A website for the serious amateur violin maker, restorer and tinkerer. A violin's front and back can be tuned using tap-tones. Measure the tap tones and plate weights and adjust them to get the best sound, the kind of sound you want, or make an instrument that is easy to bow.

This site can help if you are making a violin or you want to improve  a low cost violin or viola.

By tuning and weighing the top & back plates be confident that you will get a really good instrument that responds well and can sound like a $1500 instrument.

Like the best coffee ...  LATTE!

tapping belly 2 sml

Light-weight plates, &

  A is for Arching,

   TT is for Tap-tones, &

      E  is for Edge-work.

post-25136-1224022475 Strad back graduation V1.1 smll1
Violin plate Modes and Weights 1.1 smll

email: webmaster @platetuning.org

 Last updated:

1st.. Jan. 2021

Copyright  (C)



Evaluating a violin’s tone.

There is no accepted way of evaluating the tone of a fiddle: everyone, including me, has their own idea of what a fiddle should sound like! I bet that helps. But it also depends on what you want to do with it: a bluegrass fiddle needs to be quite brash to be heard over the banjo and various guitars - but is probably not suited to classical at all. A chamber instrument (for quartets etc). needs to be responsive and intimate or it can be overpowering.

Squawkers and squeakers.

Well, there’s a lot of squawkers and squeakers out there among the cheap violins .......... so we do need some way of assessing them with fact rather than opinion.

More objective methods.

 A simple way of checking to see how even a violin or viola is across all the strings there is a simple method first used in the early 20th Century before oscilloscopes and computers. It is described in various books on the physics of music, and it’s easy and quick to do these days. You just play every semitone from the bottom G open string upwards as loudly and evenly as possible while recording the violin. Here is an extract from Alex. Woods “The Physics of Violins” (1944) showing response curves for strads. and other violins using this method.

Ideally a sound level meter should be used with the whole experiment done in in an anechoic chamber [no echo’s from the walls etc], but a mic and a well-damped room or workshop will do. Here’s how Carleen M Hutchings describes it, and shows what an ideal (i.e a Stradivarius) violin response should look like, and where the key 3 resonances should be. Open strings tend to sound overly loud (see below) so it’s better to use 3rd position for the D, A and E strings.

Play each note for about 1 second standing say 2 feet from the computer’s microphone. Leave a gap of say one second between each note: a silence. You can even say out loud the note’s name, it helps.

Then on the playback waveform, select each note with the mouse and read off the (rough) RMS* amplitude measured in dB, from the meter in Audacity.  (It did work better with ‘professional’ software such as ‘Cool Edit 2000’  however, where the RMS energy can be read off the selected waveform using  Analyze => Statistics,  directly  in dB)  **.
bowed loudness curve vln100

    I then plot the response using these figures (dB, vertical scale, uncalibrated) for each semitone using a logarithmic horizontal frequency scale so that equal octaves are equal distances on the x-axis. Note that I used open strings (D, A, E) which therefore appear about 2dB louder than other notes.

You can see that the response of a fiddle, even a  strad., varies by some 10 dB, or about a 10 times ratio of sound power max. between various notes over its range. This seems to be a lot, but this is also where the instruments tone colour comes from too.

What does this tell us?

Well, if you’ve paid $20 k for a fiddle it tells you whether the maker got the fundamentals of the physics right. Note that Carleen was not impressed by the “Peter Guarnerius” of her example, but how a fiddle responds to attack of the bow is more than just how loud it is.

This fiddle example is showing less response for ‘G’s, while similarly 2 other fiddles I’ve plotted responses for show less output on ‘D’s. So we can use this technique to find out or to confirm our impressions of a fiddle’s strengths and weaknesses, and put numbers to them. That helps, especially in suggesting what to change and do in the next fiddle!


* RMS = root mean square:  a measure of the total energy in a sound wave.

** Select say 500 -1000 mS of waveform, and 0dB = FS [Full Scale] sinewave.

  What will it take?   You will need :-


The right tools, and

       plenty of time.

Be prepared to make a lot of mistakes, so start on a cheap violin. A faint heart ne’er won the maiden. Have a mistake on me.

If your working on the plates of a new  fiddle your making, or you have one completely in pieces you’re lucky! The weights and tap tones will be there ready for the asking or the measuring.

  It takes me about 10 hrs. per violin to take off the front and fingerboard off a ‘factory fiddle’, modify **  the front and back to appropriate tap-tones and put the fiddle back together. On top of this is any time for any repairs, like a sound-post patch and varnish touch-up:  but be warned. This is without having to take the back off the bouts (sides) and glue them back on.

Removing the front - the belly plate

      Removing the back can cause a great deal of damage ........., but I have found ways of measuring (deriving) the weight of the back, and also the Modes 2 and 5 frequencies of the free plate while it is still in the bouts with the neck in place.  

    The Mode 5 of a back plate is reduced by about 15%*, but sometimes splits into 2 frequencies up to 40 Hz apart, usually either side of 300 Hz.

    Mode 2 of a back in the bouts  is only slightly increased, but the neck (with no fingerboard) has a resonant frequency at almost exactly the same frequency, so we have to move the neck’s resonance out of the way!

    Get in touch with me if you want to know more.

    Use a shortened table knife to remove the front plate (the belly) and the fingerboard, but do be careful. You are trying to ease the knife into the joint and ‘lever’ the plate off, not to cut into the joint.

    Removing the front (belly) is made much easier if you carefully feed into the joint a little warm Isopropyl alcohol mixed with water (50/50). I have used it heated in a baby bottle warmer to 50 degC, and the belly came off in about 10 minutes with very little damage, but it can take a day. Note that the alcohol may well damage or soften some varnishes, so experiment first, and be prepared to touch up the varnish carefully later.

   When all the belly is nearly off, feed a small amount of the warm alcohol / water fluid down or via a long bladed knife to get it into the top block joint with the belly held slightly open. And with that the belly should come off relatively undamaged.


         I’ve managed to cut myself  quite badly doing this on my left hand, as I’m right handed. Always cut away from your left hand, as the knife can suddenly fly out of control as the joint gives way, so using a leather gardening glove on your left hand reduces the risk of a serious wound a lot!


*: Footnote on accuracy: As this method measures Mode 5 indirectly then there is a possible inaccuracy. I estimate that using this method there is a Standard Deviation of ~4.1%. That means that my estimates are 68% likely to be in error by not more than 4.1 %, and 95.5% likely to be out by no more than 8.2%. There's more detail on this here on the Wikip’dia page.

  ** This is by using the thumb plane(s) on the inside of back and front plates, leaving the varnish intact. Obviously you cannot alter the arching at all with this method unless the back is over-thick and you are prepared to refinish the back.