The Korf Blog

The inside story: our research, development and opinions

25 February 2019
Performance of Tonearm Headshells, Part I
In the course of our experiments, it became clear that the headshell performance varies widely, even for superficially similar designs. Unfortunately, using our standard methods, it is quite difficult to separate it from the influence of other tonearm parts.

This is why, in our new series of articles, we would try to isolate and measure only the performance of removeable tonearm headshells.

This study was made possible by a very generous donation. Mr Jam Somasundram, formerly with Pass Labs and now with Cinemag/Reichenbach Engineering, has given us his personal collection of headshells to measure.

Usually, an actual signal being picked up from a rotating LP is the best way to excite the resonances in the playback system. After all, it duplicates the real use scenario almost perfectly. This is the approach we used in measuring the tonearms and armtubes. With the latter, it was easy to extract their performance from the resonant picture of the whole tonearm. We know what it should look like, and the practical picture fits the theory really well.
Headshells, though, are a lot more complex than a simple pipe or wand. While we have some hunches (i.e. almost all of them ring at about 9-13kHz), it is impossible to truly separate the headshell's performance from that of a tonearm while the headshell remains attached.

To isolate the headshell, we have improvised the rig out of an old drill press chassis. The headshell is held by a standard SME/JIS bayonet connector which, in its turn, is gripped tight by a vise. But, obviously, this makes signal pickup from a rotating disc impossible.

In an ideal world, we would have access to a broad-spectrum vibration platform like the one used to calibrate accelerometers. It could easily be used to excite the cartridge stylus, and we will get a picture of vibrations that would be identical to using a real LP. Sadly, we don't have one, and even renting something like this is prohibitively expensive.

However, there is a very simple and cost-effective way of doing broadband excitation. It's called IET — Impulse Excitation Technique.

Instead of using a broadband shaker, we'll be striking the device under test with a very small calibrated hammer. The resulting impulse response can be then analyzed to yield resonant modes and damping estimation.

Dimensions of the calibrated hammer. The brass sphere weighs 5g.

In the engineering world, IET is mostly used to establish material properties, such as Young's modulus, shear modulus, Poisson's ratio and internal friction. While it would've been fun to be able to calculate those, it requires a sample of a known simple geometry and size. With the headshells varying widely in their basic shape, this is unfortunately impossible with the tools we have.

To create a calibrated hammer, we have used a machinist's magnetic dial holder to hang a 5g brass sphere 100mm away from headshell surface. The sphere is hanging on a thin Kevlar thread that, we hope, would not interfere with the resonances being recorded at headshell.
While for many IET measurements a microphone is enough, we have decided to use the Endevco Picomin 22 accelerometer as we did before. It gives a better low frequency resolution, and is known to be linear to its 11kHz resonance frequency. If we discover any significant resonances above that, we might use the microphones too.

Endevco Picomin 22 positioned with wax on an SME 3009 headshell

To test our rig and measurements, we first obtained some results from a "naked" headshell with no cartridge mounted. We also chose an "interesting" one, that is already known to resonate a lot.

This is what an impulse response from the SME 3009 headshell looks like:
Plotting the spectrogram (spectrum development over time) of the recorded impulse makes things very clear indeed. There are strong modes at 450 and 600 Hz, and then at 6, 8 and 10 kHz. Damping is relatively poor, with the main modes clearly visible 200 milliseconds after the strike.
Some people prefer "waterfall" plots to spectrograms. They're certainly prettier, and it's easier to read the magnitude off them. Here's the same impulse shown as a waterfall plot, with a 6 kHz mode dominating:
We've used Room Equalization Wizard software to generate the above charts. For some reasons, the software that draws nicer-looking charts, like ARTA, is limited to about 10 millisecond impulse duration.

So, what's the plan?

In the next few posts, we plan to do two IET measurements for every headshell. First, as shown above, with no cartridge attached. Then, with the cartridge screwed on, and with the hammer striking the cartridge body, not the headshell itself. This should give a better idea of real-world performance.

Once the theoretical winner is established, we'll do the subjective listening test: the worst-performing headshell versus the best-performing. This should be fun!
Please subscribe to receive blog updates in your inbox!
comments powered by HyperComments