Caruso parametric curve

[Dub Studio]

Anyone else been playing with DIYAudio Wolverine as a cutting amp, it is a fairly conventional class AB amp running at fairly high bias (0.5A or so per side), and manages a couple hundred watts in the 4 pair configuration, if you remove the input filter it is flattish out to 100k with little excess phase and has markedly low distortion. I avoided the class G/H stuff for this because I am not confident the rail boost will reliably keep up at high frequency given that there is about a 20dB treble boost in play.

Mine are running on +-70V, but that may turn out to be a little overkill, particularly if I wind up re drawing it as a bridged design.

Doing a multi tone test if you have the gear for it (DScope III or AP whatever) might work well here, as it has a suitably low average power, still best done with the 75us stage turned off however, I suspect IMD might be as informative as frequency response.

Do watch the levels, particularly if you are a 50 year old rocker with nothing above 12k or so, there is plenty of space above what you can hear to smoke coils (Fortunately only done that with ebay voice coils I was using to test a "circuit breaker" board, at a couple of quid a time who cares?).

Sorry, my electronics skills are basic to say the least. Let us know how you get on with it though.

I wonder how a feedback system would handle those high frequencies..

[dmills]

You need the amp to be flat at least in terms of phase (And that tends to imply flat amplitude as well) out to well above where the feedback compensator board runs out of loop gain because excess phase shift thru the cutting amp will hurt the phase margin available to the compensator and thus reduce the amount of feedback you can apply.

This is why choice of power amp matters to folk running feedback heads.

Incidentally the presence of 26dB or so of feedback is also why an amp far bigger then you think you need makes sense, if the amp clips, the feedback loop will make sure the amp clips HARD, and that is not a good look, either for the head or the sound.

[Dub Studio]

I guess phase shift essentially means poistive feedback starts to kick in? Would it not make sense to apply a low pass filter to the feedback signal?

[dmills]

Not really because a LPF will itself introduce phase shift (actually well before it has much impact on amplitude).

You do the frequency and phase response shaping as part of the error amplifier but you usually want to avoid any extra phase shift thru the electronics inside the loop (It is much easier to stabilise a feedback system that is wideband within the loop). The cutter head itself is an integrator (more or less) because current is force, but feedback is velocity.

At low frequency the inductance of the cutter has little effect, so the current tracks voltage rather well, but at high frequency you effectively have another integrator (due to the winding inductance) and would have a 180 degree phase shift and hence oscillation were it not for the parallel RC network that partially tunes out the winding inductance and the error amp having a deliberately falling response at high frequency, such that the loop gain falls to less then one before the phase shift turns negative feedback into positive feedback.

Then there is the minor matter of the acoustic delays due to the finite speed of sound and the feedback coils being mechanically separated from the drive coils.

Real feedback cutters actually tend to run out of loop gain somewhat before 20kHz and use feedforward to make the last octave or so about right without significant feedback correction up there.

The whole thing comes under 'Control theory' and is a fun rabbit hole.

[Dub Studio]

That makes sense, although one thing that confuses me is the ideal placement of the feedback coils. I have read (I forget where, perhaps this forum somewhere) that the closer the feedback coils are to the stylus the better... can that be true? Or is it best they are near the drive coils (magnetic interference notwithstanding)?

[markrob]

Hi,

Yes, you would like the feedback coils to be as close as possible to the stylus, since that is the motion you are trying to control. That will add some more control problems that need attention since you will now have any mechanical effects of the stylus and shank in the groove. One other thing to note that if you try to place the feedback coil close to the drive coil, you will have to deal with the electrical coupling from the drive coil swamping out the actual motion of the coil.

Mark