Jump to content

Tuning Precision & Accuracy


Recommended Posts

Dave,

since your original post I have been thinking that your reason for starting this topic was to expose the current poor quality of some restoration work, particularly in the area of tuning. Now that you have stated that to be the case I will add that over the last (very) few years several Concertinas that have supposedly been recently refurbished in the UK have passed through my hands. Not one of them was tuned acceptably well in my opinion.

 

I can see that this state of affairs annoys you and could reflect unfairly on your standards of work if it is a more general malaise than just the few instruments that I have seen.

 

With my support,

Geoff Wooff.

Edited by Geoff Wooff
Link to comment
Share on other sites

  • Replies 130
  • Created
  • Last Reply

Top Posters In This Topic

Dan,

 

I'd be surprised if Faan Harris' Anglo was anything other than "stock" Lachenal tuning, because it used to be common for players to have to replace a free-reed instrument (be it a concertina, a Paolo Soprani accordion or whatever) when it got too far out of tune/became unplayable, simply because there often wasn't anybody to tune/repair them.

 

Or do we know of anybody doing custom tuning of concertinas in South Africa in 1932?

 

Maybe you should try playing his music on a Lachenal...

Thanks, everyone, for the helpful comments. I have avoided retuning my Dipper to meantone...just being conservative, as it is my main squeeze. Maybe I should try Stephen's suggestion and look for a high class metal-ended Lachenal. If it plays fine in ET, as Chris suggests it might, then I would be home free, but otherwise I could then try meantone retuning, as Geoff and Greg suggest.

 

I did recently buy a 1950s Wheatstone to try out the Harris tunes on, but was disappointed with it. I bought it because Lachenal and Wheatstone tunings are different in the upper right hand row than my Dipper, which has Jeffries tuning. Harris really uses that upper row, and it is amazing how much easier some of his tunes are on a Wheatstone tuning than on Jeffries. I think that is merely because he arranged his tunes on a Wheatstone/Lachenal.

 

If anyone sees a high class Lachenal metal ended CG, please let me know. I don't want any rubbish, though!! And many thanks again for the ideas.

Link to comment
Share on other sites

Dan, it would be interesting to set up a simulation of a typical Faan Harris musical event to see what you might learn. Without knowing anything of his music and times, I visualise (and audialise!) a noisy dance scene with a probably well-lubricated audience. Do we have any idea of the size of the building and the number of people involved? I don't visualise a PA system, so Harris has to reach to every corner of the venue on acoustic power alone, with only guitars and cello support. In Ireland, this caused many flute players to take up sax (we had such a chap here in Australia. Old Tim Whelan prefered whistle, but had taken up sax when he ran a dancehall in Dublin. He also made up a 6-key whistle using the head of a recorder and the body of a flageolet!). Were the tricks Harris used part of his strategy to deal with big crowds in noisy venues?

 

You also mention poor quality recordings. By the end of the 1920s, electric recording was pretty well established as the norm in big centres. But it could be that he didn't enjoy that luxury, in which case the sound was cut into the wax again on acoustic power alone, in which case he might have needed as much power as he could muster, rather analogous to the dance hall. Alternatively, some of the early electric recordings employed pretty poor microphones. I'm not real good on this era of recording, but there are plenty of experts out there who could tell you more.

 

I guess what I'm trying to say is what seems like an acoustical abomination to us might have been bread-and-butter to him.

 

Terry

Link to comment
Share on other sites

 

Hmmm. Doesn't seem to respond to notes below F#3. Presumably these are below the range of the flute.

Look under the General tab in the preferences.

 

 

Well spotted, Alex. Scott put the default F3 lower limit to kill off low frequency background noise below the lowest note of the lowest flutes likely to be used. But it can be turned off or set lower in the Preferences.

 

Note you can also choose between Equal Temperament (the default), Just Intonation in D, or two custom temperaments you put in yourself.

 

You can also set the bars to change colour, from green to yellow and then to red at pre-determined deviations. If anyone's concertina shows a yellow or red on the default setting, it's in deep trouble! Not even flute players will want to sit beside you.

