How do you design these days?

Did it on a recent project: Client wanted five units to kick the tires. The difference between buying 5 bare boards and 50 boards was miniscule. So now they can make 45 more unit and the PCBs will be almost free, they just have to pay for parts, stuffing and testing.

It is, too :-)

The truth lies somewhere in between. I use whatever seems most handy given the situation. For analog I use simulation a lot but it always gets prototyped. SMD components either end up as a dead bug on prototyping board or a small dedicated PCB. RF and power means having a small PCB made. Sometimes I like to try things like a new microcontroller. I usually make universal (=a load of via's to connect wires to) boards out of such a project so they can serve as a kickstart for prototyping other projects.

...

How do you deal with "Locate capacitor as *close* *as* *possible* to the regulator output and ground pins"? (I know, you try to avoid using LDOs)

How do you deal with incorrect/optimistic data sheets? I had more than one sensor (mostyl accelerometer/magnetometer), that behaved different from what the manufacturer hoped it would.

Another problem is, that errata are written *after* the error was found.

I prefer to have seen a design working as expected. But I am only a software guy, who knows where the soldering iron's hot end is, anyway ;-)

That *is* cheap.

But I prefer etching my PCBs to using breakout boards. Can save time.

Guten Rutsch ins neue Jahr, Falk

On a sunny day (31 Dec 2009 14:12:03 -0500) it happened DJ Delorie wrote in :

theftp://panteltje.com/pub/z80/graphics_card_top.jpg second OLED

Real man use no PeeSeeBee: ftp://panteltje.com/pub/z80/graphics_card_top.jpg ftp://panteltje.com/pub/z80/graphics_card_bottom.jpg :-)

On a sunny day (Thu, 31 Dec 2009 12:18:44 -0800 (PST)) it happened " snipped-for-privacy@bid.nes" wrote in :

Yes I had that happen, where the specs changed after each council meeting.. Something to do with fire rules. But I did not have the board made until they stopped changing.

I grew up with RF-style design and EMC jobs so that really isn't a problem for me. The layouter I often use is also a bit on the older side and very experienced so I rarely have to tell him what needs special attention. Plus we are politically on the same wavelength which helps :-))

In my module specs there is always a section called "Layout Guidance". That has all the caution notes in there. And no, I do not use LDOs.

With reputable manufacturers that is rare but I had one of those happen recently. LT6700 series comparators. Turns out there's a bug on those in that the reference shoots way up after applying power and then settles after 500usec. In a nutshell, those things just don't work right until

I am a hardware guy who sometimes finds a workaround solution in software and then says "Oh, it was just software" :-)

Yeah, in Euros ...

I often design with chips like the LM3478. When you add a $2.50 adapter to each $1 chip it kind of adds up. I can get two bottles of top notch Porter for that.

Nice. Very smart design with the sliding scale QFP pattern. But being an analog guy I usually need TSSOP and MSOP.

Same to you. And careful with your Enfield motorcycle on the ice and snow.

theftp://panteltje.com/pub/z80/graphics_card_top.jpg second OLED

I think I am going to get sick ...

theftp://panteltje.com/pub/z80/graphics_card_top.jpg second OLED

SOS: System on a ..... shingle ?:-) ...Jim Thompson

amalgamate

fraction

The real problem is when they want it fixed, without changing anything. That's when the real churn starts. Compliance added another to a recent design because nothing could be changed in a previous "fix". Risk avoidance can itself be risky.

amalgamate

"Projects?" "One-offs?"

Backwards pots are code changes. ;-) (we use shaft encoders for pots).

Spec changes don't count. That's management's fault. ;-) You do point out one of the main pitfalls, though. A good spec is critical to getting success out of the first board. Changing the rules after the game starts is cheating.

IME the sheet metal always gets done before the circuits. The sheet metal fixes the number of blinkin' lights and fancy knobs pretty well.

Code.

"Whenever more than one person is involved in a screwup, blame will never be placed." Or "The first to blink, loses." BTDT, it's no fun.

On a sunny day (Thu, 31 Dec 2009 14:54:10 -0800) it happened Joerg wrote in :

Have to point out that that worked without a single problem for > 10 years. And it will likely still work if the EPROM has not lost data. But it has been superseded by better tech. You are not afraid of soldering some wires no?

PC

Just x-acto. The magic trick is to then rub it hard with a Scotchbrite pad. That removes the burrs.

John

Many

PC

Wow, that is a precise X-acto cut.

Not at all, but I use thin copper-enameled wires and run them in orderly fashion :-)

Jan Panteltje schrieb:

Why did I know that I would see an 6845 when it completely loaded?

That's the annoying thing with this kind of "implementation": It often just works.

My first CP/M computer was made the same way, but I used enameled copper wire[0] on the solder side only. My brother wrote his first 100+ job applications on this beast (Wordstar + DBase).

Falk [0]I still use it:

made

Many

is it

to a PC

ll

d in

t to

It is very hard to hook a spice model to the real transducers. Making a spice model of the transducer isn't so easy when it is pushing the limits of physics and you have to assume that the physics guy got it exactly right when he did his equations.

Very few op-amps have truly accurate spice models. They don't cover things like the recovery from hitting the rails or slew rate limits. They almost all beam electrons in from some other universe.

Models of inductors are not super accurate. Even the ones that cover saturation are only approximations.

If you have to write software to go with the hardware, it is very nice to debug on something like the real hardware.

On a sunny day (Thu, 31 Dec 2009 15:52:38 -0800) it happened Joerg wrote in :

Bu the vapour of the enamel is poisenous :-). I tried that stuff, it stinks. Normally I use flat cable: ftp://panteltje.com/pub/z80/sound_card_bottom.jpg ftp://panteltje.com/pub/z80/soundcard_top.jpg That flat vertical thing is a R2R network, the DAC. Top right an electret mike, Bottom right a DC-DC converter for the negative voltage of the TIL084s Audio amp: LM380, ADC: ADC804 SAA1099 for sound effects... (did not really use that, used wave tables). the resistors and caps in the IC sockets are part of the anti-aliasing filters. Also wrote a sound editor for it, all 8 bits of course.

I think I have similar boards somewhere but enamel wire is the way to go.