Jump to content

Recommended Posts

Recent discussions on leathers and leakage prompted me to think about doing some tests. I was particularly interested in questions like why we use chamois leather as a gasket when we know it isn't airtight. Does clamping it between two flat surfaces solve all its problems?

 

And I thought I'd make it more fun for me and you by having you looking over my shoulder, putting forward helpful hints, coming up with bright ideas, making cruel and unnecessary jibes ....

 

I thought I'd start by making up a general purpose test rig, which I could also use to test bellows, reed pans etc for leakage. One piece was like the end cap of a concertina, except no button holes or vents, other than a hole in the middle to attach to the leakage detector. The other piece was just a flat plate of the same concertina-like dimensions. Put them together and you have a thin, airtight, empty, hexagonal box. Both were made from 12mm (1/2") MDF, sometimes called craftwood. Smooth, dense, flat - just the thing.

 

So I make up my two pieces and clamp them together with 6 big spring clamps where the 6 bolts of a concertina end would normally go. First test is to confirm that they form an air-tight cavity. Plug in the Magnehelic leakage detector, et voila - hmmm, leaks like a sieve. Indeed, the Magnehelic dropped from its setup point of 8 when open to atmosphere, to about 7.8. My test rig was essentially transparent to it! A failure of stupendous proportions! What an excellent start to our journey!

 

Aha, I thought, clamping isn't enough - air is getting out through the gap between the two pieces. I can prove that by wrapping it around in duct tape. Hmmm, absolutely no difference.

 

About now the true horror is sneaking in. MDF might be dense, smooth and flat - but it ain't airtight! Hey, this science stuff is cool - we've learned something already!

 

But can I prove it, and what am I going to use for a test chamber? The answer is clearly: - let's render it airtight by doping it. So, I've lacquered all surfaces of both sections with a layer of PVA glue. That's going to take some time to dry, and it's getting late down here. We'll pick up the story tomorrow.

 

Terry

Edited by Terry McGee
Link to comment
Share on other sites

  • Replies 65
  • Created
  • Last Reply

Top Posters In This Topic

Good on you, Terry.

 

It's 1.45am so I'm trying to get to bed (I live a fascinating life)
Very interesting what you're doing. I'm very happy to confer and watch.
If possible get into the habit of making a few seconds of video at each stage, for possible use later.
Personally I don't mind quite so much about leakage because I use a strap from each end piece, going under my thighs, giving me about 30% more air to use (because I can assist the bellows pullout by using my legs, which are much more powerful than my hands are. It also firmly anchors the concertina on my legs for easier playing.
It could be good, to save time, to yak by either Skype or Google Hangout (you'd need a Google+ account for that, no cost, and better than Skype.
We could even show each other, live, what we were doing.
Cheers,
Bruce
Link to comment
Share on other sites

Yawn, it's morning! And the glue seems nice and dry. So, assemble the pieces and apply the leak detector and it's a.... 6. Hmmm, I was hoping for better than that; indeed we're going to need better than that to put the focus on the materials under test, rather than the test appliance. Now, the glue seems to have soaked in pretty well, but, rubbing the mating surfaces together doesn't feel perfectly smooth. I wonder if a slightly lumpy surface is the cause of our remaining leakage? I put a sheet of 320 grit abrasive paper on the top of a granite surface block (flat to within 2.2micrometers) and flatten the two mating surfaces. Back to the Magnehelic and, yes, a 1. I think we can live with that, by taking a measurement before and after, and subtracting the mean from the result with the material under test.

 

I do have to consider the possibility that the Magnehelic is just far too damn sensitive to work sensibly with concertinas. It's designed to detect subtle problems with woodwinds, and in particular, flute, where the slightest leakage is enough to weaken the response dreadfully. Biceps are so much stronger than lungs, and steel reeds so much stronger than the "air-reed" we visualise when struggling to understand how flutes can make sound with no moving parts. We may have to find ways to dumb the Magnehelic down. But that will become more obvious as we go.

 

OK, now, breakfast!

 

Terry

Link to comment
Share on other sites

 

Good on you, Terry.

