Group Actions
Engagement
  • Author Author: ralphjy
  • Date Created: Date Created
  • Views 6125 views
  • Likes 16 likes
  • Comments 15 comments
Related
Recommended

Integrated Op Amps - the early days

ralphjy

Great to see a Project14 focused on Op Amps: Op-Amp-a-Palooza.

I don't have time to do a clever and innovative project, but looking forward to see what other members come up with.

It got me to thinking about when I first started out using IC Op Amps back in the early 1970s.  First started using the uncompensated 709s and 308s and then quickly moved to the compensated 741s (1MHz was sufficient for most of what I was doing and these were inherently stable, although the early ones from TI were prone to latch-up).

I have a collection of vintage parts (709, 741, 747, 558).  I was surprised that I did not have some 308s - probably stashed somewhere else.  Back in those days the parts were sometimes hard to get at a reasonable price, so I resorted to buying unmarked and sometimes untested parts and testing them myself.  I would use a Sharpie or scribe to mark the parts after I tested them.

Here's a picture of some of my collection:

image

A range of vendors and styles.  Early parts were packaged in metal TO cans or ceramic DIPs (probably still are for military use) and later in plastic packages for cost and ease of manufacturing.  The parts here are from Fairchild, TI, Raytheon, Signetics, and National.  The earliest date stamps are 6916 for a National LM709 and 7016 for a TI SN72741.  The unmarked parts pre-date those.

Those were the wild west days - I remember that we would use Motorola MECL digital ICs and would bias the power supply so we could use them as linear small signal differential amplifiers - the precursor to using these op amps.

If I find some time, I'd like to try some of these out again...  

  • in reply to Jan Cumps

    I was wrong. You're not using the 'false summing node' method, you're measuring the loop gain and phase directly by injecting a signal, with the op amp biasing itself.

    That might be okay for doing a bode plot of at least some of the open-loop ac gain and phase [which is interesting in itself and very much worth a blog], but I think you'd have considerable problems extending that down in frequency towards the dc gain [the ac gain won't achieve the dc value until you get down to a few Hz, even with an old part like a 741].

    At the higher frequency end you'll need to be careful that the slew rate limitation of the output doesn't mess up your measurements. The bode plot that you see in a datasheet will be for the small-signal gain.

  • in reply to jc2048

    Just a scope capture- not interpreted, not checked

    image

    Frequency 2 kHz. I used averaging in the acquisition to filter out the noise. On channel 2 I also used 20MHz BW limit.
    Output inverted at low frequencies - 90° phase around 1 MHz. Around 1.6 MHz the two signals are in phase.

  • in reply to Jan Cumps

    ...but that turns out to be quite difficult.

    I should have added that I was trying to do it using the approach in that Analog Devices article, with a second opamp integrator to servo the first. This is what I get at the output of the DUT

    image

    The loop oscillates nicely. I got rid of that by adding some resistance in series with the integrator cap, but there are still two issues. One is that the DUT amplifies the low frequency noise at the input, so the noise 'walks' the measurements around, the other is that there look to be various long thermal time constants internally in the LM301 that take a long time to settle and all that makes getting good readings difficult.

    Perhaps I should give up and try the false-summing-node approach that you're using.

  • in reply to Jan Cumps

    This is a setup for the 3rd option. I replaced the 100R with 50R to match generator impedance. The 3 other resistors are 1K9 because I have no 1K's left.

    image

    The IC is a ST ua741, I have a lot of those