Confused about frequency response in my first TG style falcate

[Munchkin]

I haven't posted for some time but follow developments with keen interest.

I've finally got round to making a TG falcate guitar (I was put off by epoxy and CF!) and followed the books and drawings faithfully, apart from a 25½" scale length. The specs are:
Spruce
English Walnut
Wenge fingerboard
Burr Elm and ebony rosette and marker
Crazy green celluloid inlay (never again...)
Polyx oil finish

Construction went well (if laboriously) and my tap tests gave me confidence that I would hit 90, 170, 214. However, now I have the guitar under tension the measurements have moved considerably and I've realised I don't understand what is going on at all.

All readings were made in playing position without the neck attached. Here are the plots in order.

Finished body with bridge glued in place, with bone pins 90, 175, 210

Back brace infilled (infill glued in place with superglue) 93, 179, 219

So.... the plan was - add about 600g of side mass to get T(1,2) down to 170 and then trim the back brace to lower T(1,3) to 214

Bolt on the neck and string up....
What? Under tension the readings were completely different to what I expected - 90, 161, 217. (Guitar sounded great BTW)

So I took out the side masses and got this plot - 90, 171 and a split 201 - 220

The upshot is - I just don't know what I'm doing..... Can anyone illuminate or clarify for me what is going on?

Martin

[johnparchem]

Stringing up the guitar drops the top resonance T(1,1)2 about 5 Hz and with coupling drops the air and back. Looking at your set of plots I noticed you made changes before stringing the guitar. You looked pretty close with your back low. I would not have made any changes until I had a plot with the guitar under string tension. Because of coupling when you stiffened the back you raise the top and the air resonance. Side mass mostly effects the T(1,1)2 top resonance.

[Dave M]

Martin without wanting to sound too critical - in the G&G book it is clear that although tap spectra .at the intermediate build stages are interesting, it is only once the build is complete with the strings up to tension that you can evaluate the instrument. It is at this point that you can decide if any modifications are required to move resonant frequencies around.

You say it sounds good so I would only worry if there were any wolf notes that you need to deal with. I have built several straight up G&G Steel strings and they just really work. One thing I notice is that the harmonics are very strong, this may be just be that they are pretty responsive guitars.

Anyway the instrument looks good, pics of the back and sides would be interesting... and I wonder how you found the Wenge. I read that it isn’t’ much fun to work. Let’s see that rosette that you won’t repeat!

Cheers Dave

[Munchkin]

Thanks for your comments John & Dave

I've read and re-read my grubby first edition so many times and have made 5 different (4 X-brace) guitars using the techniques. Somehow I missed the bit about reading the resonances under tension. It's certainly not in section 22.3. And I don't recall seeing anywhere that the top resonance drops by ~5Hz under tension....

But this is very reassuring. The infill should be easy enough to remove/adjust - mind you, it was a bugger to fabricate after the build - next guitar I'll make it at the same time as I brace the back.

I got the Wenge from a timber yard in Barking, London. It was perfectly quartered and was part of a bulk consignment of flooring. I was able to get quite a bit of splintered stock - which is of course the challenge with this species. It doesn't take kindly to sawing, scraping or routing! And it's almost all gone now, sadly.

The rosette was fun to make - I was inspired by Martin's (Kiwigeo) multiscale masterpiece. The green celluloid purfling is really striking but proved really hard to scrape level - which then showed loads of gaps that I had to fill with liquified celluloid, which was even more difficult to level!

The hardest part of the project was binding the back of the guitar with 9mm flamed maple binding, and getting the neck angle right. I had a crisis of confidence fitting the top and over compensated flattening the neck block. I haven't measured the bridge rotation yet but I think it is a bit more than 2°.

Interestingly, the intonation program gave me string lengths shorter than scale, which is why the saddle is narrower, and placed further forward than I would have liked.

Anyhow, here are some pics, and thanks again

Martin

[johnparchem]

Gorgeous guitar! The frequency spectrum looks good with a lot of peaks and the guitar came out very close to target before doing anything. How does it sound now? How is the intonation?

