Well, I guess to the extent the cartridge is linear, it can be thought of as some coupled spring-mass systems, with some excitation from the record being played. There is some transfer function for that system that can be modeled as an electrical analog.
In the case of underdamped ringing, there is stored mechanical energy in parts of the cartridge vibrating back and forth like one or more coupled tuning forks. In that case, it's not clear to me how you can instantaneously compensate for output due to the next step excitation (say, from a square wave input) without knowing when that excitation is going to occur in advance. That is, without just filtering out the resonances and undoing the HF response extension due to the resonances.
Then again, all that means is, it isn't clear to me. Maybe I just need to think about some more.
Anyway, with an FIR filter, at least you effectively know the future output in advance, so in principle maybe you could compensate for it exactly as it occurs.
In the case of underdamped ringing, there is stored mechanical energy in parts of the cartridge vibrating back and forth like one or more coupled tuning forks. In that case, it's not clear to me how you can instantaneously compensate for output due to the next step excitation (say, from a square wave input) without knowing when that excitation is going to occur in advance. That is, without just filtering out the resonances and undoing the HF response extension due to the resonances.
Then again, all that means is, it isn't clear to me. Maybe I just need to think about some more.
Anyway, with an FIR filter, at least you effectively know the future output in advance, so in principle maybe you could compensate for it exactly as it occurs.
Last edited:
Only skimmed that, but I suspect this is the usual case of an EE looks at electrical problems, an ME looks at mechanical problems and a Physicist wonders where the spherical cow of uniform density has gone.
Mark: See where you are coming from and it is possible that the simple electrical model for mechanical resonance is too simple and in removing time domain issues with impulses you mess something else up. Certainly worth investigating.
Mark: See where you are coming from and it is possible that the simple electrical model for mechanical resonance is too simple and in removing time domain issues with impulses you mess something else up. Certainly worth investigating.
So that's someone else who's made a mistake !
Post # 7 http://www.diyaudio.com/forums/analogue-source/303396-phono-termination-calculations-calculator.html
Vo/Vin of a simple 4 element circuit? I'm not up for the tedium of doing it on paper today. It should be trivial to verify the plots in a simulator or even one of the free biquad calculators.
I just recreated Fig. 10 in LTSPICE with the values stated, so is this just a notational error in the formula or what? I admit I'm not that enthusiastic about the predictions of these MM models where the lump electrical elements contribute well down into the audio frequencies, for one thing 100's of turns of #40 + wire is distributed at some point and the mechanical stuff is just a mess when you try to create real reciprocal electro-mechanical models.
Another point, in Eric's paper he states at the start he assumes no coupling of electrical to mechanical at all, for MM/MI that is basically my opinion but I realize others differ on that. BTW Eric is an acquaintance via his involvement in the LHC. I agree on your assessment of his approach but I would hesitate without careful checking to critique his results taking into account his stated assumptions. A trustworthy test LP seems to be a necessity in any case.
Last edited:
If you load it with a short circuit, say a virtual ground current to voltage converter, does that change, or dampen, the mechanical resonance at all?
In other words, any idea how much loading affects mechanical performance? To put it another way, how efficient is it as an electrical generator?
In other words, any idea how much loading affects mechanical performance? To put it another way, how efficient is it as an electrical generator?
Last edited:
This has been argued at length and my opinion remains that the coupling coefficient in the case of MM is vanishingly small. Easy experiment, measure the complex impedance of a loaded cartridge w/wo stylus and with stylus lowered onto an LP. You should be able to "see" the cantilever resonance/vinyl compliance as a change in motor impedance.
It might be interesting to measure w/wo the other channel shorted.
Last edited:
so is this just a notational error in the formula or what?
I don't know how Mr. Margan derived that formula for damping. The transfer function is correct but I don't understand why he used damping instead of Q. I used to work in control systems and we used damping because the mechanical engineers understood damping. The conversion from Q to damping is trivial.
Anyway, Zero D at entry 26 provided a link to another thread and at entry 9 on that thread I provided an attachment showing how to derive the formula for the cartridge Q assuming a simple low pass filter model. The derivation is simple and the formula for the Q is simple, but nobody gets it right.
I don't know how Mr. Margan derived that formula for damping. The transfer function is correct but I don't understand why he used damping instead of Q. I used to work in control systems and we used damping because the mechanical engineers understood damping. The conversion from Q to damping is trivial.
Anyway, Zero D at entry 26 provided a link to another thread and at entry 9 on that thread I provided an attachment showing how to derive the formula for the cartridge Q assuming a simple low pass filter model. The derivation is simple and the formula for the Q is simple, but nobody gets it right.
Scott says "a trustworthy test LP seems to be a necessity" and I agree. These days, where can one find such? I know there are a few readily available (HFNRR, 'Ultimate' from Analog Productions, and Cardas to name the three I know of). Not all of these get high reviews -- obvious issues like warping or eccentricity. I am not sure how to know if their sweeps have a flat response. Aside from taking your chances on an old CBS (or perhaps B&K) disk from eBay, what is one to do? I'm beginning to think that it might be better to use a small piezo accelerometer as a miniature shaker table. Perhaps with some clever jigs one can get it to shake at a 45 degree angle.
Thoughts? Options?
Thoughts? Options?
I agree Brian. Oh and Scott I have just realised what you were saying and I was missing. You are right I have no issues with the conclusions Eric came to from the starting assumption, just that there is another (and possibly better) way to approach the problem that is worth analysing.
I did start another thread on how to calibrate between test records which I might bring back to life once I've worked out how to handle the logistics of it. The approach of adding a scratch which is a wideband impulse (as done by B&K in the 70s) may yet be the best approach.
I did start another thread on how to calibrate between test records which I might bring back to life once I've worked out how to handle the logistics of it. The approach of adding a scratch which is a wideband impulse (as done by B&K in the 70s) may yet be the best approach.
Thought of those and tried a RS buzzer disk with little success just to see what happened. Any of these ideas would be difficult to calibrate to a reference frequency response.
This is the best bet I could find anyone up for it?
https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=10311