syscomp resonant peak reading help.

[Mike Sulzer]

I actually use the Syscomp CGM-101 for my portable desktop testing. It works great. Just connect the Input A with the Output + to one side of the resistor and connect Input B with the pickup + to the other side. Set the probes on X10 and let her rip. I didn't like the driver coil because it was too finicky and inconsistent for me.

There are two setups described in the Syscomp document. One uses a series resistor. It is for measuring impedance. The other uses a drive coil; it is for measuring frequency response. Not really the same thing, but either tells you the resonant frequency. The first is usually easier for most people to do, as you said. Your use of a 470K resistor gives a more accurate impedance measurement for most pickups than the 56K used by Syscomp.

(But an even better way to measure impedance is to measure the current through the series resistor and the voltage across the pickup; then take Vp/I. The voltage across the resistor can be found by measuring the generator voltage and the voltage across the pickup and subtracting. Then the current follows from I = Vr/R. This method is best accomplished with a computer to make the computations easy.)

[StarryNight]

Glad to see things haven't changed much around here since i last visited Even with the bickering, I do appreciate the input (actually would be strange without it).

I had to pull this from the guitar just to test it again! Here is the bridge pickup again with the polarity flipped. It would seem the scope graphs correctly in one configuration. I don't believe I changed anything between checking the neck and bridge pickups during the first test so I'm not sure why I needed to flip the polarity but it seems to work.

[Mike Sulzer]

Glad to see things haven't changed much around here since i last visited Even with the bickering, I do appreciate the input (actually would be strange without it).

I had to pull this from the guitar just to test it again! Here is the bridge pickup again with the polarity flipped. It would seem the scope graphs correctly in one configuration. I don't believe I changed anything between checking the neck and bridge pickups during the first test so I'm not sure why I needed to flip the polarity but it seems to work.

Thanks for the update! (Possibly what changed between the neck and bride measurements was the polarity of the pickup. It is really the relative polarity between the drive coil and the pickup that counts.)

[John Kolbeck]

I had to pull this from the guitar just to test it again! Here is the bridge pickup again with the polarity flipped. It would seem the scope graphs correctly in one configuration. I don't believe I changed anything between checking the neck and bridge pickups during the first test so I'm not sure why I needed to flip the polarity but it seems to work.

As I was saying at the top of the thread, if you compare the phase plots of the two screen shots, it's the same lines 180 degrees apart, because pickup or driver coil leads were reversed for one of the two screen shots.

http://i.imgur.com/I6Edr2v.jpg

The bug is in the fact that the software draws a line between each bode plot point, no matter what the circumstance, presumably because it has no way of realizing that the phase crossed over between those two plot points. It's just dumb "connect the dots" logic. I suspect it would be non-trivial to fix, because what is obvious to us is not obvious to the computer program. It's easier for us to just be aware of what we're looking at.

[Joe Gwinn]

Phase wraps are not bugs

That jump is not a bug. It's the standard way to display a phase line whose phase range exceeds 360 degrees. The standard way to "unwrap" such a plot is to add or subtract 360 degrees as appropriate to yield a smooth plot line.

Google on "phase unwrapping" for a deluge of references.

[eschertron]

That jump is not a bug. It's the standard way to display a phase line whose phase range exceeds 360 degrees. The standard way to "unwrap" such a plot is to add or subtract 360 degrees as appropriate to yield a smooth plot line.

Google on "phase unwrapping" for a deluge of references.

I think what John suggested is that a more 'visually friendly' graphing package would OMIT the line segment that jumps from -180 to +180. The discontinuity itself is not the issue. That's what I would prefer to see.

And a really smart (the non-trivial part) graphing package would auto-roll the Y axis to include all the curve within its bounds.

[CheopisIV]

The spike on the lower graph is not really a spike, its measurement has just run off the bottom of the defined scale range that was set (starts again on top). For that particular pickup plot, right click on the phase chart and change the Phase Axis to (+90,-270) and rerun the plot. The entire plot will be shown as you would expect, withing the range defined.

Stupid question about resonant frequency test. - Page 2

[Mike Sulzer]

I think what John suggested is that a more 'visually friendly' graphing package would OMIT the line segment that jumps from -180 to +180. The discontinuity itself is not the issue. That's what I would prefer to see.

And a really smart (the non-trivial part) graphing package would auto-roll the Y axis to include all the curve within its bounds.

Yes, I think that is what John is suggesting. But perhaps the person who wrote the software thinks that when you see that jump you should realize that you have a reversed polarity and go back and fix it Then the jump goes away, and the phases are correct.

[Joe Gwinn]

I think what John suggested is that a more 'visually friendly' graphing package would OMIT the line segment that jumps from -180 to +180. The discontinuity itself is not the issue. That's what I would prefer to see.

And a really smart (the non-trivial part) graphing package would auto-roll the Y axis to include all the curve within its bounds.

Although in the present example, unwrapping is obvious and thus easy for an algorithm to perform, unconstrained phase wrapping is a difficult problem, so a general-purpose plotting package cannot succeed more than it fails. So they don't try.

All practical unwrapping algorithms depend on some added domain-specific information to make the unwrapping problem tractable.

In the present example, we know that between wraps, the difference between adjacent phase values is small, say not exceeding 20 degrees, so when one finds a difference exceeding say 40 degrees, one may safely assume that one has found a phase wrap, and proceed to fix it. The domain specific knowledge here is knowing that the difference between adjacent unwrapped phase values is always small.

[eschertron]

Yes, I think that is what John is suggesting. But perhaps the person who wrote the software thinks that when you see that jump you should realize that you have a reversed polarity and go back and fix it Then the jump goes away, and the phases are correct.

Although in the present example, unwrapping is obvious and thus easy for an algorithm to perform, unconstrained phase wrapping is a difficult problem, so a general-purpose plotting package cannot succeed more than it fails. So they don't try.

All practical unwrapping algorithms depend on some added domain-specific information to make the unwrapping problem tractable.

In the present example, we know that between wraps, the difference between adjacent phase values is small, say not exceeding 20 degrees, so when one finds a difference exceeding say 40 degrees, one may safely assume that one has found a phase wrap, and proceed to fix it. The domain specific knowledge here is knowing that the difference between adjacent unwrapped phase values is always small.

Stupid question about resonant frequency test. - Page 2

So instead of a bug we have an 'undocumented feature'? Or perhaps it is documented but not widely disseminated. The problem can be fixed with a mouse click. Good to know.

[salvarsan]

So instead of a bug we have an 'undocumented feature'?

"It's not a bug; it's a feature"?

I haven't heard that one before.