 0805s in there for the demo. :))

Cheers

Phil Hobbs

I once designed a little board with about a half-dozen chips, gave the schematic to one of the assemblers, and she followed the schematic so accurately that she put a half-dozen capacitors in parallel, right next to each other, at the power inlet to the board.

I explained that that's just a schematic convention, and it was supposed to be one cap per chip; I didn't have to revise the schematic, but it took her about 1/2 hour to add the other caps. She was also kinda relieved that since they're just bypass caps, the bank of them at the connector would do no harm, and could be left in place. ;-) (it was a one-off, so no big deal.)

Cheers! Rich

Lay the resistors in on edge; saves horizontal space and trace length may be more than sufficient to compensate on length.

Can you p+p and reflow resistors on edge?

John

I do not think a P+P machine can "turn" a part, i think it can only rotate based on how it is in the tape reel. Reflow works, once placed..

Any yaw, no roll or pitch. ;-)

Often it is the tool that does the rubberband thing, by default.

Does your PWB softwares recognize this?

Floorplanning in PWB software usually sucks. Probably the same for FPGA software, get you to buy a "bigger" part.

We don't auto-place. That always makes a mess. We don't auro-route, either.

In FPGAs, you really don't get a choice. You can sometimes, if things are dire, go in and do some manual floorplanning for the critical stuff. But mostly, if you ever want to ship stuff, you have to let the tools do their thing. The Xilinx stuff will use random seeds to try different layouts, and sometimes it gest lucky. Sometimes it does incredibly dumb sruff.

John