 

It's 1.45am so I'm trying to get to bed (I live a fascinating life)
Very interesting what you're doing. I'm very happy to confer and watch.
If possible get into the habit of making a few seconds of video at each stage, for possible use later.
Personally I don't mind quite so much about leakage because I use a strap from each end piece, going under my thighs, giving me about 30% more air to use (because I can assist the bellows pullout by using my legs, which are much more powerful than my hands are. It also firmly anchors the concertina on my legs for easier playing.
It could be good, to save time, to yak by either Skype or Google Hangout (you'd need a Google+ account for that, no cost, and better than Skype.
We could even show each other, live, what we were doing.
Cheers,
Bruce

 

 

Morning, Bruce frum acruss the dutch!

 

Video's probably a good idea, but it would require cleaning up my workshop, and that might set science back years!

 

And I must admit I've never even heard of Google Hangout, which sounds excruciatingly cool and trendy! But I am set up for Skype. I don't leave it on all the time - it seems to attract enterprising Chinese students, and the lovely Svetlana from Ukraine, who feels we are soul mates. So, zap me an email first, so I can launch it. I'm "mcgee-flutes".

 

Terry

Link to comment
Share on other sites

Hmmm, thinking about the flow rates appropriate to concertina leakage testing requires us to consider, in human terms, what's a fair amount of leakage. If you were to fill your bellows, then press the ends together at about the pressure you use in normal playing but not pressing any buttons, how long does it take for the bellows to close? How long do you think it should take to close?

 

Terry

Link to comment
Share on other sites

Yawn, it's morning! And the glue seems nice and dry. So, assemble the pieces and apply the leak detector and it's a.... 6. Hmmm, I was hoping for better than that; indeed we're going to need better than that to put the focus on the materials under test, rather than the test appliance. Now, the glue seems to have soaked in pretty well, but, rubbing the mating surfaces together doesn't feel perfectly smooth. I wonder if a slightly lumpy surface is the cause of our remaining leakage? I put a sheet of 320 grit abrasive paper on the top of a granite surface block (flat to within 2.2micrometers) and flatten the two mating surfaces. Back to the Magnehelic and, yes, a 1.

 

Terry:

 

I do not believe that 1/2" MDF passes air.

 

From the description of your test rig, it is not clear to me if you have gaskets of any sort between the bellows and your fake ends?

 

If not, then I would suggest that might be the problem.

 

If you do have gaskets (maybe aged chamois leather?) then they might be the problem. Maybe you could make up some new gaskets? In NA you can buy chamois leather from auto parts stores, you might have to pick them over a bit to find some nice pieces, but the chamois leather that I bought in a car parts store seemed OK..

 

Don.

Link to comment
Share on other sites

Heh heh, seems impossible doesn't it! But I can prove it.

 

But first I should make clear that we are talking about very small airflows here. Full scale airflow on the leakage detector is about 1 Standard Cubic Foot per Hour. At a rough calculation, that level of leakage implies you'd exhaust a full bellows in about 20 minutes. As I've mentioned I think my flute leakage detector is a bit too sensitive, but, in this initial phase, that's good. Ideally, you'd like all your test rigs and instrumentation to be at least "an order of magnitude" (10 times) better than what you're trying to measure, so that your accuracy will be at least within 10%, even without making compensations. With compensations, you can get to 1% accuracy, which is really nice.

 

And I haven't got to putting a bellows between the ends just yet (but I will!). Up till now, I've just been clamping them together, to prove (or disprove) that the rig is airtight.

 

And no, there are no gaskets as yet. "Why?" would be a fair question! Because I want to test possible gasket materials, it's best if I can get away without any on the rig itself. That way, we know that any leakage has to be through the gasket material. So, I'm clamping MDF to MDF, and hoping to get away with it.

 

Now, I claimed above that I can prove the leakage I've been seeing is through the MDF and not between it. Firstly, I can point to the results back in the first post. Leaked like a sieve when first made, didn't help when I wrapped around the joint in tape, was improved when I doped it all with glue, and came finally good when I flattened the glue lacquer back to seal the joints properly. Pretty good proof, but wait, there's more!

 

Having satisfied myself my rig is now almost airtight, I tried replacing the doped plain end (the end without the cavity and the air connection) with a fresh piece of MDF. The leakage read 6 (8 when open, 1 with my full rig). I tried flattening the new piece on the granite surface plate, but couldn't make any improvement. So the air seems to be seeping through the undoped MDF.

