Variable Q Repost with Images - Existing Circuit

This looks like a job for PSpice...

http://www.engr.uky.edu/~cathey/pspice061301.html

What you're describing is the output signal from a linear variable differential transformer (LVDT) used to provide a feedback signal to a servosystem that moves something based upon the position of the mechanical linkage attached to the iron core.

The key variable is the voltage across Rint (int=internal or integrator?) and the circuitry that surrounds it. It's hard to tell if the pair of capacitors in the bridge provide some form of filtering or integration to that voltage signal, but you might be able to tell how critical their value is by looking at them and seeing if they are commodity items or have some form of precision code on them. Fortunately they have a common value so it won't be too hard to replace them even if you have to buy a bunch and test them until you get a close match.

Thanks! I should have realized that it was an early form of LVDT. I've never explored that kind of circuitry or had a chance to play with one. Now that I've searched a little bit, I have verification as to what the inherent issues are. That's a relief!

Looks like I have several issues from component drift to phase shift matching to linearity. Now it looks like I may have to consider an upgrade as a possibility or as a long range plan. Now I wonder if the variable speed drive and processed LVDT signal are supposed to be matched? Me thinks I need to explore the whole process a little closer one day. For now, I've got to pull a rabbit out of a hat and make some non-working modules work again. I'm expecting the capacitors to be the one component that is most likely to drift out of spec.

Since the caps are shown as polarized they may not have just "drifted" out of spec. they may have dried up. Are they aluminium electrolytics?

I do not think it is as clever or linear as an LVDT. Moving the coil changes the inductance and off-tunes a resonant circuit. The graph below is a generalised resonance curve for an L-C circuit, [from "Foundations of Wireless" by M.G. Scroggie: Iliffe]. I think you have an L-C series circuit formed by the inductor and the capacitors across the bridge diodes.

The important things are :-

1. For a reasonable Q, the amplitude/frequency response is symmetrical about the resonant frequency, so the graph is half the response.
2. Amplitude response is highly dependent on Q of coil, but a resistor across the circuit (the bridge load here) reduces the circuit Q to a low value and makes the effect of variations in coil Q insignificant.
3. Low effective Q broadens the amplitude/frequency response & makes it more linear over a limited range.
4. Circuit is operated on one slope of the resonance curve.
5. I would expect the initial coil position is set so that response for movement either direction is reasonably linear.

You have not clarified if the circuit is a position measure or a "one blip per revolution" speed detector (as a speed detector the amplitude does not have to be high precision) but it could do either.

67model

The coil with the variable inductance - movable core, is supposed to affect the circuit in a way so that it drives a variable speed drive from some fixed speed in one direction to a relatively slow speed in the other direction as the inductor core goes from one extreme to the other.

The more I study this circuit and compare it to LVDT circuits, the more I dislike what I have to fix. It is in serious need of an upgrade for temperature stability, etc. There is no doubt that this circuit is a dog. But, its' my dog until I can replace it.

For now, I'm looking at the prospect of having to do a full blown time domain analysis. Uggggh. Been a long time!

I am not sure if you read my post #5 before your post #6.

Have you really got stability/repeatability problems or is it just not working? Inductors can actually be very stable.

Having thought a bit about the circuit, I note that 0.22 microfarads is ~ 3000 ohms reactance at 300 Hz. Two in series will be 6000 ohms - so the 2000? ohm load R on the bridge will reduce the Q at 300 Hz to 2000/6000 = 1/3 [Q ~ R/Xc]. So I think the circuit is simply applying an AC voltage to the inductor and measuring the current through it - proportional to its inductance. I reckon the inductor is several Henries.

I would expect the circuit to work with 50% duty cycle & much more equal half cycles - are you sure you have the right frequency? Some oscillators have different resistors setting each half cycle. Looking at the trace, the input wave would be much more symmetric if the negative half cycle were the same width as positive (about 600 Hz). Equal half cycles would reduce the fluctuation of the bridge output.

I would check the capacitors across the bridge diodes are both the marked value.

I think you are exaggerating the difficulty of analysis. In any case, it is no good if you do not have the component values.

My approach would be to disconnect the inductor from everything else electrical and measure its inductance by applying a sine wave ~ 300 Hz from a signal generator with low output impedance (50 ohm) [or better via a Hi-Fi amp, which would have <<1 ohm output impedance] through an 0.1 microfarad Mylar capacitor. Measure voltage across the inductor with a high Z (1 meg) voltmeter or oscilloscope and find the (series) resonant peak of the inductor and 0.1 microfarad. Resonant peak voltage compared to generator volts will give a measure of Q (low if you have shorted turns fault) and resonant frequency will give inductance [X_L = X_c].

Or put a resistor (low compared to the coil DC resistance) in series with the coil, to monitor current, and measure the voltage across the coil & shunt resistor at several frequencies around 300 - 1000 Hz to get a volts/current impedance.

Measure with core at both ends of mechanical travel and steps between.

67model

Thanks for the reply. Most of my data, etc. comes from the actual circuit. New thread coming up about s-domain questions shortly.

Unfortunately, I'm about 4 hours drive from the problem so it limits my ability to test things on site.

I'm trying to build a "spare parts" module based on a part that went out of spec (quit working?) I've had a brief look at a working module, but I haven't worked with this kind of circuit for a long time.

I'm trying to remember how to do an s-domain analysis (Fourier Transform) but I'm really rusty. The last time I went to that much trouble was in 1980.

I'm going start another thread about s-domain analysis. I hope you will follow me through this process, because I think I will be able to determine more about how this works from that effort.

Please clarify the following. 1. Is this a circuit diagram from the manufacturer or a self drawn one? 2. The scope traces after the Step up transformer called V(sec) core in and V(sec) core out, is that the trace measured across the 100nF Cap? or is this a trace from the chassis ref point (your 'Earth' symbol)? 3. Is the Green centre trace on the scope screen your actual 0Volt line on the scope. 4. On the red trace is the scope set on DC and is that a memory scope where you could recall the V(sec).

I the circuit the 100nF and the 2x 220nF caps can not be Polarized caps as the AC would have destroyed them on the test bench. Please check all the markings on those Capacitors to ensure yourself that there are indeed neg and pos indicator marking on.

The input to the switching transistors is probably a mark space ratio generator to determine the speed and the core on the coil is just a tuning coil that should have been left alone.

Please make sure about the Capacitors and repost.

I redrew the manufacturers diagram in MSPaint. The ~0 to 2 cm square wave in the input to the signal transformer. The transformer output is V(sec) and is measured with 0 V (ground) as the reference. The 100nF cap is always in place. With no load, the output of the signal transformer is a nice clean (fast slew rate) square wave.

Actually, I'm not even sure the 100nF does anything on the output side except to help slightly with phase control. As I researched LVDT circuits I realized that this one was a very early version (so I would call it an LVT position sensing). From my reading, this circuit is subject to error due to phase shift, temperature variation, and other maladies. The only electrolytic capacitor was the 220uF. All others are plastic film (except for one defective module that appeared to use silver Mica caps).

If it were up to me, I'd upgrade before wasting time trying to make this work. I re-posted with some comments in an attempt to model this and use Fourier Transforms to analyze it. The drive signal is a simple op-amp circuit with a pot to adjust the duty cycle a few percent either way around the 50% mark.

This circuit is supposed to generate a slightly negative DC voltage up to a few volts to feed the analog input of a Variable Speed Drive. The red trace is the bridge output with respect to ground while the blue(or green) trace is the input to the bridge. The scope has a common ground on both input channels and DC coupling was used. The arrow (or triangle) on the left is the zero set point for the channel.