disgusted, back from Selem

[yaddatrance]

You might want to check the bandwidth again... We looked at the SSM2241
when we were looking for simple methods for driving differential signals,
unfortunately those didn't work for several reasons.. first they had an
interesting unity behavior where when you fed a 5V signal into it, it would
give back 5.6V back (and -5.6)... the second, and most important thing was
that their slew rate was much slower than one would expect...When we
scoped it, it seemed to prefer drawing 45 degree lines with standard
graphics...

Doc, the requirements might be different, we often do not use our own
lasers, so we need to rapidly set tuning based on which system we're
attaching to... the converter needs to be as close to the laser as possible
for the compensation to be as accurate as possible. we looked at using scale
and offset, but digital pots were too expensive. and I like being able to use a
lookup table to set gamma, etc... we're also not using a classic DSP though,
we're using a fast general purpose micro, which are much easier to deal with
when simultaneously juggling multiple channels (unless you want to use an
expensive DSP)... The only requirement for proper behaviour is sampling at
2.2x the desired frequency... In our case, (for simplicity) we sample at a hair
over 264kHz (not exact because we wanted to use an even clock muliplier)
and output is very simple and clockedat a constant 120kHz... quad and even
octal dacs tend to be surprisingly cheap... We don't have to worry about the
rate the color is being changed,we just need to examine behavior of the laser
itself and feed it what it wants.

[mixedgas]

there is/Was a ilda standard diff driver - receiver I will drag it up over the weekend and scan it. Before the 25 pin Db revision the ISC (ilda standard connector) was a 30 pin CPC Series 5 plastic amphenol, and in that standard was a 2 opamp driver circuit and 2 opamp receiver circuit. So there is a standard reccomended circuit.

Somebody get me a gallery, I have no web hosting of my own now.
I'll start posting schematics like there is no tomarrow if I have space.

Space, the final frontier, to boldly scan where no one has scanned before...

Sorry, couldnt resist.

Seasons Greetings, and "Bah Humbug!" if your a grouch or atheist.

Steve

[yaddatrance]

Oh yeah! before I go off on a tangent, Doc, that's a good circuit...

Very few systems use differential color anymore, I've only used a few
of them (like the NEOS 8 channel driver) and even those are very
finicky about ground paths... I believe at this stage, most everybody
has the nominal -2.5V lines actually tied to ground (effectively forcing it to
be single-ended aka non-differential output) because then you don't have to
worry about how the system builder wired the color channels internally. I'd go
as far as to say it should be safe to use on everything, since a true
differential input receiver should handle one leg tied to ground.

For differential driving scanheads from a single-ended source, I usually use
some form of this circuit... Make sure that ground is referenced from the laser
board's ground, if ground is left floating, behavior is undefined.

[The_Doctor]

..and "Bah Humbug!" if your a grouch or atheist.

Steve

Thankyou kindly, good sir. Likewise.

Yadda, cheap DAC's might help get round the pot problem. Was going to suggest digital pots but you already tried those, I see. If DAC's are cheap, they can control FET's. The control range need not be high res because that's taken care of by the analog scaling on the main DAC out. Digital pots are pay-for-convenience, plus perhaps a floating pot that is isolated from input control signals. If you can work round that so it doesn't matter if they share common ground, you can get by with the cheap DAC's and FET's to control the scale amps. And you even get your look up tables.

My circuit actually has a PROBLEM. Daft not to see it before. It inverts. I'll replace the self-hosted copy of it soon, with a non-inverting version. It's meant to run on a single supply. I was so busy thinking I needed attenuation (noninverting amps can't scale below unity) that I failed to see that I could go back to noninverting, having accepted the simple passive scaling resistances to get the advantage of a single stage.

[The_Doctor]

Ok, fixed, as is previous post:

It's a passive averager, feeding a high impedance x2 noninverting amp, so the error will be set by the tolerance of the resistors used (likely 1%). Operates off a single supply. Scale and offset controls are independent and easy to set. Start from zero, set offset for threshold, scale for full output.

For reference, the old design:

Must use a dual rail supply. Might be useful, but I can't quite see how.