Tuning Advice

[RunClimbPlay]

I have a new (found in antique store) Giulietti LMMH accordion that I want to get in as good shape as I can. I play for my own enjoyment, so apart from wanting to make this accordion awesome, I don't have any professional/performance aspirations.

Tuning:
I loaded the Github Tuner (Snowball mic) and went through the low and one set of middle reeds. I played with gentle pressure, mostly letting the bellows open on the way out and trying to match the pressure on the way in. I hope this table makes sense to you - the numbers represent the cents above the note. I'd appreciate any advice on how to proceed from here.

In particular:
- Is this good enough? How close should then be before I consider tuning? (I have a tuning bellows I just made and reed blocks I can practice on, but I haven't tuned before)
- I took a fiberglass pen to some of the reeds that had rust on them. It took the rust off beautifully, but I also noticed a fair amount of dirt. Would cleaning the reeds change the tuning in any predictable way? (I would only use the pens on them, not remove them from the blocks)
- None of the registers seem to allow me to play the other middle or the high reeds. I guess these will have to be checked out of the instrument?
- With a few exceptions, they all see to be in the same ballpark. Or am I deluded?
- Anything else?

Thanks!

 

[Glug]

You should be able to select L,M,H reed banks easily. For the M+ bank put masking tape on the bottom of the M reed blocks - sticky side of the tape onto the block. Then when you play violin (M,M+) you will only get M+.

I'm going to suggest you record the actual frequency of the notes (pull + push) in a spread sheet.
The reason for that is you can calculate how far off the notes are and draw graphs etc. It's also possible to work out what the tuning is - is A4 442Hz or something else ?

How many cents off tune is too much is a tricky question. Some people here can hear small differences, other (me) can't.
So I mostly retune if the tremolo is annoying or varies too much, or some notes need replacing.

Here's my Scandalli LMMH tremolo after I retuned it to 12 cents at A4 (cleaned, rewaxed, revalved, replaced 6 reeds):

 

[Glug]

And for illustration here's the L reed tuning records:
I fill in the black columns under 'Measured' and the spreadsheet calculates the cents and colour codes them green/blue/red according to how off the tuning is.

Edit: From looking at the spreadsheet green is < 2 cents, blue is 2 - 3 cents, and red is > 3 cents.
I think I was aiming for 1 cent tuning accuracy. But there is a difference between a tuning target and what is OK before tuning.

 

[RunClimbPlay]

This is awe inspiring, and a bit scary. How long did this take you??

I like the tape idea and will do that. I can hear tuning differences but it isn't keeping me from enjoying playing (can't say the same for my family). The musette sounds pretty consistent, but there are some notes that 'warble' differently from the others ( l generally like the musette tuning - my Paolo Soprani is too wet for my tastes).

There is other work to do: the keys are loud, so I think putting in new pallet material (forgot the name for it) and key cushions are next.

Since I haven't tuned before, do you think I could start by tuning a few of the worst offending keys (I'd start by practicing on spare reed blocks)?

ps. Interesting: I just asked Chat GPT about the difference in cents between 440 and 442 - it's about 8 cents, which is where my reeds seem to be around.

 

[Glug]

I definitely wouldn't start learning to tune on your favourite accordion - I bought a cheap (£75) junker to learn on.

I've tuned about 3 accordions so I'm not an expert, but I did use the same spreadsheet.
NB. you can have it generate a map of which reeds to tune that exactly matches the layout on your reed blocks



I would strongly suggest you record the current tuning before you change anything.
That should let you identify which notes you think are a problem.
Also be aware that old/bad valves can throw off the tuning so don't dive in to scraping the reeds - you need to clean, rewax, and revalve first.
But you could do that to individual troublesome reeds.
But you need to know what to tune to - so work out what A4 is supposed to be first.

It's often a guess what the A4 was originally tuned to. You may be able to guess based on model but customers can specify tuning and it may have been retuned. My accordions (4) are 1935ish to 1958 so it was a guess:

Estimate 1:
Compare the average cents tuning error based on A = 442 or 443 (or 440 or 439) and guess original pitch from distribution of errors.

Estimate 2:
Assuming each note is correctly tuned (it isn't but ...) compute what A4 would have to be and plot the distribution of A4s

If you've got the current tuning in a spreadsheet you can generate those plots and see which you believe.

