Tuesday, April 3, 2012

Equalizing the strain gauge

I won't have time to illustrate this fully now, but just a quick overview.  The strain gauge cartridge is position sensitive rather than velocity sensitive like a normal cartridge.  Relative to a position sensitive device, a velocity sensitive device has a rising frequency response characteristic, rising by 6dB per octave.

For phonograph records, the output of a velocity sensitive device is equalized downward by a familiar RIAA de-emphasis curve.  That curve slopes downward mostly, but is nominally flat below 50Hz and between 500 and 2122Hz (and, though this isn't usually spelled out, it should go flat above 20kHz and then ultimately curve down to infinite frequencies).

For a position sensitive device, we are do no equalization where the familiar curve slopes down, but must equalize up with frequency by 6dB per octave where the familiar curve is flat.  This means specifically the bands below 50Hz, between 500 and 2122Hz, and above 20kHz.  Well we really don't want to equalize up above 20kHz, so just forget that one.  Below 50Hz causes problems too, but I'll get back to that later.  The critical region requiring equalization is 500Hz to 2122Hz, right in the middle of the midrange.  In this region, the treble must be shelved up at 6dB per octave to be the same as a velocity sensitive cartridge with no EQ.  Since this region is just over 2 octaves, the total amount of equalization should be about 12dB.  I have read on a blog that the exact number is 12.6dB, and that is quite plausible, though I didn't calculate it myself.

All this is basic audio engineering, and was incorporated into the strain gauge cartridge electronics created by Panasonic in the 1970's primarily for the purpose of reading CD-4 records.  CD-4 records are just like regular records except they have a high frequency subcarrier for the second set of channels.  CD-4 playback equipment then obviously needed to have very good and predictable high frequency response.  Panasonic chose to deal with this using a strain gauge device.

However, almost from the beginning, audiophiles realized that these cartridges were special in various ways, and desireable for ordinary stereo playback.   Almost before you could blink, ingenious Americans like John Iverson co-opted the Panasonic cartridge for his legendary EK-1 phono playback system.  I don't know much about the EK-1, those are very rare.

But other strain gauge promoters who were not so hung up on ordinary audio engineering practices started producing high end strain gauge cartridge systems without using ANY electronic equlization.  They did in fact have cartridges that needed less equalization than perfect strain gauge cartridges would because they already had a slightly rising frequency response characteristic.  One of these Strain Gauge promoters was Sao Win, and his preamplifiers used NO equalization (and sounded mostly OK).

It turns out that even the Panasonic cartridges are not exactly perfect either.  Just like the Win, they have a slight (but very slight) rising frequency response above 500Hz.  They still need equalization for truly flat (correct) frequency response, but less than the 12.6dB that would nominally be required for a perfect position sensitive device.

A few years ago there was a resurgence of interest in strain gauge cartridges, including NOS Panasonics, used Panasonics, and cloned Panasonics.  Though these cartridges, as always, need about 10dB of equalization, new promoters are one again trotting out the audio BS to justify why their system requires no "phase altering" equaliztion.  (Actually, the basic cartridge system is minimum phase, so correcting the frequency response *IS* correcting the phase.)

I couldn't remember what exact amount of EQ was required when I started setting up my system last Saturday night.  I thought it was about 10dB so I tried that.  It sounded OK, maybe a bit thin.  Counting on two octaves of equalization, I then tried 12dB of EQ.  That was horrible.  Rolling the EQ down to 6dB gave a nice warm bassy sound.

Setting up a shelving EQ in parametric form on the Behringer DEQ 2496 is strange.  You can't select two frequencies, you can only select one frequency, the direction (highpass or lowpass), and then the total amount of EQ.  Then while you nominally select highpass or lowpass, you can set the amplitude either positive or negative.  If you use negative amplitude with a lowpass it's actually more like a highpass.

Rather than trying to sort this out by ear, I found a test record with wide band pink noise (unfortunately, only about 20 seconds) and then octave band filtered noises to test each octave.  I made adjustments and then tried them out on either the pink noise or the octave band tests.  I can set the behringer RTA display to show the frequency response of the equalized output, which makes this very nice.

I ended up setting the fixed frequency at 2122Hz and then using 9.5dB of cut below that (highpass with positive amplitude).  I also equalized the frequency extremes from 20-40Hz, 40-80Hz, and 10k-20kHz very slightly using the 1/3 octave graphic equalizer.  There was a 2dB bump  in the 10-18kHz range so I adjusted that 2dB down, but diminishing at 20kHz where the bump had dissipated.   Iin the 40-80Hz range so I adjusted that 2dB down.  There was a 3dB depression below 40Hz so I adjusted that 2dB up, but after playing an organ record with deep subsonics, I decided to roll that back to 1dB up, and only 0.5dB at 20Hz.  I am a bit concerned that the test record (Soundcraftsman) might not be entirely accurate at the extremes, but it was intended for setting octave equalizer levels for the extreme octaves, and should be pretty close, and shouldn't be designed to make you overcompensate.

It now sounds spectrally correct and wonderful mostly now.  Neutral certainly.

This gives me roughly +/-1dB response at least from 25-20,000 Hz.  I think the rolloff below 40Hz might be a function of my tube strain gauge preamp which has no equalization.  I should probably check that out and fix it.

The "preamp" by the way, I don't know much about it, it has some dymo labeling tape on it saying "Bruce Moore Audio Stage" and "Strain Gauge."  I don't know whether that means it was made by THE Bruce Moore (who now seems to have a thriving audio business) long ago or whether somebody copied a Bruce Moore design and modded it for strain gauge.

It's in a thin gauge aluminum box and has exposed (but covered) tubes, capacitors, and transformer in the back.  The tubes are 4 6DJ8's (2 per channel) and 2 12AU7.  I believe the circuit is two 6DJ8 cascodes followed by a 12AU7 double cathode follower.  There is a little switch for Win Labs.  That switch does nothing other than reduce the gain by 6dB in the Win Labs position.  I found I needed to use that switch to keep output levels below 2V for the DEQ which I run on its 2V setting.  Alternatively, I could use the 10V setting and turn the switch to the other position.

But the strain gauge magic is actually in a little box that goes between the turntable output and the preamp.  The little box is supposed to have two 9 volt batteries placed on the inside (but I put them on the outside).  A resistor network reduces the 18 volts to 15 volts IIRC, which is applied in series across both channels, with the channels out of phase with each other.

Given that the output of the magic box is fairly high (close to line level) and low impedance,  I think you could use just about any line level preamp or possibly no preamp at all, just straight from the box into the DEQ !!!  You also need some way of reversing the polarity in one channel (the DEQ can do that).

No comments:

Post a Comment