Hi all,
Has anyone here measured the drive to feedback coil transfer function of a reasonably modern cutting head, and if so, would you mind sharing?
Either Bode or Nyquist plots would be helpful (or even, in an atypical use, S Parameters (S21 is the one that matters)!).
I need both amplitude and phase response, against frequency, so if you have a Nyquist plot it would be helpful if there were frequency markers.
'Basic Disk mastering' has one or two amplitude plots, but that is of limited utility as the phase data is not there.
I am playing with an idea and want to see if the loop dynamics supports it.
Thanks to anyone who can help, I am sure the guys designing the old lathes had this stuff but it does not appear to be widely published.
Regards, Dan.
Bode plot of a feedback head wanted.
Moderators: piaptk, tragwag, Steve E., Aussie0zborn
Re: Bode plot of a feedback head wanted.
Hi,
Here is a page from an Ortofon brochure. Hope that helps.
The loop compensation is easy if you have a mechanical system that is as free from secondary resonances as found n the Ortofon. The real design challenge is to make that happen.
Mark
Here is a page from an Ortofon brochure. Hope that helps.
The loop compensation is easy if you have a mechanical system that is as free from secondary resonances as found n the Ortofon. The real design challenge is to make that happen.
Mark
You do not have the required permissions to view the files attached to this post.
Re: Bode plot of a feedback head wanted.
Thanks for that.
I can see that even lacking secondary resonances getting the feedback coil as close as possible to the drive coil will matter at high frequency where the phase shift becomes determined by the acoustic distance between the coils (Also by standing waves in the assembly).
So, feedback signal is clearly proportional to velocity (Faradays law of induction), and within the loop the drive current gives force hence acceleration.
The amplifiers are voltage output but current is the integral of (V-IR)/L dt and velocity is the integral of acceleration, ok, I see where the large LF phase shift comes from.
The ~180 degrees of phase shift thru resonance (second order) is also understandable so there does not appear to be any surprises here.
Looks like a gyrator or state variable notch inside the forward path of the loop to add a zero to cancel the resonant pole plus a fairly standard type II compensator would get it done.
Does this sound about right? I am just trying to make sure I actually understand how these machines work.
Annoyingly I have the schematics of a system including HALF of the relevant card, but not the bit that actually has the error amp....
Regards, Dan.
I can see that even lacking secondary resonances getting the feedback coil as close as possible to the drive coil will matter at high frequency where the phase shift becomes determined by the acoustic distance between the coils (Also by standing waves in the assembly).
So, feedback signal is clearly proportional to velocity (Faradays law of induction), and within the loop the drive current gives force hence acceleration.
The amplifiers are voltage output but current is the integral of (V-IR)/L dt and velocity is the integral of acceleration, ok, I see where the large LF phase shift comes from.
The ~180 degrees of phase shift thru resonance (second order) is also understandable so there does not appear to be any surprises here.
Looks like a gyrator or state variable notch inside the forward path of the loop to add a zero to cancel the resonant pole plus a fairly standard type II compensator would get it done.
Does this sound about right? I am just trying to make sure I actually understand how these machines work.
Annoyingly I have the schematics of a system including HALF of the relevant card, but not the bit that actually has the error amp....
Regards, Dan.
Re: Bode plot of a feedback head wanted.
Hi,
You don't have to cancel the main system resonance. Since the feedback coil is in effect a derivative, you get a +90 degree phase shift to start out. On the other side of resonance, you shift -180 degrees referenced to that, but only have -90 total to deal with. So a simple summing amp closes the loop with plenty of phase margin. You do have to deal with the excess phase shift due to prop delay. Also, note that the plot was done using a constant current drive method rather than they typical constant voltage. This kills any response roll off due to the L/R time constant of the voice coil.
The hard part will be to achieve an open loop response this good. The Westrex heads had a secondary resonance in the 10-12Khz range and had to be compensated. I believe they used some lead-lag compensation.
Mark
You don't have to cancel the main system resonance. Since the feedback coil is in effect a derivative, you get a +90 degree phase shift to start out. On the other side of resonance, you shift -180 degrees referenced to that, but only have -90 total to deal with. So a simple summing amp closes the loop with plenty of phase margin. You do have to deal with the excess phase shift due to prop delay. Also, note that the plot was done using a constant current drive method rather than they typical constant voltage. This kills any response roll off due to the L/R time constant of the voice coil.
The hard part will be to achieve an open loop response this good. The Westrex heads had a secondary resonance in the 10-12Khz range and had to be compensated. I believe they used some lead-lag compensation.
Mark
Re: Bode plot of a feedback head wanted.
Thanks Mark, makes sense.
That plot is indeed very, very much better behaved then I was expecting, I was figuring on a mess of series/parallel resonances starting at maybe 5kHz or so, and becoming progressively more annoying to compensate.
Is the secondary Westrex resonance the reasoning behind that parallel RC network in the cutter head wiring do you know?
Regards, Dan.
