Dividing MM Carts into electrical parts with individual Transfer Functions

Member
Joined 2005
Paid Member
In a joined effort, Bill, Dagfinn and I went into much detail by developing replacement diagrams for MM cartridges, divided into two parts, first the output section with coil, that we called the Generator, and second for the complete Cantilever assembly.
An achievement never realized before to the best of our knowledge.
Having developed the transfer functions for both both parts, we were able to prove that these two parts don't "see" each other, in other words, the external load on the Generator does not influence the Cantilever's movement in any way.
Another benefit by having the replacement diagrams, was that it was now possible to find the optimal load in LTSpice for a specific cartridge as regards to it's FR.
Because being too long to put it in a posting, we have added the results in the attachment as a PDF.

Happy reading,

P.S. A few typos haven been corrected and further explanation has been given to indentation.
 

Attachments

Last edited:
Hans, this is very good work. Two comments if I may.

1. I didn't understand section 4(a). It's not clear what the 'concave' and 'convex' parts of the sine wave are. Is this section talking about vertical modulation? I think it needs expanding.

2. The claim that 'one can also put a signal on a generator, but that will never lead to a moving needle' seems to defy the laws of electromagnetism. A generator can be driven as a motor, and vice versa. Is this really what was meant?

In general I found it all a little terse and in need of expansion.

Great stuff though.

EJP
 
A MM generator is at best 0.5% efficient, so no defying the laws of physics, you just cannot put enough power in to make it damp let alone move the needle.

The Terseness is my fault as I was trying to pad out the technical bones with some flesh and ran out of time. All feedback is useful in helping point to where a few more paragraphs will help the reader so we can improve it.
 
EJP, thx for your positive reaction.

1) Obviously 4a) was to brief because there were more people asking for more detail. That's why this section was extended with extra text plus an image.
When reading the PDF again in the first posting, you will notice this.
In case it is still not completely clear, please let me know.

2) As mentioned in the paper, measuring the Generator with a very sensitive VNA, it was impossible to see any difference with or without cantilever connected.
So laws of reciprocity made me wonder, how about the other way from cantilever to Generator, that's why this paper was seeing the light.
Fully respecting the laws of magnetism, but having a magnitude of a coupling so marginal, makes that for practical reasons a Cart's model can be seen in simulations as composed of two completely independent parts not "seeing" each other.
Also the often expressed opinion that having DC flowing in the generator will offset the cantilever or dampen its movement can therefore be completely rejected.

Hans
 
Last edited:
Super job!!!
Your findings explain why so many nice ideas, developed with a simplified cartridge model in mind, could never really realize their initially claimed advantages.
In my view, the most important finding is your model explaining the sag in FR at around 5 kHz visible in FR for the vast majority of MM and MC cartridges.
 
4(a) is much better now, thanks. I think you could explain why it peaks at ~5KHz, specifically what happens at lower frequencies; and why cutting the corner results in lower amplitude rather than 2HD, which I assume is from the velocity-sensitivity.

I've now proof-read the whole document and annotated all typos in the -1 version. I'll send this to you via PM as it is pretty tedious (=thorough).

EJP
 
An observation.

The needle in the groove is assumed to be linear parallel? Most record players have an arm where the needle swings through an arc which changes the needle profile (the needle rotates in relation to the grooves). Now it depends on how the record was recorded but either linear arms or normal arms are therefore incorrectly matching the transfer function at this point.
 
4(a) is much better now, thanks. I think you could explain why it peaks at ~5KHz, specifically what happens at lower frequencies; and why cutting the corner results in lower amplitude rather than 2HD, which I assume is from the velocity-sensitivity.
In fact the indentation itself does not peak, but the resonance at a much higher frequency starts to lifting the FR, see figure 9. In some cases it can even completely mask the indentation.
So, for the manufacturer it’s all a fine lnterplay and tuning of all individual parameters.

Also be aware it’s not like hard clipping or cutting corners but a much more gradual process of diminishing the amplitude, but yes it may be that this will lead to some uneven harmonics, but most likely not noticeable.

Hans
 
Last edited:
Nice work Hans, though it makes me think I could have overloaded the Audio Technica M95ML. Although the input is about 47K the bipolar inputs appear having lowered the overall input impedance to around 10K in testing. This appears as potentially problematic given your research.
 
A MM generator is at best 0.5% efficient
Bill, the comments below are not directed at you specifically. Anyone who would like to offer some clarification would be great.

1. Why does it seem like 0.5% is being turned into what amounts to a very solid 0.0% for some of the explanations in this thread? If we were talking about some kinds of distortion then 0.5% might not all that small of a number?

2. Also, I'm not sure about Hans' comment as to something being 'below our auditory threshold.' Such a threshold implies 50% of the population should be able to hear below that threshold, doesn't it?
 
This is nice work. The flatness of the measured frequency response with the test disc is impressive.

I think that some record companies do not aim for flat response but rather something that sounds good on average cartridges. This produces tipped-up treble when played on a good cartridge. The mastering process seems to be filled with kludges.
Ed
 
Nice work Hans, though it makes me think I could have overloaded the Audio Technica M95ML. Although the input is about 47K the bipolar inputs appear having lowered the overall input impedance to around 10K in testing. This appears as potentially problematic given your research.
Not sure to understand what you mean.
How could you have overloaded your Cart, in what way ?
And what amp are you using that has a 10K dynamic input despite a 47K input load.
Please provide a bit more details for me to understand your potential problem.

Hans
 
Bill, the comments below are not directed at you specifically. Anyone who would like to offer some clarification would be great.

1. Why does it seem like 0.5% is being turned into what amounts to a very solid 0.0% for some of the explanations in this thread? If we were talking about some kinds of distortion then 0.5% might not all that small of a number?

2. Also, I'm not sure about Hans' comment as to something being 'below our auditory threshold.' Such a threshold implies 50% of the population should be able to hear below that threshold, doesn't it?
Mark,

Bill was talking of 0.5% efficiency at the most, anyhow way too little to be confirmed in the most sensitive measurements.
This figure, or however low it may be, should not in any way be confused with non-linearity or distortion.

Assuming that you are referring to tracking errors, “below auditory threshold” means to me that seemingly nobody can hear these errors and certainly not just 50% of the population when evaluating very expensive arms.

Hans
 
By "overload" I just meant having lowered the load resistance from 47K to 10K, being far below the recommended 47K. In other words, from your study the impact to frequency response is variant enough to be concluded as not recommended under normal circumstances, this leading to the conclusion the cartridge is therefore being "overloaded".

The network is a simple differential amp using the LM394 without feedback. I have the MM inputs connected between the bases with the outputs feeding a current amplifier that essentially shorts the collectors together. The inputs are across 2 x Vbe as generating collector current. I am guessing the input resistance would be something like the Hfe x some form of effective emitter resistance. It seems this could get to 10K as dependant upon collector current. In a feedback system the input resistance can rise dramatically dependant upon the feedback ratio. What I am doing is a little radical being open loop.
 
  • Like
Reactions: Calvin