 

[RunClimbPlay]

Well, you're clearly a spreadsheet nerd (don't worry, I am too). Did you calculate the cents based on the Hz you measured, or was it a separate measurement?

Everything you're saying makes total sense. It makes sense to clean, rewax, and revolve before scraping. I will do that with my sacrificial block first for practice. And I like the idea that I could do this to individual reeds. The wax and valves see generally good, so if I don't have to do an overhaul, that would be nice -- although it may get to that if I get obsessed.

 

[Glug]

Myself 30 years developing software, but didn't really use spreadsheets. Now I'm retired I'm using them a lot - Open Office is my (free) poison.
Measured all the treble reeds open/close, estimated A4 from those measurements for M and L banks, decided 442 fitted best.

Individual dodgy notes - try 'pinging' the reed and think about changing the valve (or in some situations an adjacent valve).
'pinging' - lift the reed tongue using a feeler gauge or other instrument and let it ping back into place - shakes loose any crud block free travel.

 

[RunClimbPlay]

Yeah, I was already able to free 6-8 stuck valves by pinging with my newly-constructed bellows/tuner - very proud of myself.

Just to clarify: you measured the Hz of each reed and then calculated the cents from that?

 

[Glug]

Yep
1) Measure frequency of reeds
2) Assume A4 tuning (make it a cell on the spreadsheet so you can try different values)
3) calculate cents for that tuning

Then you've got the raw data to estimate current A4 tuning.
And if you get bored you could work out how wet the musette is tuned, and what curve was used. But that's a huge kettle of snails.

 

[Ventura]

most tuners can be set to 442 instead of 440
which makes it easier to see on a chart the deviation
high and low and average, and spot the worst reeds
more easily

when you then tune a reed on the workbench bellows,
you don't tune to the exact frequency because it will change
when the block is back in place, so you use your chart to
guide you to tune THE NUMBER OF CENTS down or up
for each reed so that when the block is back in place
you are really close to correct, and then only minor
fine tuning (hopefully) is needed, because for that you
have to keep popping the bellows on and off and on and off

yes accumulated dirt can affect tuning if it changes the
mass relative to the flex point of the reed.. a clean shiny reed
is what i like to work with.. a drag to get a 50 year old grubby reed
tuned just so then some of the old dirt starts to fly off as you
play the accordion over the next month changing the tuning again

 

[debra]

I have a new (found in antique store) Giulietti LMMH accordion that I want to get in as good shape as I can. I play for my own enjoyment, so apart from wanting to make this accordion awesome, I don't have any professional/performance aspirations.

Tuning:
...

The table with deviations seems to suggest this accordion's base tuning must be 442Hz. When you measure using 440Hz as reference frequency it is normal to see mostly deviations near +8 when the accordion is tuned 442Hz.
So I suggest you set your tuning app to 442Hz and then measure again. more of the values will be around zero (some positive some negative).

 

[RunClimbPlay]

And for illustration here's the L reed tuning records:
I fill in the black columns under 'Measured' and the spreadsheet calculates the cents and colour codes them green/blue/red according to how off the tuning is.

Edit: From looking at the spreadsheet green is < 2 cents, blue is 2 - 3 cents, and red is > 3 cents.
I think I was aiming for 1 cent tuning accuracy. But there is a difference between a tuning target and what is OK before tuning.

Sorry to bother you with this: what do the "adjustment: open/close" black numbers refer to?

(You wouldn't want to share the formulas you used for Hz->cents or tremolo, would you?)
EDIT: ((I've been reading up some of your earlier posts on tremolo - I recognize how complex it can get. I have a lot of learning to do!))

 

[Ventura]

honestly, the right musette is not by Math,
it is by your ear

i mean ok you make the M set perfect first,
because ALL the musette pitch adjustments are
made on the M- or M+ set

but using numbers as you go up in pitch, well the rumble
really doesn't stay the same or consistent.. you just
gotta put them in, listen, take em out make tiny pitch
adjustments put em in, listen some more, take em
out again make teensy adjustments

trust your ear until you get them so you can't help yourself
you just gotta play a whole damn song with 'em and
it sounds GOOD and you SMILE and all those hours were worth it

 

[Glug]

Sorry to bother you with this: what do the "adjustment: open/close" black numbers refer to?

(You wouldn't want to share the formulas you used for Hz->cents or tremolo, would you?)
EDIT: ((I've been reading up some of your earlier posts on tremolo - I recognize how complex it can get. I have a lot of learning to do!))