[Munchkin]

Thanks John

Intonation is very close to perfect - but I'm letting her settle down for a few more days before I refine the saddle and action. I'll update you then!

If anything the sound was best with the side masses attached, but she was a hefty handful to say the least. And the wide fingerboard and 16" radius is a bit of a challenge to play...
I'm hoping to get two commissions based on the performance of this guitar, so fingers crossed.

Martin

[johnparchem]

Besides lowering the top resonance (T(1,1)2) the side mass increases the size of the monopole, So having a 160 Hz guitar with a large monopole would be a different sound than you ended up with. I would take a 170 Hz guitar with no side mass over a 160 Hz with 600 grams of extra mass.

[kiwigeo]

The rosette was fun to make - I was inspired by Martin's (Kiwigeo) multiscale masterpiece. The green celluloid purfling is really striking but proved really hard to scrape level - which then showed loads of gaps that I had to fill with liquified celluloid, which was even more difficult to level!

Jason Kostal was the inspiration for my segmented "stained glass" rosette

[Dave M]

Thanks for putting the photos up. That Walnut is really nice - well matched. I have used American Walnut which is OK but much plainer. A great looking guitar, and a nicely shaped heel.

I was new to CF and epoxy but once into it it was fine, though pretty messy in the first go rounds!,

It is quite funny how design elements descend through a series of builders, re Martin.

Did you really do abalone purfling round the top. That must be a huge amount of work...

Cheers Dave

[Trevor Gore]

Nice guitar, Martin.

Regarding the changing frequency responses, your early responses without the neck mean you couldn't have had strings on, and you may not have had saddle and bridge pins in. Just adding the mass of the strings can make quite a difference which people often put down to loading the soundboard with string tension, when I usually find that the change can be accounted for by the added mass.

To properly identify what is going on you need to know which resonance each peak represents, which can be determined using Chladni patterns. With added side mass and possibly saddle/pins/string mass, the resonances will be redistributed and one may have emerged from beneath another.

The only sure way to find out is to get the Chladni gear out. It would be very interesting to know what is responsible for your peak at 201Hz. I have an idea, but it would be good to see it independently verified.

Your tap responses are very clean, by the way, so well done there. As many have a lot of trouble getting clean plots, would you like to share how you did that? TIA.

[johnparchem]

... Just adding the mass of the strings can make quite a difference which people often put down to loading the soundboard with string tension, when I usually find that the change can be accounted for by the added mass. ...

Interesting, I do test my steel string guitars with pins and saddle both before and after stringing the guitar and see the T(!,1)2 drop. But there is the added mass of the string balls as well as the strings. I also constantly see this with classical guitars. I have not tested no tension vs tension to see if the string mass is the cause. So that is worth looking into.

[Jim watts]

Nice guitar, Martin.
.......
Your tap responses are very clean, by the way, so well done there. As many have a lot of trouble getting clean plots, would you like to share how you did that? TIA.

Please!

[Munchkin]

I did struggle with setting up Visual Analyser at first but once I'd got the sample size and buffers right it works pretty well. So it must be my gear...

I use an AKG D5 vocal mike on a boom mic stand (playing in a band helps!). This is connected to the laptop PC via a Yamaha AG06 mini mixer, so I'm not relying on the sound card. The mixer connects via USB and the mic appears in the list of window sound devices. I hold the guitar on my lap and tap around the bridge with a home made hammer consisting of a plastic eraser mounted to a length of dowel. The gear itself is not high end and relatively cheap to buy.

I got very similar results using LSA. from Alain Lambert. But there is an interesting free Android app called Luthier Lab which has a heap of features including a spectrum analyser which uses the phones microphone and gives pretty accurate results. Here's a plot I took with one tap just now.

The biggest challenge of that was sharing the plot as it involved emailing a pdf to myself, printscreen and Microsoft Paint....