 

Here's my concertina with colours added. 1 cent or better is green, up to 2 cents yellow and 3 cents or worse in red.

 

post-11004-0-32899900-1395359515_thumb.gif

 

Seems to me that widespread availability and use of Flutini or ideally Concertini (with Dave's World Standard for Concertina Tuning Acceptability programmed in as default) would go some way to putting pressure on the shonky operators. Nothing like getting a few instruments sent back with colourful graphs (and equally colourful cover notes) to make people lift their game.

 

Interesting question how would you make use of two colours. Does it turn Yellow at 1.5 cents, saying "ideally this should be better", or turn red, saying "this is out of tune"? Or would we reserve red for Rossing's jnd (8 cents) or some other "this is out of tune" criterion?

 

Terry

Edited by Terry McGee
Link to comment
Share on other sites

Terry,

 

Flutini is great, but I do not see not much use as a workshop aid on English & Duet systems, other than for health checking. Why? two reasons:

 

1. it only outputs in integers cents

2. These two systems play the same note in each direction of play, so you end up with a gross or averaged error for each note, add to this the en-harmonics of G#, Ab, or D# and Eb, you don't know which of the four reeds is in error

 

You can, however, show that an instrument is 'in tune' or if it shows errors A Concertini version would need to be able to manage a multiplicity of reeds playing the same note. Obviously there are doubled reeds on Anglos which will have the same problems

 

Dave

 

I think #1 could be easily dealt with. Tartini seems to offer 0.1cent resolution - I imagine Scott has rounded this to 1 cent in Flutini, so it should only be a matter of unrounding it.

 

I think #2(a) would just require adopting a strategy. On sweep one, just play push notes. Flutini doesn't mind waiting - you don't have to keep the notes coming. So play as many notes as will fit on one breath, open, and carry on. And go round the loop a number of times if you want more averaging of your bellows pressure.

 

So supposing you are exploring a 56 key English, you might give each note about one second, so it will take a minute to play them all (plus gasps-for-air-time). You might regard that as good enough for the initial test. Print that screen to save it. Then spend another minute doing the pulls. So in maybe three minutes all up, you have a printout of the pitch of all the notes on the instrument.

 

Now suppose you've done your tuning and put the instrument back together. And you have a very picky customer who's prepared to pay for proof. You might go around the push loop say 5 times, to average out bellow-pressure effects. Then the pull loop 5 times. In less than 15 minutes you've printed out a list of the (hopefully) finished product you can give the customer.

 

Your #2 part b, how to deal with enharmonic notes, would require some fiddling. Currently Flutini regards anything from Eb-50 cents to Eb+50 cents as Eb/D#. We would need to add more note bins to collect the data into. The D# bin would include anything from Eb-50cents to Eb-0.1cents. The Eb bin would collect anything from Eb+0 cents to Eb+50 cents. There is nothing tricky about doing this. Flutini currently has 12 bins, we just need to add the remainder and set the limits accordeoningly (as we say in the music industry).

 

I've had to do something similar to deal with a peculiarity of early 19th century flutes. They can have low notes so flat that low D, for example, crosses over into C# territory. Which gets very confusing, and completely inaccurate around the crossover as the notes vacillate between two bins. So I fiddled the bin separation points, creating a special case where low D is anywhere between -100 and 0 cents flat, above which it is an Eb, and so on. I stretched my measurement scale to match the situation. So my graph range is no longer +/-50 cents, but -100 cents to +50 cents. I did that in the RTTA Polygraph, as the language R it is written in is quite easy to understand, even for humans. That might be an option if we can't find anyone to rejig Flutini.

 

Terry

Edited by Terry McGee
Link to comment
Share on other sites

I find the very low notes are the most difficult to tune as they tend to vary the most with bellows pressure...

 

Not just with bellows pressure, it is common for tuners to have trouble locking to the fundamental partial in low notes when the first or higher partials are louder. Locking the tuner to the fundamental helps.