That plot is indeed very, very much better behaved then I was expecting, I was figuring on a mess of series/parallel resonances starting at maybe 5kHz or so, and becoming progressively more annoying to compensate.
Is the secondary Westrex resonance the reasoning behind that parallel RC network in the cutter head wiring do you know?
Regards, Dan.
Re: Bode plot of a feedback head wanted.
Hi,
I believe that RC network was to used add a bit of phase lead to compensate for the L/R roll off.
As someone who is working on an open loop head design, I can attest to the difficulty of getting an open loop response that looks as clean as the Ortofon.
You might want to look at getting a hold of the 2 volume AES Disk Recording Anthology. That publication brings together most of the seminal papers on cutter heads and playback pickups.
https://customer258769455.portal.membersuite.com/onlinestorefront/BrowseMerchandise.aspx?contextID=dfb59c4a-0066-c63d-c975-0b3b93bf27d5
Well worth the price.
Mark
I believe that RC network was to used add a bit of phase lead to compensate for the L/R roll off.
As someone who is working on an open loop head design, I can attest to the difficulty of getting an open loop response that looks as clean as the Ortofon.
You might want to look at getting a hold of the 2 volume AES Disk Recording Anthology. That publication brings together most of the seminal papers on cutter heads and playback pickups.
https://customer258769455.portal.membersuite.com/onlinestorefront/BrowseMerchandise.aspx?contextID=dfb59c4a-0066-c63d-c975-0b3b93bf27d5
Well worth the price.
Mark
Re: Bode plot of a feedback head wanted.
Thanks Mark, $60 well spent!
Those are good reading, and as we all know three weeks in the lab can save all of 30 minutes in the library!
This is not a cheap hobby, but is turning out to have a rather international flavour, PCBs on order from China, transverse slide mechanics from Germany, main gearbox from Korea (Ex robotics parts), motors are Swiss/American, semiconductors from the Philippines, and god only knows where the tooling plate is manufactured.
I have not even got to the head yet, still contemplating the possibilities of modern PZT or turfenol D and strain gauge feedback, but also looking at a broken 1 inch compression driver and scratching my head, we will see.
Everyone should try proper precision mechanical stuff sometime, makes the software and electronics seem very, very easy.
Those are good reading, and as we all know three weeks in the lab can save all of 30 minutes in the library!
This is not a cheap hobby, but is turning out to have a rather international flavour, PCBs on order from China, transverse slide mechanics from Germany, main gearbox from Korea (Ex robotics parts), motors are Swiss/American, semiconductors from the Philippines, and god only knows where the tooling plate is manufactured.
I have not even got to the head yet, still contemplating the possibilities of modern PZT or turfenol D and strain gauge feedback, but also looking at a broken 1 inch compression driver and scratching my head, we will see.
Everyone should try proper precision mechanical stuff sometime, makes the software and electronics seem very, very easy.
Re: Bode plot of a feedback head wanted.
Hi,
Glad you find that useful. I'll be interested to see if you get any results using a PZT element for feedback. One big issue for any feedback sensor is the susceptibility to pickup of the drive coil EM field. PZT being very high impedance might require quite a bit of shielding to keep this from being a problem.
Take a look at the Fairchild 642 head design on page 253 of Vol 1. They used a RF eddy current position sensor much like the sensors Kaman produces. By getting away from the baseband, they keep interference pickup down to a minimum. There is also a wealth of info on general head design in this paper.
Mark
Glad you find that useful. I'll be interested to see if you get any results using a PZT element for feedback. One big issue for any feedback sensor is the susceptibility to pickup of the drive coil EM field. PZT being very high impedance might require quite a bit of shielding to keep this from being a problem.
Take a look at the Fairchild 642 head design on page 253 of Vol 1. They used a RF eddy current position sensor much like the sensors Kaman produces. By getting away from the baseband, they keep interference pickup down to a minimum. There is also a wealth of info on general head design in this paper.
Mark
Re: Bode plot of a feedback head wanted.
I was actually planning to run the strain gauges at 1MHz or so and do synchronous demodulation to remove the baseband bleed, there is much to be said for getting away from DC when doing sensing of mostly DC things.
My notion was to use a flexture similar to a thorlabs PK2FSF1 but set up for two stacks in push/pull (to cancel the hysteresis and deal with the fact that PZT does not take tension well). I have some significant reading and thinking to do however. Going down the magnetic route (at least for the drive coils) may very well turn out to be easier overall (PZT being rather capacitive after all).
The electronics are not a concern, I know what I am doing there, it is the mechanical bits that are causing headaches.
Regards, Dan.
My notion was to use a flexture similar to a thorlabs PK2FSF1 but set up for two stacks in push/pull (to cancel the hysteresis and deal with the fact that PZT does not take tension well). I have some significant reading and thinking to do however. Going down the magnetic route (at least for the drive coils) may very well turn out to be easier overall (PZT being rather capacitive after all).
The electronics are not a concern, I know what I am doing there, it is the mechanical bits that are causing headaches.
Regards, Dan.