Project "Groove Scribe"- A DYI stereo 45/45 head.

[Bahndahn]

https://vimeo.com/131547513

new test. Using MyShank black blank. I am very impressed with the quality of the cut.

Congratulations!! This is great!

I would love to hear the audio directly from the mixer if you get around to it.

[EpicenterBryan]

I used the wire from an old speaker, then I do not know the gauge. I used Mu metal to shield, but need higher magnetic isolation.

Hey Ciuens,
great stuff on that test too!
I think I'm on to the issue as well after playing the the old drivers and Caruso Massimo. I hope to have some solid data in a few days. I had a nice chat with MarkRob and he had some great ideas. I had a chance to try some things that should have worked and didn't. I think I know why...

If you have only one layer of MuMetal, I don't think it is enough and if it's easy to add layers, I think you will see a big difference. I found more layers is better, and each layer decreases the field straight on axis. But it kind of works like a heat shield. It will block some straight forward, but it really diverts it to the edge. More layers is like applying more sun screen and does a better job blocking straight on, but the MuMetal really doesn't absorb it, it redirects it as I have found out. That is where I think I went wrong with trying to apply it after the fact on Groove scribe. It did help quite a bit in reducing the strength when you use a screwdriver to feel the difference, but it seamed to concentrate it at the edge of the shield and I didn't make the connection until tonight. In my situation, an edge is right below the phono cartridge which can still pick it up...More than I thought. More later...

In your case, your coils are in the center and you do not have a direct magnetic couple from the driver to the coil with a steel rod like I do. So I think you can get away with just increasing the number of MuMetal discs you have.

I ran into the same symptoms using the old drivers in the Caruso Massimo update video (see other thread), and after playing with a few other things that failed to impact the big phase issue, I disconnected the link between the drivers, and there was the same crazy phase shift and a bunch of signal! It's clear that comes from magnetic coupling...

I plan to try my MuMetal "heat shield like" theory out using the other drivers. What I plan to do is make a "cup" of MuMetal to go between the driver and the other one that's being used as a pickup. I'll face the cup away from the pickup. If it works, I'll try to make a 5 layer cup and compare results. The cup idea is not only to reduce the on-axis coupling, but also to get the concentrated magnetic field away from the feedback device. If you imagine a doughnut of magnetic field on the edge of a disc of MuMetal, I will move the doughnut so the minimum is pointing toward the pickup.

toroidalfield01.jpg

So pretend like this picture shows what the field looks like at the edge of a MuMetal disc. Pretend like the disc gets changed into a cup... Then the magnet doughnut starts at the edge of the cup. Then shouldn't the doughnut move to the top of the cup, and be much less strong at the bottom (where your coils are, and where my phono cartridge is)?

Think of it like this (horseshoe magnet, and spin it in a circle):

horseshoe_246p_256c_b_C6D01184-FF37-3286-CEDB044E99B3ED74.gif

Hey MarkRob what do you think?

Bryan

[EpicenterBryan]

Hey Ciuens,
It occurred to me that you may have an issue with your magnet holder / mount not being stout enough. If your magnet vibrates, it causes a signal just like when your links move. Take a good look at that. Your shielding may be OK.

You may have both a moving coil and a moving magnet transducer at the same time. Should be an easy fix.

Bryan

[EpicenterBryan]

A quick update:

I did a bunch of reading about MuMetal and shielding applications and discovered this cup / cylinder idea is a common solution. So common, you can buy them in cylinder or box form. There is even a YouTube video showing their machine in operation.
You can get these from: http://magnetic-shield.com

deep-drawn-mumetal-cans.JPG

It turns out there is a size that fits these drivers and a size for the Groove Scribe drivers. One size is in stock, one is a 6 week lead time. They are $50 each ($45 each in quantity of 2). Both the box and cylinder version have a radius on the corners which is the ideal path for the field to travel...

So before I spent the cash, I decided to try the low tech method. Here I made up a duplicate test jig but with the feedback sensor inside the can, and the driver outside. I love olives and ate them.