Link to comment
Share on other sites

Thanks, everyone, for the helpful comments. I have avoided retuning my Dipper to meantone...just being conservative, as it is my main squeeze. Maybe I should try Stephen's suggestion and look for a high class metal-ended Lachenal. If it plays fine in ET, as Chris suggests it might, then I would be home free, but otherwise I could then try meantone retuning, as Geoff and Greg suggest.

 

If anyone sees a high class Lachenal metal ended CG, please let me know. I don't want any rubbish, though!! And many thanks again for the ideas.

I'm not sure buying a high class Lachenal is a definitive solution. If by high class you mean one which plays easier and louder than the average Lachenal then you will by definition be buying better reeds, which will mean better clearances, and more of the baggage you are trying to avoid, more higher partials.

 

If it is a matter of degree, (is there anything that isn't?) then perhaps there is a compromise you can live with, and maybe a high class Lachenal is it, but I wanted to warn you it is not a binary equation, more a choosing your position on a slope.

 

Might it be possible to get someone with a high class Lachenal to let you have a try or to record the intervals you want to use and send them to you to save you the potential $ pain if it is not an adequate solution?

 

 

Cheers

 

Chris

Edited by Chris Ghent
Link to comment
Share on other sites

 

I find the very low notes are the most difficult to tune as they tend to vary the most with bellows pressure...

 

Not just with bellows pressure, it is common for tuners to have trouble locking to the fundamental partial in low notes when the first or higher partials are louder. Locking the tuner to the fundamental helps.

 

 

And that of course is deadly in an RTTA approach, as the tuner then contributes the low note's data to the octave above (or whatever partial it thinks is the fundamental). So you end up with not just a missing low note's data, but a contaminated middle note.

 

I had a quick go on Flutini (set for F0 and higher) and seemed to get good reliable results over the range of my anglo (C3 to B6). I think Tartini's pitch detection algorithm is pretty robust in this sense, as flutes played in the hard Irish style still register correctly, even when the player has shifted most of the energy from the fundamental into the first two harmonics.

 

It's good to watch the Samples column in Flutini as it tells you which bin the current data is being directed into. Vital to avoid playing octaves upwards legato when measuring, as you can fool it for a while. Playing octaves downwards legato doesn't fool it. Playing non-legato is the safest bet, as it forces Tartini to recompute the fundamental. Poop, peep is good. Poo-eep is not.

 

Playing scales is fine however, even legato in either direction. Presumably the shift in pitch retriggers the fundamental recalc.

 

Terry

Link to comment
Share on other sites

Whilst there has bee a lot of thread drift, I have found much in these little excursions for me to think about, and I do need to do more reading around the physics of sound, and to understand more about other instruments and how that can be brought in to my own tuning practices. I have made no secret of the fact that I am quite disturbed by the number of instruments, 'refurbished & tuned' before sale that I am seeing (seemingly almost monthly); and that there are times when I do feel that players are not always getting the standard that they are paying for. I do feel that the responsible concertina repairing, restoring, making community should be consistent in what we would all expect to see (hear) and how we advise players on what they should expect from a re- tuned or new instrument. We need to be 'of one voice' in this.

 

Dave, Geoff and others, it would be instructive for the rest of us to see what you are seeing. Can you give us an idea of how poorly these instruments are tuned? E.G. how close to zero are the best notes likely to be, how far from zero the worst notes, what would be an average deviation?

 

Do they mostly get it right, with occasional wild deviations, or is it a general dog's breakfast?

 

I'm assuming in this that you are seeing a scatter of notes from a rough'n'ready repairer, rather than a "fixed offset" because they used an out-of-tune tuning standard.

Link to comment
Share on other sites

Dave,

since your original post I have been thinking that your reason for starting this topic was to expose the current poor quality of some restoration work, particularly in the area of tuning. Now that you have stated that to be the case I will add that over the last (very) few years several Concertinas that have supposedly been recently refurbished in the UK have passed through my hands. Not one of them was tuned acceptably well in my opinion.

 

I can see that this state of affairs annoys you and could reflect unfairly on your standards of work if it is a more general malaise than just the few instruments that I have seen.

 

With my support,

Geoff Wooff.

Geoff,

 

I have been away tutoring for a few days, so i am now catching up, and yes sloppy and lazy workmanship does annoy me, but is it that simple?