The 'adjustment' column is what changes I need to make, they're calculated from the measured value:
target frequency - calculated by exponential from A4 tuning
open reed adjustment <- reed that works when opening bellows = inside reed block
close reed adjustment <- reed that works when closing bellows= outside reed block

I agree the right musette is not by maths but for mere mortals maths can get to the right ballpark.
I'm still going to say it's worth recording what is the current tuning - for example you might want to put the musette back to how it originally was.

On that L tuning page
row 6
open cents = Q6 = 1200*LOG(P6/$A$2;2) - 100*($A6-5-2*12)
= 1200 * LOG2 (measured freq / A4 frequency) - 100*(key no -5-2*12)
= 1200 * LOG2 (87.67 / 442) - 100*(1-5-2*12)
= 1200 * -2.334 - 100 * -28
= 0.8 cents

But that looks confusing because it includes calculating the target freqency.
You will find it easier to add a calculated column for the target frequency and then reference that and use 100 cents = half a tone

Edit:
100 cents = half a tone
so 1200 cents = 1 octave = twice the frequency

so 1200 * log2 (F1 / F2) is the number of cents between F1 and F2

log2 = log base 2

 

[Ventura]

but for mere mortals maths can get to the right ballpark.

yes get them into the ballpark first, and your approach and
preparation sounds very nicely thought out

taking the time to prepare well saves so much time later

 

[RunClimbPlay]

I'm a long way away from worrying about the musette, but I will take both of your advice as a starting point!
Now I just have to become a better accordion player.

 

[RunClimbPlay]

You should be able to select L,M,H reed banks easily. For the M+ bank put masking tape on the bottom of the M reed blocks - sticky side of the tape onto the block. Then when you play violin (M,M+) you will only get M+.

I'm going to suggest you record the actual frequency of the notes (pull + push) in a spread sheet.
The reason for that is you can calculate how far off the notes are and draw graphs etc. It's also possible to work out what the tuning is - is A4 442Hz or something else ?

How many cents off tune is too much is a tricky question. Some people here can hear small differences, other (me) can't.
So I mostly retune if the tremolo is annoying or varies too much, or some notes need replacing.

Here's my Scandalli LMMH tremolo after I retuned it to 12 cents at A4 (cleaned, rewaxed, revalved, replaced 6 reeds):

Okay, looking more closely at this chart (great chart!), question: how did you calculate bps? And what does it mean to have a 'bps open' and 'bps close' --is that even possible?

At the moment (one more register to measure) I'm torn between spot tuning and waiting until I have the tools and skills to do a full clean/wax/valve/tune procedure like you did.

 

[Glug]

Okay, looking more closely at this chart (great chart!), question: how did you calculate bps? And what does it mean to have a 'bps open' and 'bps close' --is that even possible?

At the moment (one more register to measure) I'm torn between spot tuning and waiting until I have the tools and skills to do a full clean/wax/valve/tune procedure like you did.

The chart is tremolo tuning, so for LMMH it's the different between the two M reeds (either M and M+ or M and M-).

bps = beats per second = absolute difference in frequency
cents = difference in frequency expressed as cents difference from M reed to M+ or M-.

"bps open" is the bps on the reeds that are active on bellows open
"bps close" is the bps on the reeds that are active on bellows close
Ideally those are the same.
"bps avg" you can probably guess.

Historically / artistically it sounds better if the tremolo increases with note pitch.
The increase can be done in many ways, my 1958 Hohner Lucia seemed to be linear.
So in this case I've done 1.53 bps at F3 increasing linearly to 5.24 bps at A6.
That gives 2.95 bps at A4 = 11.7 cents. I was aiming for "12cents" tremolo - always quoted at A4.

I would suggest you will need some practise at tuning so pick one reed that's really bad/annoying and see if you can work out why it has a problem.
Could just need waxing/valving or might actually need tuning after you've tried those first.

 

[RunClimbPlay]

Do you have something that can measure BPS? I can't imagine that you can personally hear if something is 2.95 BPS!

 

[Ventura]

your brain can more or less measure BPM with some
experience

easiest way would be to use your home computer with
software like Band in a Box that lets you easily change
BPM as a setting with backgrounds you can hear and
play along with

there are probably apps for that too, but not nearly as much fun