I've not set up to do Chladni analysis yet but I had a go with Luthier Lab (which has a tone generator) and a bluetooth speaker. I was able to get a main monopole but I don't think the speaker was powerful enough to excite the top at higher frequencies. LL let's you photograph the pattern together with the frequency and stores it in the project - pretty cool. I've attached the reports for your interest. I will pursue further and report back.

Thanks for all your kind comments.
Martin

As you can see from the LL plot the split t(1,1)3 appears to have been a ghost. If those readings are correct then the guitar is pretty much perfect with no side mass and a stiffened back brace. I will check later with VA. I'm fiddling around with the action at the moment to get the playability right before addressing anything else.

[Munchkin]

So.... Set up is pretty much there. Intonation is a little sharp, which is disappointing - I'll get on to that later.

Here is the tap test on VA - 90, 173, 209

I decided to add 230g side mass to shift T(1,1)2 and got this result - 90, 169, 208

I've already infilled the back brace - any suggestions as to how I can lift T(1,1)3 to 214Hz? Perhaps a more substantial, taller infill, perhaps extending the infill wider on the brace?

TIA
Martin

[Dave M]

Martin thanks for putting this stuff up. I don't think any of us find it easy so info on how other people do it is always useful.

Just for information, to excite Chladni patterns I use an old domestic stereo amplifier (doesn't everyone have one in a cupboard somewhere?) a cheap car speaker which I built into a simple MDF box, and VA to provide a frequency signal. It works fine. The 4 inch or so speaker is about the right size to excite the smaller areas that are resonating.

Even then it is not always straightforward to sort out the various resonances.

Cheers Dave

[Munchkin]

It has become apparent that I need to set up a Chladni testing environment....

I found Trevor's detailed description of his DIY Sig gen/amp unit but wondered if I could make a unit based on the Luthier Lab android app I mentioned above.

To this end I've just ordered a Bluetooth Amplifier board rated 2X50W from Amazon for £13.99 and a pair of JVC 4" speakers 210W peak power 45 - 20,000Hz for £15.50. The plan is to build a MDF box and to power the unit with an old laptop PSU with a spare speaker in case I blow one. So that is A$53.50 in real money....

If the project goes well I will start another thread with details.

[Munchkin]

I'm getting frustrated now. In the quiet comfort of my studio I set up VA to do another tap test.... and got 90, 165, 210

Which hints that I didn't need to add 230g side mass.

I may be getting clean tap spectra, but the inconsistency is annoying. I was playing around with Luthier Lab and got this spectrum with a single tap, which is what motivated me to check....

I honestly have no idea why the tap test in the workshop is different apart from the ambience.

Has anybody else experienced this variability?
Martin

[Dave M]

Is it some time between the measurements? Could the guitar be settling in? It is the top that is changing, could the string tension be changing the top resonance?

[Dave M]

To check this you would have go back to the previous test conditions.

[Munchkin]

The guitar has been settling in for a few days - and the top is visibly less arched. I guess this means my bracing was not stiff enough...

Sounds really good though!

I intend to stiffen the back when I get back to the shop next week - and to work on my Chladni testing gear. I don't intend to take the top off to reinforce the bracing, however.

Thanks for your feedback Dave

[mike_mahar]

Hi, I'm one of the developers of Luthier Lab and I just discovered this forum and that some of you are using Luthier Lab. From Munchkin's post, it appears that he has a pretty good handle on how to use it. If you have any suggestions on improvements, please let me know.

For getting Chladni patterns, I use the method that Al Carruth taught me. I take a small 4" speaker with no cabinet and hold it close to the top or back. I move the speaker around the sound board looking for some of the glitter to dance. I raise or lower the frequency to get the maximum activity in the glitter at the same time that I am moving the speaker around. Sometimes moving the speaker a centimeter or two can make a huge difference. I support the soundboard at the edges with small foam pyramids and move then around the perimeter so that they support the sound board at the node lines of the current mode. You can get quite clean patterns that way.

You can take a picture of the pattern with the phone's camera in Luthier Lab and the image will go into the current album in the Chladni tool. I like to create a new album for the top and back both before and after shaping the braces.