[ Home | Staff & Contacts | DIY & Tweaks | Listening tests | HiFi Playground | Music & Books ]

The Great VTA experiment or "Knowing Me Knowing You - AHA" (apologies to ABBA...).

[Italian version]

Author: Geoff Husband - TNT France
Published: October, 2004

Phew! Well I thought my VTA article would be controversial, but I underestimated how upset some people would get for merely 'thinking aloud' about the problem. And yes it is a problem. We have a controversy where people have deeply entrenched views. In one corner we have Roy Gandy, designer of the most popular and influential tonearm of the last 50 years who thinks the whole VTA thing is rubbish. In the other we have Roy Gregory who thinks that with some records, changes in arm height of 0.018mm produce "vital" changes in sound. They are both credible figures. They both have armies of supporters in the industry and amongst informed listeners.
They both can't be right and to just to float around in some compromise 'mid ground' is a cop-out. The single biggest problem is that most of the argument on 'VTA' towers and the like is based around personal impressions or laughable 'experiments' that would end up in the bin if submitted to any scientific journal. It's my aim to get as many people as possible to undertake the following experiments so we can at least have something with vaguely scientific credibility to argue over. Please spread the word far and wide and get as many people as possible to take part and pool their results.

An apology - In a previous life I was a science teacher, if during the following I sound pompous, hectoring, patronising and smug, forgive me because it comes with the territory. Think yourself lucky you are out of range of the board rubber... "pay attention class!"

What makes a good experiment?

Let's look at a bad one... Magazines tend to avoid 'blind tests' because they take a lot of time and may throw up results uncomfortable to them. But they do produce tests which masquerade as 'blind'. We've all seen them. The favourite subject is cables because they are easy to swap around and also leave plenty of 'fudge' room afterwards :-) In a typical example a 'tester' invites a few pals round to his house. Over a cup of tea he (sorry - it's always 'he') plays some music. Every so often he goes up to the system, everyone closes their eyes and he swaps speaker cables. They all listen some more, if it's a good test they don't talk to each other, and make notes but usually they chat. Then the tester goes off and whilst everyone closes their eyes he changes the cables again.

Great - a perfect scientific blind test :-) - er not really... The whole test is utterly worthless. The single biggest flaw is the 'human factor'.

The Human Factor

We have evolved for millions of years to become the most socially sophisticated animal on the planet. We talk, we gesture, we interact. This is all conscious, but we also do a mass of things unconsciously. 'Man' is, for example, the only animal to lie in a sophisticated way. We can spot lying. Sometimes we don't even know we do it. When we lie we send out a whole series of subtle clues, most of them 'displacement activities' of some form or another, and others can spot them.

But more important are the unconscious abilities we have that help us maintain a bond in a group. That group of friends on the couch listening to cables, will not know the cables being tested, but they will be on-their-guard and looking for clues to stop them looking a fool. The tester will no-doubt already have opinions on the cables in question, his body language or speech will be different for a cable he likes to one he dislikes. I'm not talking about a 'parody' like "now guys listen to THIS one!", but subtle unconscious clues that the panel will unconsciously pick up. If the tester tries to put in a 'control' by not changing the cable (though strangely this basic requirement of such a test is rare!) the tester will radiate 'lying' clues.

The group will also want to conform. Given a tricky question like "can you hear this?" the group will look for clues and try to find the position of others - are they making 'positive' of 'negative' moves? So, such poorly run 'experiments' often end up with wonderfully conclusive results with everyone agreeing with each other. And because they carry the 'blind' moniker they gain totally unjustified credibility.

You doubt the influence of these factors? Well some people are specialists in consciously reading these signs. We call them 'magicians', or fakirs, or 'mind-readers', or 'mediums' or 'gurus'.

A true story - last week a friend of mine went to a wedding anniversary party and there was a magician there. He specialised in close up magic, card tricks, slight-of-hand etc. The last thing he did was to go round the table with a pack of cards - imaginary cards... He went to each guest and said "pick a card, any card" and held out his imaginary pack of cards all fanned out. The bemused guest pulled out an imaginary card and he said "Don't show me, now look at the card and remember it, now put it back in the pack". The guest put the card back in the 'pack' and he said "do you remember the card, sure? It was the four of clubs!". The astonished guest admitted he was right. He then went round the table and did the same with every guest. By the 12th and last people were really on their guard, but still he got it right. It was a stunning demonstration of the power of suggestion and 'reading' peoples unconscious signals. Compared to that, figuring out whether the guy next to you likes the cable he's listening to or not, or what the tester thinks is a doddle even for the layman!

