Complicated Opamp circuit

[veryevil]

Hey, working with this circuit and was not sure exactly what the first opamp is doing.



It has a gain of -83.33. the cap in the feedback path acts as low pass filter but I'm not sure about the diode and the resistor / cap on the negative input.

I've also never seen a cap placed across the two inputs.

The input is from a sensor which has a dc offset and then very small signal on top of it.

Any help would be great.

[T3sl4co1l]

The diode keeps the output from rising more than 0.7V above -in.  If the input has a little bias, then the output will be constrained between about 0V and input + 0.7V.  Just some clamping I guess.  Doesn't seem to make any difference, seeing as the next stage has even more gain, so the diode clipping will still result in saturation of the following amp.

The cap between inputs is probably just to reduce HF noise.  With the cap from out to -in, it's probably stable.  But not usually something you want to rely on for filtering.  (If it still looks weird, recall an op-amp is more like an integrator, so there is a nonzero input signal at AC.  Obviously, in slew rate limiting it jumps up to whole volts!  The effect of the cap depends on the amp's GBW, so it's not the best thing to rely on.)

Tim

[jpb]

If the diode was the other way around it would be a log amplifier complicated by other components. Perhaps it is acting as an exponential amplifier if it is designed for very small signals - though I guess it would be very dependent on temperature.

Is it some sort of peak detect circuit? When the input voltage is above that on the 4.7uF cap then the output goes high the diode conducts and the cap charges to a higher voltage, when the voltage is below the diode doesn't conduct and the 10M is very high impedance so is essentially open circuit.

[uwezi]

The diode keeps the output from rising more than 0.7V above -in.  If the input has a little bias, then the output will be constrained between about 0V and input + 0.7V.  Just some clamping I guess.  Doesn't seem to make any difference, seeing as the next stage has even more gain, so the diode clipping will still result in saturation of the following amp.

But the second amp is AC coupled to the first stage. A DC clamping of the first stage will still allow for it to amplify AC signals which are then further amplified by the second stage.

[uwezi]

The input is from a sensor which has a dc offset and then very small signal on top of it.

It might help if you told us what kind of sensor we are talking about...

[veryevil]

Interesting ideas.

The full schematic is in this pdf http://www.murata.com/products/catalog/pdf/s21e.pdf on page 11

[T3sl4co1l]

Yeah, they just need a little bit of gain at a low frequency, but not to DC.  So it's AC coupled, with a bandwidth of, apparently, 0.5 to 5Hz or so.  Which corresponds roughly to the expected frequency from one of those multi-faceted-lens people detectors: as long as a warm body is moving across the field of view, the image alternately comes into and out of view of each facet, producing a fairly specific alternating signal.

Probably, the diode clamps the first stage's output to about +/-1V around quiescent (very roughly, depending on just what the bias ends up at), so the second stage isn't driven beyond the rails (otherwise, the coupling cap could push the input above or below the supply rails, which an LM358 doesn't appreciate much).  The second stage doesn't need this, because its output is centered around Vcc/2 and drives a window comparator (which has an input range which includes the 358's output range).

Tim

[uwezi]

The full schematic is in this pdf http://www.murata.com/products/catalog/pdf/s21e.pdf on page 11

...and I had almost started to complain to you why you were drawing such an ugly schematic...

[Dave Turner]

Any idea why 1kV capacitors have been specified.

[uwezi]

Any idea why 1kV capacitors have been specified.

where did you get the 1 kV from?

I interpret the capacitors as Y5P ceramics (B) with ±10% tolerance (K)

[JoeO]

According to the App Note, it is a portion of a human detector.

[Dave Turner]

uwezi

A number of the capacitors are described as B222K. Since I wasn't certain of the description I googled it. The results were description and data sheets of high voltage (1kV) capacitors. Hence my query.

[uwezi]

uwezi

A number of the capacitors are described as B222K. Since I wasn't certain of the description I googled it. The results were description and data sheets of high voltage (1kV) capacitors. Hence my query.

That's what I also did, googled I mean. Google immediately suggested some kV in addition to my search term, which I decided to ignore. I then ended at a Murata datasheet over their ceramic capacitors at http://www.rockby.com.au/DSheets/10059.pdf

On the page 3 it describes Murata's part numbering scheme for ceramic capacitors, where the leading B indicates the tempco  as ±10% within -25°C to +85°C, a bit further down the book this is stated as the tempco without DC bias. This then suddenly reminded me of another manufacturer which uses the same classification (to everyone: is this a standard?), but even defines the ceramic material as Y5P: https://www.elektronik.ropla.eu/pdf/katalog_dongil.pdf

Murata indicates the voltage rating in clear numbers after the tolerance letter.

[Dave Turner]

uwezi - Thanks. I didn't look deep enough, but I get it now.

[qno]

I think the diode is for fast charging the 4.7 uF capacitor during switching on.

[T3sl4co1l]

Good point!

[uwezi]

I think the diode is for fast charging the 4.7 uF capacitor during switching on.

Not only during "switching on". The diode allows the (-)-input to quickly follow the (+)-input during positive transients from the sensor, while the system reacts much slower to negative transients.

From a quick Ltspice simulation I get a rise time of 0.3 s and a fall time of 3 s with the values from the original posting as a reaction to a 1 V step at the input.

Actually this leads to a sharp negative spike whenever there is a small negative transient on the input. The AC coupling of the second stage will preferably pick up these spikes. The attached screenshot shows the reaction to a square wave with rise- and fall times of 1 s - quite slow signals as expected from pyroelectric sensors...