yikes

.pdf

r 1

ce

s.

I can see a dot,

Arie, Fun, thanks. I've never worked in your 'world', but I've always felt that it's the constraints of a problem that make it interesting.

George H.

df

The ratio's work out OK if you stick with inverting amps. no? GH

Isn't on whatever device they sent me, and the Mouser listing has two pictures, a regular 16 pin SOIC and the wider package, they sent me the narrower one.

It is a special kind of fun and can be stressful, but very rewarding when it works out fine. Apart from making a perfect robust design (you can't fix 10K bad units) you have to be sure to always be able to source components, and deliver on time. I was part owner and main designer of a design house, when I lost a design to a competing designer, his was 5 euro cents cheaper (he used a PIC which I hated, I used Atmel). It was going into 200 thousand thermostats, so every cent counted. It was the only design I lost to a competitor in 45 years (just retired now). It is also a nice niche market - making professional stuff for consumer pricing. Lots of design houses make designs that are more expensive to produce.

Arie

Doing amazing things for very low cost is great fun. (Of course this requires a reasonably permissive definition of 'amazing'.) ;)

Cheers

Phil Hobbs

There are lots of ways to connect four resistors to an opamp.

The rule of thumb I read was "Invert when you can, non-invert when you must" due to common-mode distortion, the impedance of the input diff-pair current source is not infinite.

I don't know if this is still relevant for RR input stages, though.

Especially at AC.

More, rather than less relevant. RRI amps have two diff pairs of opposite polarity, and gradually switch between them someplace up towards the positive rail. They have different offset voltages in general, so you wind up with some gross CMR problems in the debatable land between.

Inverting lets you avoid that area. Of course then you don't really need RRI. ;)

Cheers

Phil Hobbs

Hey, those different pairs will have different PSRRs! To different rails!

Duh! Do people still say duh?

Well, RRIOs and other single-supply amps tend to have fairly horrible negative PSRR--after all, ground doesn't move. ;)

To be fair, I haven't checked on a lot of the more recent ones, so they may be better.

Cheers

Phil Hobbs

I use RRIOs with dual supplies! I should pay more attention to PSRR. But RRIOs aren't the amps one would select for low noise anyhow.

My applications probably average about 80 dB less noise critical than yours.

If you run a non-RRO op amp near its output limits its PSRR degrades too; like running a low-value resistor from the rail to the output the noise gets in and there's nowhere for the negative feedback loop to go to help it.

OK, in an RRO the fets are in their ohmic regions, similar issue.

I sometimes use them as comparators, but I assume the output fets are hard against the rails.

But they're 100 times faster, so it all evens out. ;0

Cheers

Phil Hobbs

Lately I'm groveling around at 400 Hz. And a 30 dB s/n ratio would be fine.

I'm sure you won't see mis-rotated parts anymore. I guess it was a small batch only.

Sounds like a good tool - which model and do you need a close up lens ?

It's obsolete now, an E45 with the close-up germanium lens. It will focus when the lens touches the part, and can image the hot spot on an

We got a freebie benchtop unit,

but it's fixed-focus, and the post gets in the way, so it's mostly useless.

Does anyone have an affordable imager that's good for close-up electronics?