IMG_3783.JPG

I did a quick check tonight with 1 watt and then disconnected the link between the drivers and the feedback sensor with the audio still on. Damn if there wasn't still a bunch of signal.

Then something occurred to me...
I turned off the audio and spoke into the can.

It turned out I re-invented one of my favorite childhood toys!
My "feedback sensor" which was a spare driver, still had a cone, surround and one of my plastic threaded link discs attached. So it was acting like a microphone as well as movement sensor.

UNTITLED.jpg

So here is my minimalist sensor mod for the night. The glue needs to dry but I will start testing tomorrow.
There is not much left for sound to push against and what's left is the bare minimum - so if this doesn't do the trick...

IMG_3786.JPG

It's the simple stuff I always forget about. Duh.

Bryan

[EpicenterBryan]

So here is my minimalist sensor mod for the night. The glue needs to dry but I will start testing tomorrow. There is not much left for sound to push against and what's left is the bare minimum - so if this doesn't do the trick...

A quick check showed a marked improvement relating to unwanted acoustic coupling. But I think I'll need to punch holes in the rubber spider to get near the minimum pickup possible.

Going back to the other drivers for Groove Scribe, I have two coils in the works. I did a test using contact cement earlier in that day that failed.

I contacted a company that sells speaker rebuild kits and parts including glues for various speaker applications and asked them what the ideal glue they sell would fit the application. They said "I would go to the hardware store, and buy a 2 part epoxy that dries NON-flexible.".

Sorry, I've used generic epoxy on stainless and it doesn't stick.

MarkRob and I have both used a product called JB Weld that sticks like crazy and although Mark recommended it, I really didn't want to use it for several reasons, but I gave up and started the application tonight.

I love the stuff. I hate the stuff. It really works but drives me insane most of the time. The stuff bonds like crazy. There is no chance to ever take it apart for any reason. The other issue is it takes literally 24 hours to fully cure. In between, it remains semi-fluid and flows where it wants to hours after you think it is set up. It also has iron in it so it wants to flow to a magnetic source.

Tonight I decided to take advantage of that to direct it where I wanted while it sets up. In the shot below, you see some hard drive magnets. They are there to make sure it flows and stays at the union of the aluminum voice coil and the stainless spring. Oh, and that fixture is at about 30 degrees...

IMG_3788.JPG

It may take a day or two to have all 4 quadrants fully attached so hopefully this weekend I'll have some results.

I hope someone out there has a better way to re-attach these coils.... This is silly.

B

[Ciuens]

Bryan, I'm thinking of using Mu Metal removed from old hard disks.
Using 2 will get the necessary magnetic insulation.
More and more tests, hahahaha

Example
https://www.youtube.com/watch?v=hM1hJL3mYJk

To reduce the mass, I completely removed the metal coil. 3g weight

Another test I'm doing:

Decrease the springs

[markrob]

Hi,

I did some experimenting this weekend. I used Bryan's method for removing the coil assembly from the spider and CNC cut a new one out of PCB material (G10 glass epoxy). My CAD package predicted (correctly) that this would weight just about the same as the supplied metal spider minus the extra stiffener disk. That works out to about 2.8g. It is much thicker at .062" and has a copper clad surface which is nice for sensing using my Kaman Eddy current sensor (or a home brew version). I attached the coil to the new spider using Duco cement. Porbably not the best long term solution, but it is easy to apply and remove if you need to play.

BCT-2 Mod Top.JPG

BCT-2 Mod Bot.JPG

Below is the new displacement response:

BCT-2 Modified.jpg

With the mass reduced from about 6.7g to 4.8g and the stiffness increased, the resonance is now in the 1Khz range. There is still a secondary resonance at about 7 Khz. I think that can be damped with some adjustments. I suspect that the drive levels are quite a bit higher than the stock head. I'll run some tests to see how badly reduced it is.