 

My motivation is two fold:

 

i do not think that there has ever been a concertina standard as such, certainly not a tuning standard that i am aware of; or concensus discussed never mind agreed, so how can we actually say that there are poor standards of work. i would like to see some standards in place, if only 'technical performance standards' and tuning is the best place to start

 

Players and customers need to be able to objectively assess if it is reasonable to refer an instrument back to repairer, and be confident in doing so. The statement of 'i think ths sounds off, or it's a bit flat' are subjective and are uncomfortable for a customer to substantiate. This leads to disatisfaction and continuing sloppy work.

 

I am not saying that the Uk is riddled with poor concertina craftsmen, but we do have global issue in that no one can (could) say what a reasonable tolerace ought to be.

 

Dave

Link to comment
Share on other sites

  • 2 weeks later...

Just a bit more on using Flutini to map out a concertina's general state of tuning. Here's an instrument I've been looking at (a metal-ended Lachenal). Unfortunately, the forum has compressed all the blank spaces between the columns into one, but you can still make out the meaning. The note name comes first, followed by the number of samples taken by Flutini, followed by the deviation in cents. You can see I did the C row, then the G row and finally the accidental row:

 

C-row: Note Samples Delta in Cents
C3 75 -6
G3 121 -3
B3 24 -5
C4 64 -9
D4 47 6
E4 83 -11
F4 87 2
G4 89 2
A4 81 9
B4 96 -7
C5 87 -2
D5 98 -2
E5 51 -3
F5 73 -2
G5 81 1
A5 58 -9
B5 88 -3
C6 102 32
E6 64 -11

 

G row: Note Samples Delta in Cents
B3 67 -5
D4 115 -6
F#4 60 13
G4 50 -3
A4 67 6
B4 65 1
C5 44 -2
D5 53 6
E5 54 -2
F5 1 2
F#5 42 1
G5 38 -6
A5 36 -13
B5 39 2
C6 36 -1
D6 70 22
E6 55 -32
F6 82 31
G6 36 17
A6 60 21

 

Accidental row:

Note Samples Delta in Cents
D3 32 -14
A3 39 -17
B3 40 -6
C#4 36 -5
D4 39 -7
E4 34 -2
F#4 29 3
G4 35 2
A4 33 -2
B4 35 -2
C#5 39 -13
D5 37 -22
E5 35 -13
F#5 36 -9
G5 41 -13
A5 37 -2
B5 38 10
C#6 57 17
D6 133 -2

 

The only problems I encountered were the repetitions of a few notes within their row. They could be easily filled in by hand. Otherwise the whole operation took about 5 minutes.

 

This test also goes some way to answering my question of how deviant are some concertinas. By a rough count, we have:

 

3 worse than 30 cents

4 in the twenties

11 in the teens

17 between 6 and 10 cents,

6 between 3 and 5 cents

15 at 2 cents

and only 4 at one cent.

 

Even if we took Rossing's 8 cent just noticeable difference as our guide, and so set a limit of say 5 cents to come safely within it, we'd have 25 notes that passed the test, and 35 that failed. On Dave's +/- 1.5 cents, we'd have somewhere between 4 and 19 notes that passed, let's take it as 12, and so 48 fails. Hmmmm.

 

Terry

Edited by Terry McGee
Link to comment
Share on other sites

I use a Sanderson Accutuner for my finish tuning and will occasionally let a midrange reed go if it is as far out as 1 cent sharp since most of the time, they will eventually go flat. High reeds seem to react differently for some reason. I find even with equal temperament, 5 cents sounds rough in general. The difference between a instrument tuned 0-1 cent and one with even just some of the reeds 5 cents out is quite noticeable. The really in tune ones sound so nice! But... as Greg says, what pressure do you use. I find mid range reeds less pressure sensitive than low or high ones. I can only guess that the low and High reeds which are relatively less stiff than the mid range are more easily influenced. Poorly centered reeds also detune more with pressure, making the whole process full of compromises. One saving grace though is that with chords, both notes tend to detune similarly enough to stay relatively in tune with each other. Listeners are generally more concerned with relative dissonance than exact frequency. Some kinds of music are also more tolerant of tuning errors. One thing that cause more global tuning changes is air density changes. Something flute players are all to familiar with. Also reed pans can pick up a surprising amount of moisture / weight in humid weather which can cause tuning changes. Concertinas can't be adjusted on the fly like most single voice orchestral instruments. People who play them need to realize their limitations and work within them.