So any experiment must rigorously exclude any possibility of the human factor and that is why the experiments below set such strict instructions as to how they must be done.

The Experiments

The first thing I should say before we begin is that to prove a theory is impossible. Here we have a hypothesis, "that changing arm height has a critical effect on sound quality". The following experiments may disprove the hypothesis, but will not be able to absolutely prove it is correct. They may be able to put up very powerful evidence for it but no more. All scientists are 'devils advocates' and will look for the tiniest chink in the armour of any theory.

So after that rather long preamble here are the two experiments. If you don't have time just do one of them, the choice is yours. You'll see that I have not used the abbreviations VTA or SRA because I am not convinced that either is the critical factor that we are changing. We may be optimising something, but to use one of those terms pre-judges the result. The only thing that I am prepared to accept we are altering is arm height, so the pedant in me forces me to use the abbreviation AHA (Arm Height Ajustment).

For these experiments you will need one or more flat records. I don't mean flat I mean FLAT! It's pointless undertaking an experiment on AHA when the basic material has a runout of +/- 0.1mm - Finding one may be the most difficult single task you have to perform. As the test progresses, especially if several 'subjects' are tested, you may need more than one disc as otherwise a track may be played more than 50 times and thus be damaged. Sorry but you'll also need an arm with micrometer adjustment of AHA. I know that excludes many of you, but surely you could find a dealer prepared to put this to the test, organise a group of friends to go - make the effort! And on the question of effort, I accept that the tests are going to be time consuming and not particularly riviting to undertake, but some of you may be tempted to spend thousands of pounds on an arm you don't need, or alternatively you may find you would really benifit from such a purchase. Don't you want to know if it's worth it?

Experiment One - An attempt to find the critical limits of AHA in your system and your ears

Equipment for this experiment = a blindfold + a stopwatch + a pre-prepared results sheet.

Before the test the 'subject' should alter AHA to what they consider the best possible AHA for the disc in question. This will become the 'baseline' position for the experiment.

The subject should be seated in the best position relative to the Hi-Fi. The 'operator' and turntable must be totally unseen, make no noise whatsoever (if you cough you have to repeat that set of tests) during the entire time of the experiment. The operator could be behind a screen or the subject could wear a blindfold - after all, the subject only has to make ten marks on a piece of paper. No-one else can be in the room and the operator must not move any more than is absolutely necessary to operate the turntable. If ANY of these rules are broken the experiment is invalid.

For each 'run' the operator will play a section of a track of an album, chosen by agreement with the subject, eleven times. Each run will begin with the AHA at the baseline position. After that baseline play, on five occasions the AHA will not be changed but the adjuster will be turned and returned to the 'baseline' position during the intervals between music. The other five occasions the AHA adjuster will be altered to raise the AHA 1.0mm. The order of play should be random and the operator needs to write down the order in which they are going to do the test beforehand and stick to it. Remember that before each run the 'baseline' position must be played for 30 seconds. The orders should be be one of the following, used at random - 'A' being 'change, 'B' being 'no change'. A,B,B,A,B,A,A,B,A. or B,A,B,A,B,B,A,B,A,A. or B,B,A,B,A,B,A,B,A,A. or A,A,B,A,B,B,A,A,B,B. It's easiest if you use squared paper and leave spaces next to the operators plan for the subjects results, this will aid later analysis.

The operator needs to practice the procedure before the experiment so that he/she can play a 30 second section, then in the following 30 seconds alter the AHA and then after that time recue the same section of the record once again. The subject will hear 30 seconds of music, 30 seconds of silence, 30 seconds of the same music ... and so on. After each section the subject writes down 'A' if they think they hear a change from the baseline, 'B' if they think they hear the baseline position.