Mark

[Ciuens]

Mark, in professional cutting heads there this fall after resonance?

[markrob]

Mark, in professional cutting heads there this fall after resonance?

Hi,

Yes. Its pretty much dictated by the physical laws of the head construction. The pro heads will have lower moving mass, better damped or no secondary resonances, better thermal characteristics, etc. But at the end of the day, this is the open loop response we have to deal with. Keep in mind that the plot I show is displacement vs. frequency. If you are using a moving coil feedback element, you would be working with a velocity vs. frequency relationship. It is easy to convert between the two (velocity is the first derivative of displacement). The differentiated (velocity response) will rise at +6db/oct to the system resonance and fall at -6db/oct on the other side of resonance. This page from an Ortophon manual might make this clearer.

IMG_0745.jpg

This open loop response is why you need such a high power amplifier to drive the head of this type even though you would burn it up in seconds if you fed it anything close to the power amp rating. When you consider the high frequency boost in the RIAA curve and the fact that the head is down about -20db at 10 Khz vs. the 1Khz response, you can see how much power is required to make the stylus move at these frequencies. Fortunately, normal program material does not contain much energy in this region. But you still need large reserve power to handle short term peaks.

Mark

[markrob]

Hi again,

I did the test to compare the modified vs stock BCT-2 driver response. Here is a plot of the results

BCT-2 Stock vs Mod.jpg

The green plot is the stock driver and the yellow is the driver with the G10 spider. Purple is the noise floor of my measurement system. Both were driven with the same power level. You can see that the modified driver requires about 6db more power (4X!) to achieve the same excursion at low frequencies (in the compliance controlled region). Its hard to draw conclusions at the high frequencies since I'm getting close to the noise floor, but it looks like the stock head does a bit better here as well. So, increasing the stiffness this much (along with the reduction in mass) to attain the higher resonance is not a good way to go. I'm going to play with a plastic spider, which I think will weigh in at a gram or so (my G10 spider is at about 2.6g right now). This coupled with a lower stiffness might be a better solution. I like the truncated shape Ciuens proposed.

Mark

[EpicenterBryan]

Bryan, I'm thinking of using Mu Metal removed from old hard disks.

Great idea. I have been saving magnets from hard drives for years, and never though about saving the metal backing. Duh! Mu-Metal!
So I tore down some drives this weekend and built what I call a "Ciuens Shield". More on that later...
And about your spring idea, I'm glad you are trying those. Now that Mark has tried something like the original shape in a glass epoxy material (PCB blank) and you and I have tried them in various thickness of plastic with similar results it's time to try a different shape!

To Quote Albert Einstein:
"Insanity: doing the same thing over and over again and expecting different results."

I did some experimenting this weekend.... That works out to about 2.8g. It is much thicker at .062" and has a copper clad surface which is nice for sensing using my Kaman Eddy current sensor...

Another great idea there Mark. It's so cool that your sensor works on the copper surface too. So you have something like a CNC PCB machine?
I have an 8x10 sheet of double sided 0.058 that I could probably etch one side off and send to you if that helps. Not sure what the final thickness would be but you are welcome to it...

It is easy to convert between the two (velocity is the first derivative of displacement). The differentiated (velocity response) will rise at +6db/oct to the system resonance and fall at -6db/oct on the other side of resonance.

This one is going to take some time to get my head around. It's been many years for me...
So let me get this straight... if we had the expected rise and fall of 6db/oct the displacement chart shown should be a straight line, and the velocity chart would be a perfect triangle?

So what I'm getting from this chart is the LF rise is very close to 6db/oct which results in a straight line as the rise over run =1 as frequency increases. Then it increases to a higher rate of change heading into the resonance, where rise over run is greater than 1. And has a higher than ideal fall off right after the resonance where rise over run becomes negative. But the fall off then increases (rate of change in the negative direction) to a much higher rate over time going into the second resonance... It's hard to visualize this.

