vacuum tweezer thingie

That was one of the things I wanted to make sure worked.

STM32F2 microcontroller, using a 32MHz SPI interface in "TI" mode, and DMA into two alternating buffers. A timer output at 1/2 the frame frequency drove the ADC channel select toggle. Worked really well, once the registers are all setup the data just "appears" in the buffer with no further intervention. You can even halt the CPU in debug mode and the data keeps getting updated! First time I have actually made that work, only took couple of hours too.

The ADC worked really well as an ADC, too. It must be 14 bit inside, with two bits masked off. (And there is a 14 bit part with same pinout). No visible transition noise, datasheet quotes 25uVrms IIRC.

A welcome change from the on-chip MCU adcs I've been used to recently. The STM32F2 on-board ADC has at least 22lsb p-p noise!

Yes, I linked to one in my reply to krw: .

Didn't want to wait weeks for it in this case.

include

Keep meaning to do that (for past 5 years....)

We've found the ADCs in the NXP ARM chips to be pretty terrible, even if you oversample and filter out the noise.

I like the ADS7866 for simple stuff. It's a SOT23, a 12-bit SAR that's clocked by the SPI readout. Just read it whenever you like.

include

Someone in this ng, with some spare time, could accept requests from various people for little circuits. They could consolidate them into one board layout, order some boards, shear them up after receipt, and distribute the bits. It could be reasonably priced to the clients and fairly profitable to the guy who does it. They could wind up with a sellable series of adapters/filters/whatever in the process, basically for free.

rote:

the

hode

u might

you

ocked

...

ld include

ous

yout,

It

y who

atever

really no need to put them all on one board, several that do that all ready

places like seeedstudio, iteadstudio, elecrow will make 10 pieces

-Lasse

might

clocked

include

layout,

who

adapters/filters/whatever

Do they include the layout, from a schematic sketch, too? That's what I had in mind.

the

of the

cathode

you might

ce

the

if you

clocked

hould include

arious

layout,

s. It

guy who

/whatever

ad in

ok, that makes more sense,

I thought you meant getting a bunch of gerbers from people and putting them on a panel

elcrow does 4 Layer 10 * 10cm Max - 10pcs for 60$

-Lasse

These are great for cutting copper clad.

cathode

might

you

clocked

include

various

layout,

who

adapters/filters/whatever

in

elcrow? Link please?

Thanks

Phil Hobbs

d the

e of the

d cathode

nd you might

and

once

ith

d the

rk,

n if you

's clocked

=220...

should include

various

rd layout,

its. It

he guy who

rs/whatever

I had in

missed an 'e'

and i misread the 60$ is for 10*5cmx5cm, 10*10cmx10cm is 95$

is about the same price

-Lasse

Yes, looked at those along with all the others but too slow for this application. I needed the two channels at 1MSPS each. TI has the best parametric search page for this, select >=1MSPS, >=12 bit, sort by price, done. Of course picking the cheapest risks getting junk but it turns out to be a beautiful little chip for what it does, $1.95.

In fact I found nothing comparable elsewhere; ADI ones get expensive in multi-channel, LTC are expensive period, intersil can't do it, and I don't know any other precision analog manufactures come to think of it.

Oh there is Maxim, had them blocked mentally. MAX11102 looks similar.

what a mess...

I did some stuff on a transistor insulator pad once. There are a couple 0402s on there.

Cool.

Building on copper-clad is generally better for most things, signal-integrity wise.

Yeah, well, don't look back! I can't IMAGINE trying to manufacture what I'm doing now, manually! UGH! But, then I had to learn how to optimally use the P&P machine, make my own solder stencils that actually give the RIGHT amount of solder, deal with the change to lead-free soldering, deal with the change from DIP to SOIC to SSOP and ever smaller components and finer lead pitch.

Jon

Yup, I did a board with the AD CMP604, only available in that 12-pin chip-scale package. I never got the soldering system to work on those, and we ended up changing to a different part with a more traditional leaded package. The package was not the ONLY reason for the change, but was part of the reason. We got lots of solder bridges under the chip, and when a bunch of power rails are shorted at some number of the 64 parts/board, it was a bear to find.

Jon

I actually quite like the QFN packages (that was a 16 pin one). Must be good from a noise point of view, with the substrate is connected to the thermal pad. They take up significantly less space than the same pitch QFP. They can be hand soldered (not so easily but not too bad). Just a pig to prototype, in the way I did there anyway. But a QFP would be worse I think.

I have had good results in production with a 40 pin QFN, no defects that I recall in >1k units. Better than a QFP for us.