Arm tests

Analysis of an arm with clearly linked vibrational modes. Not all are like this, but a basic mode around 200Hz for a 12in arm, up to 500Hz for a light, stiff wrapped carbon fibre tube is a common basic pattern.

WHAT IT TELLS US

For a cartridge to read the music in a groove it must remain absolutely still in the headshell. The headshell must not move at audio frequencies or information will be lost. In practice arms vibrate in various complex ways, excited by the cartridge. Our measurement of vibration in the arm shows how severe this problem is, and what form it takes.

Vibrational modes and arm stiffness impact sound quality in audible ways. Stiff arms that are undamped ring in the midband and have a zing about them. The Linn Ittok was an example of this. Arms tubes flex (first order bending mode) at around 200Hz. If severe this degrades image solidity and sharpness, and also stage width, across the lower mid-band. It also weakens dynamics. The basic stiffness of the Rega RB300 tapered, cast alloy tube gave it superb sound staging and great dynamics, audibly superior to other arms at its introduction. Measurement shows this.

As well as various bending modes, often but not always harmonically related as above, arms also suffer high frequency effects that our measurements suggest are headshell related.; they are eliminated by clamping the cartridge body direct to the arm tube. These modes and others around the midband can colour the sound slightly. Eliminating them reduces colouration and makes for a neutral and well resolved midband.

The popular Rega RB301 arm. Its main bending mode lies at a high 280Hz and harmonics can be identified on the harmonic markers.

Our B&K accelerometer is powered by the on-board supply in the HP3561A analyser.

HOW WE MEASURE IT

We attach a Bruel & Kjaer accelerometer Type 4517 to the headshell, above the cartridge and just forward of the headshell screws. Position is important. The position we use gives a general and representative result. Further forward by a few millimetres, at the front of the headshell, makes resonant modes far more apparent, proportionally more than the change of distance.

A position behind the cartridge lessens amplitudes but gives a similar pattern, as does attachment to the counterweight and side of the cartridge.

As the accelerometer is sensitive only along a perpendicular axis through it, surface headshell attachment (i.e. on a  horizontal plane) needs excitement in a vertical direction. We use a vertically cut sweep on JVC TRS-1007 test disc that runs from 20Hz up to 20kHz at a reasonably slow 50 secs.

To measure lateral excitation the accelerometer must be attached to the side on the cartridge or headshell and a lateral cut used. This gives similar results to a vertical measurement though, understanding that mechanical asymmetry assures they will not be identical.

The accelerometer is powered by the on-board supply within a HP3561A spectrum analyser and its output read by the analyser

The Bruel & Kjaer Type 4517 accelerometer, attached to the top of an arm, just above the cartridge.

An accelerometer gives an output proportional to acceleration, but test discs do not have constant acceleration cuts; bass excursions would be too great and treble ones so low as to be buried in groove noise. TRS-1007 is constant amplitude up to 1kHz and constant velocity above 1kHz. Since a cartridge is constant velocity device (i.e. output is proportional to velocity) the accelerometer should ideally read velocity; its output falls at -6dB/octave relative to a constant velocity cut  and -12dB/octave relative to a constant amplitude cut. This means our results visually understate conditions at low frequencies and a post equalisation network that corrected this would make low frequency disturbances look far more severe relative to the treble disturbances.

Absolute levels are interesting and more important. In a resonant arm a 0.3g peak at 200Hz is common and it has a velocity of 0.2cm/sec. The mechanical velocity on the disc at 200Hz is 1.38 cms/sec. This puts the headshell resonance -17dB below groove excitation, so it is no small effect.  Arms nowadays have peaks around 0.1g and if at 200Hz such a peak will be -26dB below the signal in the groove, a better result than that of the ‘ringy’ arms. However, the energy in the system is a function of the area under the curve and short narrow peaks encompass little area, whilst Rega RB300 based arms, for example, lack high peaks but have distributed energy seen as broad low level disturbance, our measurements show. Rega arms still perform very well against most though. Few manufacturers yet measure the resonant behaviour of their products and our accelerometer measurements of pickup arms are unique at present.