Can you confirm from the chart for simple math people like me ... flat line = constant slope/rate of change of velocity. Rise in displacement indicates increasing positive rate of change in velocity(more than +6db/oct). Decreasing displacement indicates decreasing rate of change of velocity(less than -6db/oct).

What does the displacement graph look like if the frequency response is exactly a triangular rising to and falling from a resonance, or like with a standard 3db point curve at the top of resonance?

Bryan

[EpicenterBryan]

So, increasing the stiffness this much (along with the reduction in mass) to attain the higher resonance is not a good way to go. I'm going to play with a plastic spider, which I think will weigh in at a gram or so (my G10 spider is at about 2.6g right now). This coupled with a lower stiffness might be a better solution. I like the truncated shape Ciuens proposed.

Interesting stuff Mark. Especially seeing both charts overlapped. And the power discussion is interesting too. Wow, 4x!

I keep thinking about the 4 identical springs segments. What say we try to come up with a spring that has 4 different effective lengths. Perhaps with different widths or thicknesses so they have the same flex but not the same resonance. Think of it like a suspension bridge that is designed to break up the resonance of marching solders. Could that be part of our problems?

Or take this example of an intentionally poor bridge design and see the many possible resonances from the same structure:

Bryan

[markrob]

Hi Bryan,

On the subject of velocity vs. displacement response, maybe this will clear things up. If you check out a post I made some time ago, you'll see I plotted 2 curves. The first shows 2 sine wave of the same amplitude (think stylus excursion), but with one at twice the frequency of the second.

http://lathetrolls.com/viewtopic.php?f=22&t=4046&p=33209&hilit=constant+velocity#p33209

You will notice that the slope or rate of change (think stylus velocity) of the higher frequency sine wave is twice the slope of the lower frequency at the zero crossing. This implies that as you double the frequency (1 octave) of a sine wave but keep the amplitude the same, the velocity doubles. So we have a response that rises at +6db/oct.

Below resonance, the head has a response that maintains a constant amplitude, so if you if you decide to plot the velocity response in this region, it must rise at +6db/oct. Once you get on the other side of resonance, the amplitude response falls off at -12db/oct. This is because a spring mass system acts like a second order filter at high frequencies. Since we have established that the velocity response has a +6db/oct response vs frequency when you convert from excursion to velocity, the -12db/oct response is reduced to -6db/oct (+6db/oct + -12db/oct = -6db/oct). This results in that triangular response that you see in the Ortophon manual. Your are viewing the exact same response from a different perspective.

I hope this didn't add more confusion.

Mark

[markrob]

Hi Again,

To answer you other questions. Yes I have a small CNC lathe/mill that I used to make the copper clad PCB based spider. I also made a plastic version with the truncated springs, but the plastic I used proved to be difficult to glue to the voice coil. Even JB weld failed. Not sure what material it was. Just something I had laying around. The exercise killed the BCT-2 driver ( I have 1 more to play with). However, I tore the driver apart and took some more measurements that should be useful.

Coil former .540g
Stiffener (minus rivets) 2.06g
Threaded pusher 2.24g
Spider 2.87g
Coil Wire 1.41g
Total 8.85g with pusher 6.61g minus pusher

Approx 17' of Metric 2.0 (32 ga AWG) gage wire wound on 4 layers of 5mm length
Aluminum Coil former diameter 28mm

There is quite a bit of wire on this driver. I'm thinking that I might try and custom wind a new voice coil and spider to see if I can get the mass way down.

I also found this company that seems to specialize in adhesives used in the speaker manufacturing indusrty:

http://www.hernon.eu/hernonmfg/index.php?option=com_content&view=article&id=107&Itemid=200

http://www.hernon.com/search-for-adhesives/540/Voice%20Coil%20Bonder%20360


I think the spider construction is causing the secondary resonance, but I have not tried to run a finite element analysis of the shape to see if that is the case. I have an open source FEM program, but I find it very hard to use. Probably because its not something I'm very familiar with. I remember Flo indicating that he did just that to prove out the Caruso head design. It seems to me you want all 4 spring sections to have the same stiffness or else you would not have a piston like motion to the spider. However, I suspect that each of the sections has its own resonance defined by the local distribution of mass and the the stiffness of the PCB material.

