The attached diagram is for an op-amp circuit for a seismograph that records earthquakes at Griffith Observatory in Los Angeles.

The ground motion sensor is shown in the first image and it has a magnetic variable reluctance transducer where a mass is suspended from a spring and the displacement between the mass and the frame varies the air gap between a magnet and two iron cores with multiple output coils.

The output of the transducer is amplified to drive the pen galvanometers on a rotating drum recorder like the one shown in my avatar.

The diagram shows a four stage amplifier composed of four 741 op-amps with a "compressor amp".

However, what does the compressor amp do?

 

The compressor is the lamp and photo-resistive cell in the feedback of the second stage.
As the signal increases in amplitude, the lamp increases it's light output which lowers the resistance of the photo-resistor, reducing the stage gain to keep the signal from overdriving the galvanometer.

 

It gives a log response instead of a linear response. The bulbs and ldr cut the gain at higher input.

Edit: all my time and effort and I'm beaten to the punch.

 

OK - so it's a form of automatic gain control (AGC) to keep the pen from being driven off the edge of the chart paper.

 

Yes, it's an AGC circuit- but how does the thing display a meaningful value if it's got an AGC?

The scale factor is then unknown?

Interesting!

 

OK - so it's a form of automatic gain control (AGC) to keep the pen from being driven off the edge of the chart paper.

Sort of.
But there's no filter in the gain control so it will respond relatively rapidly to signal amplitude and, given the low frequency of seismic waves. likely acts more like a limiter (compressor) to the peaks than an AGC (which is normally designed to have a constant gain over a complete waveform cycle).
So it would generate sort of a log of the peaks as GopherT suggested meaning the display would be relative, not absolute.

 

There are some problems with that diagram.
The circuit shows the lamp with a 100k pot in series with it. That is much too high a value for either an LED or an incandescent lamp.
There is no connection shown for the non-inverting input of the compressor amp.
The timing marks are fed through a 100k resistor into the very low output impedance of the '741. They would be very much attenuated.

 

However, what does the compressor amp do?

To expand thee dynamic range of the circuit. Essentially lowering it's gain. Typically done with a log amplifier.

In this case, the lamp plus photo cell do the job.

 

Yes, it's an AGC circuit- but how does the thing display a meaningful value if it's got an AGC?

The scale factor is then unknown?

Interesting!

This seismograph is just for a public display about how earthquakes are recorded and it is not intended for scientific use.

The Los Angeles area often experiences local earthquakes (epicenters less than 100 miles from the observatory) which could drive the pen off the chart if AGC was not used. However, the instrument is also used to display earthquakes that occur 1000s of miles away which usually don't deflect the pen full scale. Hence, with AGC, both local and very distant seismic events can be recorded. However, AGC removes the ability to gage the magnitude of these events.

Seismographic stations for scientific analysis are located in "Far Field" which is a very large distance from the regions where the earthquake epicenters are located. Therefore, the ground motion at these distant stations will usually be within the range of the instrument and amplifiers with constant gain are used.

Also, computerized recording enables a very wide range of motion to be recorded on a given size scale without regard to the mechanical limits of the old fashioned chart recorder. Accelerometers are often used to detect ground motion and the output is integrated twice to obtain the actual displacement. With accelerometers and computerized recording, a ground displacement of 10 feet or more can be displayed within the size limits of a flat screen.

 

This seismograph is just for a public display about how earthquakes are recorded and it is not intended for scientific use.

The Los Angeles area often experiences local earthquakes (epicenters less than 100 miles from the observatory) which could drive the pen off the chart if AGC was not used. However, the instrument is also used to display earthquakes that occur 1000s of miles away which usually don't deflect the pen full scale. Hence, with AGC, both local and very distant seismic events can be recorded. However, AGC removes the ability to gage the magnitude of these events.

Seismographic stations for scientific analysis are located in "Far Field" which is a very large distance from the regions where the earthquake epicenters are located. Therefore, the ground motion at these distant stations will usually be within the range of the instrument and amplifiers with constant gain are used.

Also, computerized recording enables a very wide range of motion to be recorded on a given size scale without regard to the mechanical limits of the old fashioned chart recorder. Accelerometers are often used to detect ground motion and the output is integrated twice to obtain the actual displacement. With accelerometers and computerized recording, a ground displacement of 10 feet or more can be displayed within the size limits of a flat screen.

Please stop saying it is AGC. It is not. The whole concept of the Richter scale is a log response. An earthquake of 8 is about 31x more powerful than a 7. The compression stage gives the log response to the amplification to mimic the desired scale.


From Wikipedia (Richter Scale)

• Magnitude 3 = 2 gigajoules
• Magnitude 4 = 63 gigajoules
• Magnitude 5 = 2 terajoules
• Magnitude 6 = 63 terajoules
• Magnitude 7 = 2 petajoules

 

I was considering replacing the bulb with an LED. However, the LED will not turn on until the voltage reaches the forward conduction threshold of the diode which I believe is 1.23 volts for a gallium arsenide LED. Therefore, a separate LED would have to be connected in parallel with the LED in the feedback loop and connected to the power supply through a dropping resistor. This arrangement would be a voltage divider that would bias the diode at cutoff and essentially act as a "current mirror".

Or are there opto-couplers that are a completely package device that provides a light output in linear proportion to the voltage in the feedback loop and shines the light directly to a photo transistor?

 

I was considering replacing the bulb with an LED. However, the LED will not turn on until the voltage reaches the forward conduction threshold of the diode which I believe is 1.23 volts for a gallium arsenide LED. Therefore, a separate LED would have to be connected in parallel with the LED in the feedback loop and connected to the power supply through a dropping resistor. This arrangement would be a voltage divider that would bias the diode at cutoff and essentially act as a "current mirror".

Or are there opto-couplers that are a completely package device that provides a light output in linear proportion to the voltage in the feedback loop and shines the light directly to a photo transistor?

There are (were) resistive optocouplers available.
https://en.m.wikipedia.org/wiki/Resistive_opto-isolator

The cool thing about the old school resistive (CdS) photoresistor is the long linear response that, in a logarithmic amplifier is perfect to get many orders of dynamic range that no other sensors can achieve.

Also, they work on AC or DC. That means, when the rocking occurs, the "to" and "fro" motion are both captured.

Finally, the slow response time of the photoresistor was plent fast for the application because the output was a relatively slow pen plotter system. which was not able to record anything over about 20 to 50 HZ (estimate). .