 

Now why only a 6, which, while not good enough, is better than I'd found with the full rig, pre doping? I need to explain how the rig is constructed. I had cut three hexagons of MDF. One is the plain end, the other two make up the measuring end. One of these is left plain, other than having the measuring connector through the middle. The other one has the middle cut out, leaving about 10mm all the way around. Those two are glued together, thus forming a cavity.

 

That last piece offers "sidegrain" access to the air inside. I suspect the leakage is greater through the "sidegrain" of the MDF. Indeed, I suspect the stuff is built up of deposited layers, possibly with spaces between them in places. You sense that when drilling especially with a spade or forstner bit - the bit suddenly breaks through a layer, pauses, then repeats.

 

I also tried sealing my measuring end on a sheet of thick plate glass and got good results. Interestingly, even with that tiny airflow, the piece "hovers" and can be slid around the glass effortlessly. Great stuff!

 

Hope all that makes sense, Don, and feel free to challenge any of my assumptions. That's how we get somewhere!

 

So, we're getting close to being able to actually use the rig to measure things....

 

Terry

Link to comment
Share on other sites

Hmmm, I'm suffering a work attack, so this might have to slow down for a bit until the fit passes. But I did have time for a couple of quick experiments...

 

I put a strip of masking tape on the backing plate so that it passes from the inner cavity to the outside world, then clamped the rig up under the 6 big spring clamps. Leakage doubled from 1 to 2. I doubt if the air is passing through the tape, I imagine its seeping through the gaps on each side of the tape.

 

I don't have any chamois leather on hand but in my rag box I found some similar looking cloth - smooth on one side, furry on the other, probably taken from a windcheater. Quite transparent to airflow through it (whereas I imagine chamois to be somewhat more resistant). Clamped in the rig, the reading dropped from 8 (full open) to about 7. So definitely not attractive as a gasket. I need to find some chamois....

 

I clamped one end of my Anglo to the measuring end of the rig. I'd previously tested each hole with the leakage detector and found them pretty good (average 1.4). But 18 holes mount up, and the combined leakage came up to about 7.5.

 

All of this tells me that my rig is good, but my meter too sensitive. So, next step is probably finding a way to modify the meter to be about 10 times less sensitive.

 

At this point it would be very good to get some views on the questions I posed above:

  1. If you were to fill your bellows, then press the ends together at about the pressure you use in normal playing but not pressing any buttons, how long does it take for the bellows to close?
  2. How long do you think it should take to close?

 

Up-oh, back to the flute mines....

 

Terry

Link to comment
Share on other sites

Terry,

you have asked two "How long's a piece of string" questions there. For instance, I have concertinas of various sizes , with Bellows of 5,6,7and 8 folds... then the effective length of the various Bellows are different depending on the depth of the folds. Then there is Condition... some of my squeezers have New Bellows whilst others are original 90,100, 110 years old. Then there is all the end condition (pads, cracks, springs) ... and then how hard does a person think they press when playing.... you'll get a huge variety of answers to your first question.

"

Second question has a simple answer . " As long as possible"..... perhaps?

Link to comment
Share on other sites

OK, OK, point taken. Let's narrow it down a bit.

 

It's a 6 fold concertina of classic proportions, in good to very good condition. Possibly even new. (Someone must have one of these surely!) Open the bellows and play a note at typical volume, noting the force involved. Now open the bellows fully, but don't play a note, just press with the same pressure. Count 1, Mississippi, 2, Mississippi, and so on until the bellows stop moving. Report the number of Mississippis encountered.

 

Email me at terry@mcgee-flutes.com if you prefer!

 

(You might wonder why I don't do this myself. I have two concertinas, both Anglos, one period, one recent. But I don't have the experience to know if they are typical.)

 

[Turns to audience, looks cute, and sings:] I need someone, older and wiser....

 

Terry

Link to comment
Share on other sites

OK .On that basis my most airtight concertina which is 116 years old but has a new-ish bellows ( 7 years old by Wakker) of your stated size (6 1/4" Hex x 6 folds) takes 110 seconds to shut from fully open.

 

For this test I played an A (440hz) then turned the concertina to a vertical position and let gravity play the A... (gravity was louder than my 'normal pressure') so I let gravity close the bellows with out the note and that gave me the 110 secs.

Edited by Geoff Wooff
Link to comment
Share on other sites

Thanks Alex. Let's "normalise" your 5-fold up to 6 fold by multiplying 23 seconds by 6/5 = 28 seconds. And thanks for the descriptor "slightly leaky" - that's a useful reference point.