Mark

[EpicenterBryan]

Hi Bryan,
On the subject of velocity vs. displacement response, maybe this will clear things up. If you check out a post I made some time ago, you'll see I plotted 2 curves. The first shows 2 sine wave of the same amplitude (think stylus excursion), but with one at twice the frequency of the second.
http://lathetrolls.com/viewtopic.php?f=22&t=4046&p=33209&hilit=constant+velocity#p33209

That is a tremendous post Mark. And for others, they need to read your full post.
Let me say this Mark - the posts you make take us regular guys time to digest. I for one need several days to process what you have to say - and I hope I speak for others here when I say that a lack of response to one of your posts is really more about us absorbing and or digesting your inputs. If other readers have the same delayed response - chime in on this post and say "Mark - Me too".

From the other post and the two graphs, I see what you are talking about. For lower frequencies... and as displacement increases prior to resonance the rate of change increases as referenced to a lower frequency.

When the higher frequency is twice the lower frequency BUT is at a lower amplitude, the rate of change between the two is nearly identical (ideally) which explains why your displacement chart is a flat line up to (near) resonance. OK I'm fine with that... But I still need to get my head around it.

Ignore what actually is going on at resonance for now, or secondary resonances...
After resonance, we have a natural fall off in amplitude in the frequency realm, because of the mechanical system.
From what I understand from your post (in theory) it should be at the same rate. It is not and falls off at a much higher rate or the displacement curve would be a straight line.

For the casual reader, any time you see one of Marks displacement charts deviate from a flat line, it's really telling you the frequency response is going out of an expected response - and the magnitude will clue you into the frequency involved.

How did I do explaining that?

B

[Sillitoe]

Mark- me too!

[Ciuens]

Mark- me too x2, hehehehe.

you are the man.

I'm working right now on the lightweight version of spring in plastic. I'm riding along with the magnetic isolation "Ciuens Shield" (thanks Bryan). then put the results obtained.

[Soulbear]

Hi Bryan,
Good Point to make about concepts and explanations taking time to digest and sink in, (I'm well into my 7th Decade and in my dotage I think some of my intellect has gone A.W.O.L I'm afraid so," YUP ME TOO MARKROB" Te He Soulbear

[EpicenterBryan]

I've been playing with some crazy ideas for the last two weeks. Results have been mixed but interesting. In the first round trying two different ideas on two drivers and having to swap drivers in and out of the test jig - it became clear that something needed to change with the test jig. I was not getting the same results once a driver was removed and re-installed with no changes to the driver... even though everything was adjusted and shimmed again during re-install. So obviously, the measurement system disturbs the measurement. That's not a good thing.

I still plan to try the first crazy idea again once I have more repeatability in the test jig. The idea relates to my attempt to magnetically dampen the divers using eddy currents. I'm going to hold off on some details until I have a chance to try again - but I will say that I added some washers (spot welded together) and glued them on top of the magnet stack to extend the magnetic field 3mm into the unused area of the coil former and tried a mystery mod on the former... More later if it works. To get an idea why I though this might work, here is a YouTube video that shows the effect I was trying to take advantage of. And I have to say that I ran this idea past Mark via PM and he didn't think the effect would be strong enough. You got to go with your gut, but Mark is usually right! I have some special sauce to try before I give up though... It's an interesting idea and I hope you get where I was heading. Use your imagination for now.

After messing around with all the adjustments time after time and watching things in real time with the iPad transfer function App, I decided it was time to simplify the test jig and eliminate a support point and some adjustments - but also some possibe resonances related to the length of the carbon fiber rod and such. So I wanted to shorten the rod length and move the pickup closer. But the reason the phono cartridge was so far away was to make sure there was no magnetic pickup from the driver...

