Op Amp Selection

[rstofer]

I am looking for suggestions about op amp selection for an analog computer.  The signal swings over a 20V range in either 1 second of 2 mS depending on the time base selected - slew rate is not an issue nor is high frequency response.  Dual 15V supplies with signals swinging over +- 10V.  Rail-to-rail is not a criteria.

Low leakage seems important as is high gain and high input impedance.  This will be a redesign so having to correct for offset voltage is doable but wouldn't it be nice if I didn't have to?

I am currently using an obsolete device, the TL081.  It seems to work ok but the range of offset voltage seems high.

Cost is not much of a consideration because I only need a dozen or so, if that.  I am presently using socketed 8 pin DIPs and I imagine I will stay with that form factor simply to allow possible interchange.  Soldering SMD is not a problem, all of the other components on the PCB are already SMD.

A side issue:  Anybody have a source for low leakage, high accuracy capacitors?

[void_error]

I'm clueless about analog computers but I can make a general suggestion: go to Mouser and/or Digikey (or other distributor) websites and use the filters to narrow down the options under the op-amp category. Alternatively just browse for op-amps on the manufacturer's (Texas Instruments, Linear Technology, Analog devices, etc.) and use their parts filters to find any suitable candidates. Then check if any of the distributors most convenient for you have those parts in stock.

[Edwin G. Pettis]

Linear Tech makes several op-amps that meet your requirements, the LT1056 and similar ones are available in DIP form and low cost, you can check the LT site for other similar op amps.

Polypropylene capacitors are the best type from a cost and availability view, many manufacturers make them.

[rstofer]

I'm clueless about analog computers but I can make a general suggestion: go to Mouser and/or Digikey (or other distributor) websites and use the filters to narrow down the options under the op-amp category. Alternatively just browse for op-amps on the manufacturer's (Texas Instruments, Linear Technology, Analog devices, etc.) and use their parts filters to find any suitable candidates. Then check if any of the distributors most convenient for you have those parts in stock.

I guess I'm just not smart enough about op amps to know which questions to ask or which parameters to filter.  I suppose if the word 'Precision' shows up, that is a good thing but the range of op amps seems overwhelming.

My primary application is, of course, integrators.  Nice, linear and with little to no leakage.  Leakage messes up long running simulations, eg generating a circle with two integrators and an inverter will show a declining radius over time.  Well, sure, energy is lost to the environment and not replenished so, yes, there will be leakage at the system level.

I found an old article (50 years old!) written by one of the founders of Analog Devices that discusses in great detail the problems with integrators.  Maybe when I digest it, things will become more clear (or not...).

ftp://ftp.analog.com/pub/cftl/ADI%20Classics/Ray%20Stata%20Articles-1960s/Operational-Integrators.pdf

Analog computing is a kick!  I have built up some modules to use with LTSpice and MatLab includes everything necessary but it's just not the same as patching up a circuit to analyze damped harmonic motion or just a falling object.  I really wish we had had analog computers when I was studying differential equations.  It would have made the subject more 'visual'.  Somehow, an exponential decaying envelope over some sin() terms just doesn't mean as much as a trace on a scope.

[rstofer]

Linear Tech makes several op-amps that meet your requirements, the LT1056 and similar ones are available in DIP form and low cost, you can check the LT site for other similar op amps.

Polypropylene capacitors are the best type from a cost and availability view, many manufacturers make them.

I'll look at LT along with TI and ADI.  It's odd that the LT1056 datasheet pans the plastic packages, highly recommends the TO-5 package and then they turn around and obsolete the TO-5!  Just one of those ironies you run across every once in awhile.  Datasheet page 10, 3rd full paragraph:

http://cds.linear.com/docs/en/datasheet/10556fd.pdf

But what they are saying about the JFET inputs would apply to any manufacturer, I would suspect.  Just something else to consider.  Maybe I should look around for TO-5 packaging.  I would expect it to be scarce.

I have been using Polypropylene capacitors and they work quite well.  And they are available up to an beyond my requirement of 1 ufd.  But they are only 1%.  That's pretty good but, in the old days, the integration capacitors in analog computers were packaged in a heated enclosure and capable of 0.1%.  Sort of like oven controlled crystal oscillators.  Out of my price range!

[Kleinstein]

In an analog computer you may need different OPs, depending on the part they are used for. For a slow version input bias is important in many cases, but not all.
Good candidates should be OPA140 or the cheaper OPA141, OP177/OP277 and as a low cost version maybe OP07 in cases where bias is not that important. If money does not matter, LTC2057 is good where low drift is needed.

Integrators can sometimes use a combination of a lower voltage AZ OP and a second driver (e.g. LT1052 or AD8551 and NE5532).

Not many OPs found in metal case like TO5 - there are other options.

For the caps you might need to measure some - low tolerance caps are rare and expensive.