 

Anyone else have a really leaky, leaky or slightly leaky concertina to contribute?

 

Remarkable figures (110 seconds) from your "most airtight" instrument, Geoff. Again an excellent reference point. I'd be inclined to call that "near perfect".

 

I do like the idea of letting gravity take responsibility for the pressure, and the inverted mode gets around some of the problems of the drop test. Some concertinas have issues as you come up to closing, for example not wanting to close fully. If that's suspected, it would be fair to measure the time to close to halfway, and then double it.

 

And very good figures (77 seconds) from your Dippered Jeffries, ex-sqzbxr. Shall we describe that as "tight" or even "very tight"?

 

A few more figures would be great. Squeeze people, squeeze! In the meantime, I'll do some calculations on volume....

 

Terry

Edited by Terry McGee
Link to comment
Share on other sites

But while on the topic of closing times, how about a table like this for something to aim at?

 

Unplayable:

Really leaky:

Leaky:

Slightly leaky: 28 seconds

Tight:

Very tight: 77 seconds

Near perfect: 110 seconds

 

"Perfect" is, by definition, an infinite closing time and probably unobtainable.

 

Any proposed changes to the descriptors above?

 

Terry

Link to comment
Share on other sites

Now, to get our leakage rate, we need closing time figures as we've been discussion, but we also need to estimate the capacity of our concertinas. See what you think of this as a way to do that.

 

(I'll work it out for a hexagon, but the approach should work for any other shape. And we're not really concerned with total volume, but the change in volume as we open and close the bellows fully.)

 

My Lachenal/Crabb has bellows trapezoids whose long sides are 85mm, and short sides (measuring to the middle of the folds) of 65mm. We can thus say the hexagon's average side length is 75mm.

 

The surface area of a hexagon (3√3 * s2)/ 2, where S is the side length, 75mm. For our averaged side length, our hexagon has an area of 14614 sq mm.

 

The concertina measures 380mm end to end when fully extended, and 130 when fully compressed. The difference in length is thus 380-130 = 250mm of bellows extension.

 

The volume of the concertina is the product of that extension (250mm) and the averaged surface area (14614 sq mm), or 3653645 cubic mm, or 3.6 litres. Or nearly a wine cask, as we'd say in Australia.

 

My Simpson has trapezoid sides 83 and 60, so an average of 71.5mm. That makes the area 13282mm2. The bellows extension is 170mm, making the volume 2.26 Litres. Just over half a wine cask.

 

Happy with the method? (We're only looking for an approximation.)

 

Given that disparity (3.6 vs 2.3 Litres) it would be good to get a few more dimensions to average out. I'm happy to do the maths if you give me:

 

Concertina details:

Number of sides:

Length of long side of trapezoid:

Length of short side (to centres of folds):

Length of instrument fully open:

Length of instrument fully closed:

 

Metric or Imperial measurements OK.

 

Historical note. When wine casks were first introduced they held a gallon (4.54L or 6 bottles), but that got rounded down to 4L. For some technical reason I don't fully understand, the price didn't get rounded down. Grrrr.

 

Terry

Link to comment
Share on other sites

Terry,

 

A couple of variables;

1) concertina weights vary such that there could be a .25kg difference in the weight of your "gravity".

2) spring pressure. Set it to, say, 120gms, and come back tomorrow.

 

Historical note. When wine casks were first introduced they held a gallon (4.54L or 6 bottles), but that got rounded down to 4L. For some technical reason I don't fully understand, the price didn't get rounded down. Grrrr.

It was a tax on poor taste in wine..!

 

Cheers

 

Chris

Link to comment
Share on other sites

Agreed on the variation in weight, Chris. Fortunately I'm only looking for a ballpark indication, and the difference is probably no worse than just guessing the right pressure!

 

And I guess you're expecting to find that leakage through the pads probably exceeds that through the gasket. I'll be pleased to put that to the test once we get the test protocols worked out.

 

Now, speaking of button pressure, do we have any established view as to what constitutes a reasonable pressure?

 

And on the cheap wine front, I can only say Hurrumph! I'm but a poor flute maker, and cannot hope to enjoy the luxuries that must be commonplace at the tables of the fabulously rich concertina makers that loll idly round on this list....

 

(Heh heh)

 

Terry

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...