To Greg's healing reed question, Reeds are hardened steel with balanced stresses. When you remove metal, you change the stress balance, and until the reed has vibrated enough to balance out the stress again, the reed will warp slightly and change it's set even if it is only slightly. That will change it's pitch a little. In industry, some castings are put on vibration machines to de stress them. That is one answer. might just be fairy magic. You never know.

Dana

Link to comment
Share on other sites

  • 3 years later...

I have been using a desktop strobe tuner for some time to evaluate a reed tone
when in question:

http://www.katsurashareware.com/strobe/strobe.html

More recently, I have questioned the software and found a tone
generator to test the above software:

http://www.szynalski.com/tone-generator/

To my surprise, the software seems to have been doing a good job.
I ask this because I have a bunch of old accordion reeds that I
would like to tune (just for FUN), and was considering a
lower end Peterson tuner for the project.

Does anyone have any experience / thoughts / opinions regarding online strobe tuners?

Edited by Noel Ways
Link to comment
Share on other sites

I have been using a desktop strobe tuner for some time to evaluate a reed tone

when in question:

 

http://www.katsurashareware.com/strobe/strobe.html

 

More lately, I have questioned the software and found a tone

generator to test the above software:

 

http://www.szynalski.com/tone-generator/

 

To my surprise, the software seems to have been doing a good job.

I ask this because I have a bunch of old accordion reeds that I

would like to tune (just for FUN), and was considering a

lower end Peterson tuner for the project.

 

Does anyone have any experience / thoughts / opinions regarding online strobe tuners?

The trouble with that is if you use the same computer to generate the tone and measure its frequency, it should be no surprise that it's in tune with itself. When I'm feeling paranoid, I try using several different devices to generate tones and checking that the tuner agrees with them all (so far it has, to a small fraction of a cent).

 

Incidentally I started out using Katsura Strobe Tuner but have switched over to Tonal Energy Tuner on an iPhone because it has far more features and seems to work really well. The display doesn't try to imitate an electromechanical strobe, but I don't think that actually matters with reeds because you're not turning a tuning knob in realtime.

Link to comment
Share on other sites

The problem with all the electronic wizardry and super advanced software is the instrument and how it is tuned, In an engineering sense, just because you can measure something, you don't have to. Just because you can measure to five or six decimal places, you don't need to. It is more important to measure with consistency and at least 20% , tighter precision (discrimination) than you need to actually work to. To be confident in working to a mm unit accuracy you need to be able to measure to 0.2 mm across the measured range, or tighter.

 

This thread set out to establish an agreed common tolerance for tuning concertinas, I put forward +/- 1,5 cents from nominal. Some non-technical types chose to demand perfection, which is in the eye of the beholder and does not exist. Some argued that it is a best efforts exercise, as close to nominal as you can get. Many manufacturers/ repairers chose not to disclose what they thought was acceptable, I assume because they were working to a slacker tolerance. I think that Generally +/- 1.5 cents was deemed acceptable.

 

If we are working to say 1/2 cent in tuning, then instrumentation need to discriminate to a unit decimal of a cent to 0.1 of a cent in its display, behind that the actual precision is probably 0.01 cents to guarantee an accurate 0.1 cent incremental read out. Frequency readings from a free reed do tend to be a bit irregular, so some form of smoothing or averaging function needs to be built in. In short, analogue read outs can give parallax errors which can cause some irregularity, and need a big (expensive) display. Strobe I never got on with, I found it not steady enough and extended working times. I prefer a straight digital output, giving a single level of decimal precision, one that is responsive but does not flicker. I also like to have the read out display change colour green between +/- 1,5 cents. Red if flatter than this, and blue if sharper.

 

Dave

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...