Then 2 weeks ago Mark and Ciuens both posted some really interesting stuff on this thread. Mark's post took some time to digest and I'm still to this day processing it all. Ciuens also posted an idea I now call the "Ciuens shield", using the backing metal from hard drive magnets to shield magnetic fields and I ran with it and tore down a bunch of drives. I ended up finding two parts that were large enough to over lap and I milled the surfaces so they mated in one larger part. Then I drill holes so it would mount to the driver and let a 3mm rod come out of the driver. I also used another chunk to make an auxiliary shield near the phono pickup. So far so good.

IMG_3803.JPG

Even with this setup, when the driver gets installed, the rod end support (with a rubber bushing) also needs to get shimmed and adjusted. And the phonon cartridge needs to get shimmed and adjusted forward or back (and up / down) so the stylus sits in the groove cut in the carbon fiber. It's less effort and fewer things to be wrong - but when you actually watch final adjustments in real time EVERYTHING effects the response.

So in one final redesign to the test setup, I've purchased several things. First, I decided it's silly to try to shim the cartridge up from the bottom and have no idea if the actual pressure is in the ideal tracking range for the cartridge. Tonight I bought a Micro Seiki DQ41 direct drive turntable from a guy locally for $20! It has an MA-707 tone arm on it, and a B&O cartridge. The exact same tone arm is on E-bay for $329 right now with no head shell (one guy wants over $100 for that alone). The platter speed controller is messed up and it runs at 100prm, but the tone arm is fine!

00v0v_5xebfTb3ctm_600x450.jpg

I also bought some 4mm Ruby spherical probes, and I'll be using 2 to form a super low friction V for supporting the end of the carbon fiber tube on the test setup. I also bought a spare for use as a cutting depth limiter for the lathe.

$(KGrHqZ,!rgFIsdjEN-DBSLjpU(0tQ~~60_57.JPG

CONTINUED ON NEXT POST

[EpicenterBryan]

CONTINUED FROM PREVIOUS POST:

This all leads to my next crazy idea. I wanted to see if changing the effective length of each spring so all the "spring soldiers" are not marching at the same frequency and if it effects the resonances I keep seeing. I played with several ideas that could be implemented in plastic to have many connection points between the inner coil and outer housing connection, all of which could have slightly random spring lengths... but decided it would be easy to try the idea out with some simple external mods that could also be changed.

In the first round of mods, I decided I needed to decrease the width of the springs. That lowered the stiffness of each spring. Then, I could shorten each spring without increasing the stiffness too much. For the first test, I decided to reduce the spring width by about 0.5mm each (from around 1.98mm to 1.50mm). I did this on a fine wheel grinder. I would recommend doing so before gluing the spring back onto the coil! I caught an edge of a spring on the grinder and it sucked the disc right off the coil and I didn't die! Bad.... but no damage and I was able to re-glue the spring.

In this first shot it shows 4 different spring clamps I made out of an aluminum hard drive disc spacer salvaged while harvesting MuMetal for the "Ciuens shield". The diameter is not ideal but was close enough. I just drilled 2mm holes in 4 locations and did rough cuts to separate the segments. Then I cut the longest one to a length that was something less than half the normal spring length. Each successive clamp was cut 1mm shorter (for no specific reason just for a sanity check). The plastic parts were printed as one solid ring that fit at the correct thickness plus 0.25mm so they would compress. And short segments were put in position near each spring mount temporally. Then the outer clamps were installed and a razor blade matched each plastic section to each outer clamp.

Each plastic part was 1.25mm thick, slightly thicker than needed so the outer clamp would compress the spring at that point.

IMG_3808.JPG

The plastic parts were then glued in place. The idea was to put the two longest parts opposite of each other, and the two shorter parts opposite of each other...

IMG_3809.JPG

Here is what the clamp scheme looked like on the final driver.

The response was interesting. But I don't want to post it until I have total confidence in the new test jig - so stay tuned....

Bryan

IMG_3810.JPG