stability for an op-amp integrator

[JeffM]

you're still a plagiarizing hack

People who can't be bothered to do that sort of research
are called nit-wits, or "inventors".
Bill Sloman

Restated (I keep this one over my desk):
A couple of weeks in the laboratory
can save you a few hours in the library.

 

[Fred Bloggs]

Plagiarists are writers who copy other people literary efforts.

That should be copy with the purpose of passing it off as their original
work. It is ironic that you should be married to a relatively elite
linguist and have so little of that ability yourself.

Engineers who copy successful circuits are called experts who have
done their research.

Engineers don't "copy" circuits, they use established and proven
reference designs with more or less modification. This implies an
element of understanding that is absent in the mere act of copying.
People who copy are engineers of questionable ability and unquestionably
not experts.

People who can't be bothered to do that sort of research are called
nit-wits, or "inventors".

Get real- engineers with a good understanding of their field know when
it's time to dig and invent because they know there is no established
technique.

I get very nervous if there is a known good
solution to a problem, and I find that someone has invented a "better"
solution.

Yes- well I have observed that you are rather stubborn about certain
things- like your failure to recognize and accept the value of Walsh
function shaping of synchronous detector low pass filters.

 

[Bill Sloman]

That should be copy with the purpose of passing it off as their original
work. It is ironic that you should be married to a relatively elite
linguist and have so little of that ability yourself.

No more ironic than if I were married to an Olympic class runner and
couldn't myself run a mile in less than five minutes.

Engineers don't "copy" circuits, they use established and proven
reference designs with more or less modification. This implies an
element of understanding that is absent in the mere act of copying.
People who copy are engineers of questionable ability and unquestionably
not experts.


Get real - engineers with a good understanding of their field know when
it's time to dig and invent because they know there is no established
technique.

Exactly. I've had to do it from time to time and it does tend to
stretch development times.

Yes- well I have observed that you are rather stubborn about certain
things- like your failure to recognize and accept the value of Walsh
function shaping of synchronous detector low pass filters.

I've yet to find a text which gives me any clear idea of what Walsh
functions are good for. It's a question of ignorance and unfamiliarity
rather than stubborness - I've looked at Walsh functions from time to
time, but I've not yet had occasion to get serious about using them.

 

[starfire]

Thanks for your reply.

It was what I would consider "constructive" input.

....by the way, when I said "constructive", I was just hoping to reduce the
number of replies of the sarcastic nature. It's getting very difficult
anymore to phrase a post where it doesn't violate someone's sense of
morality or decency... Even ax-murderers have a sense of "morality"

Dave

 

[starfire]

Thanks for those part numbers!

I've done quite a few Google searches for parts and hadn't seen those
particular units. While I don't think it will work in my particular case (I
need a bipolar power supply because the output ramp will be negative-going
before I invert it), I do have other applications that are unipolar and
would definitely benefit from these parts.

Thanks again.

Dave

Have you looked at these?
http://www.fairchildsemi.com/ds/NC/NC7SZ66.pdf

These are really amazing:
http://www.onsemi.com/pub/Collateral/NLAS4684-D.PDF


Best regards,
Spehro Pefhany

http://www.speff.com

 

[starfire]

What you're suggesting was the first approach I tried but 12 bit resolution
wasn't considered adequate. The physicist I'm doing this for says he wants
total analog output and no stair-stepping. I suggested doing it with a
16-bit DAC to reduce the size of the stair-step but that idea was rejected.
I have no choice but to approach this with an analog solution.

Thanks.

Dave

 

[John Larkin]

Thanks for your reply.

It was what I would consider "constructive" input.

...by the way, when I said "constructive", I was just hoping to reduce the
number of replies of the sarcastic nature.

So, ironically, that's exactly what you got.

John

 

[starfire]

Good point... I'm sorry I didn't mention the required precision. It was
stated as being 0.0375% differential linearity over the full range output of
8V in 35ms. That translates to about 3mV full scale differential from one
ramp to the next. That's pretty tight!

The charge injection problem has me a little concerned, also. According to
a paper written by Maxim, the charge injection goes up with the reduced "on"
resistance. This being a result of the increased area of the die decreasing
the resistance but increasing the capacitance.

Thanks again.

Dave

 

[Ian Buckner]

I've yet to find a text which gives me any clear idea of what Walsh
functions are good for. It's a question of ignorance and unfamiliarity
rather than stubborness - I've looked at Walsh functions from time to
time, but I've not yet had occasion to get serious about using them.

ISTR Walsh functions were used way back for minimising
crosstalk in long multiple cable runs in telephone trunking.

Regards
Ian

 

[Ian Buckner]

What you're suggesting was the first approach I tried but 12 bit resolution
wasn't considered adequate. The physicist I'm doing this for says he wants
total analog output and no stair-stepping. I suggested doing it with a
16-bit DAC to reduce the size of the stair-step but that idea was rejected.
I have no choice but to approach this with an analog solution.

Thanks.

Dave

<snip>

I find it ironic the physicist demands no quantising in nuclear
measurements ;-)

Regards
Ian

 

[Ken Smith]

Hi all -

I'm working on a project that requires a stable, repeatable integration of a
35ms, 4.1VDC pulse input to produce a negative-going ramp output.

Are you sure that this is what you really want to do? From what you
described elsewhere it sounds like this is part of a larger system. It
sounds quite a bit like some sort of anti-coincidence circuit for a
spectrometer.

Since the amount of time involved is fairly long, how about using a
digital counter as the "ramp". Digital counters are very-very linear and
quite repeatable. The counter could count a 10MHz crystal oscillator.
This would give a step size of 0.1uS out of your 35mS.

 

[Ken Smith]

The charge injection problem has me a little concerned, also. According to
a paper written by Maxim, the charge injection goes up with the reduced "on"
resistance. This being a result of the increased area of the die decreasing
the resistance but increasing the capacitance.

The good news is that the charge injection of two devices in the same
package are usually almost equal. You can add a second switch and
capacitor combination on the inverting pin of the op-amp and make the
injected charge nearly cancel.

 

[starfire]

The 35ms period is the nominal ramp period but the ramp is to be reset upon
receipt of an external t0 pulse and a new ramp begun. You're right in that
this is part of a larger project. I'm only privy to the pieces given me,
though, and really don't have any idea of what the "bigger picture" is.

With the approach you're talking about (0.1us step size), I assume you have
to take the counter output and run it through a D to A converter? If that's
the case, I've already been told to not implement it with a D to A. That
was my first approach.

Thanks.

Dave

 

[Cornholio]

Dave,
I think the voltage follower on the input may be unnecessary, as it
may not handle the pulse well, due to its limited bandwidth.

I'm not sure what amplitude your ramp needs to reach, but this large
resistor and cap value will only give you milivolts output for a 35ms
pulse.

Metal film resistors give you less noise.

This sounds like a decent capacitor, however, you will want the lowest
dielectric absorbtion rating possible. Something like a hermetic
sealed polystyrene is best.

More details about what kind of output you need would help.

Also, a 1.82M resistor from the non-inverting input to ground with an
op-amp that has 8nA input bias current will give you 1.82e6 * 8e-9 =
+/-14.5mV at your +pin.

You will probably want to use a smaller resistor in the input and
nothing on the n/i pin.