At the end of the ten plays (+ the start baseline) you can stop and compare notes if you wish (just make sure the subject doesn't see the operators notes for the next test!)

Now repeat only this time the operator lowers the AHA 1.0mm for five of the ten excerpts, the other five being no change.

You now have results that show the subjects ability to distinguish AHA variation down to +/- 1.0mm. If the results show that the subject cannot reliably identify changes i.e. they make one or more errors, then repeat the experiment with identical settings just to make sure. If there are still mistakes the subject cannot reliably distinguish AHA variations of +/-1.0mm and so the next test should be for +/-1.2mm Keep repeating the experiment until the subject gets straight 10/10 for two repeated samples. You need to repeat like this because by chance you are quite likely to get 10/10 sometime, just like sometimes you can throw three sixes in a row.

Assuming the subject gets a straight 10/10, you then repeat the experiment with the up/down change in AHA being 0.8mm. Then 0.6mm, and so on right down to the resolution of your arms AHA adjuster.

When the subject reaches their limit of resolution go back one and repeat the previous test to make sure the 10/10 they got with that test was not down to chance.

Interpreting the results

What this test should show is the limits for critical AHA adjustment for the combination of subject and system. If you find that the resolution is say +/-0.6mm you know that you are safe in setting a single AHA which will cover the thickness of most records - congratulations you can go back to fiddling with your records and not with the arm... It would also be pointless to continue with experiment two.

There is of course one gaping flaw in this experiment. It doesn't cover the problem that the subject may be hearing the altering of the mechanism of the arm as described in the accompanying article. I've tried to cover this to a small extent by making the operator move the adjuster even when the AHA is not to be moved, but the possibility still remains. If this is the case though I would expect the results to look rather 'scattered' and independent of the distance AHA is altered. But mechanical change is a highly possible explanation for a set of results.

The second flaw is that this experiment only measures whether we can hear 'change'. Not whether this change is significant, or musically important i.e. does it sound better! This is the purpose of the second experiment.

Experiment Two - To find how accurately the subject can identify the AHA position which gives the most musical result

N.B. If the subject has already failed to distinguish small AHA changes then the next test is pointless and you can all have a drink and listen to music.

The subject chooses a favourite record that they are happy is sensitive to AHA change. It must be very familiar to the point of boredom if possible. The subject now can spend as much time as they want getting AHA just right for that record. This will be the first recorded AHA position.

Once again the operator must be out of sight, silent and unmoving. The subject must be 'blind'.

Regaining the preferred seating position the operator will move the AHA adjuster away from the preferred position, either up or down, sometimes a long way, sometimes not so far. The subject then calls out 'up' or 'down' to the operator who acts as a kind of remote control i.e. efficient and SILENT :-) The subject, when they are perfectly happy, just says "OK that's it" and the operator notes the position, randomises the adjuster and repeats the experiment. There is no limit to how often you do this and to how many records you chose, but a minimum of ten tests should be done.

Interpreting the results

Now you should have a 'scatter' of results, which will show to what accuracy the subject can pick 'perfect' AHA, from a musical point of view, with a variety of records.

The flaws? Well the subject could still be zeroing in on sweet-spots in the arm mechanism, but if you use different thickness records this would mean that you would be 'passing through' several sweet-spots. If you find that every record, regardless of thickness, needs the same AHA then you probably do have a sweet-spot problem.


This is a serious attempt to solve a long, long argument. If you don't follow the experimental conditions to the letter then your contribution will be worse than useless as it may incorrectly skew the results. Please be honest :-)

Reporting back

Please send all your results to be at this address - please include the system and records used. Then after I have collated the results I will post them, and some tentitive conclusions in an article here. Thanks in advance

Note to all editors, writers, shop owners, journalists, clubs etc. Please feel free to copy and use this article, all I ask is that the results come back to me. If you are running a hi-fi show what a wonderful experiment to carry out on visitors? HELP!!!! It is my hope that with a big enough sample we can at least learn something about what goes on with AHA and whether further investigation is necessary.

© Copyright 2004 Geoff Husband - www.tnt-audio.com

[ Home | Staff & Contacts | DIY & Tweaks | Listening tests | HiFi Playground | Music & Books ]