Stereo Cutting Head - based on Grooveguys design

[markrob]

Mr. Inventor,

Sounds like you have it all worked out. Have at it and let's see what you come up with.

Mark

[dmills]

Now Mark, play nice...
Mr Inventor, sit down and do the maths, it is revealing as to just how hard some of this stuff is. Also, grab the AES Disk cutting anthologies, a few hours in a library can save you weeks in the lab.

While it is true that we have some better materials today (Magnetics being a big one, also computation which gets you fun things like coreless BLDC motors), it is also true that back in the day there was effectively infinite money behind cutting a better record. Having everything hardened and ground was doable, custom grey cast iron assemblies, doable.... That sort of thing makes a difference.

Have fun, but expect this to be way harder then it looks.

[markrob]

Hi Everybody,

Sorry about that! Just rubbed me the wrong way. Definitely get the AES papers to start.

BTW, some here have done pitch controllers using an Arduino, so its not impossible. IIRC, the original Zuma was done on an S100 based computer. The ATMega2560 is probably on par with that hardware. I think they had quite a bit of external hardware to assist the CPU. If you are planning to make the next Neumann, I think you will want to throw more MIPS/FLOPS at the problem given how little that costs today.

Mark

[dmills]

Yea pitch (And depth, actually more important!) are way easier when you can just throw RAM and CPU at it. Something like a STM32F7 feels like a good spot for this (But you try getting one!).
I would think that on a low end the preview->displacement filter would probably be hardware (At most basic, IRIAA plus an integrator, but there are cleaner approaches), and some of the groove snuggling stuff from the VMS80 might get rather approximate.

The VMS80 (Circuits are out there) is a fascinating example of what is doable without any CPU (in the modern sense) at all, they just used an infinite quantity of 4000 series cmos.

[Dogtemple]

as someone who is restoring a lathe and having to build a lot from scratch, I can echo what has been said here already, these things are simple in principle but in practice there are a lot of nuances and they're actually pretty complex despite how they first appear.

its relatively easy task to make something that works to a degree but to make them really really good takes a lot of precision and knowledge, and that part is difficult. I think theres a reason why the gakken record maker is £100 and the other options are loads of cash or don't really exist to buy any longer...

I've been reading solidly on the subject for over a year now and keep discovering more that needs to be kept in mind. and I'm sure when I have put all this in to practice it won't work and will need lots of tuning to get just right.

[dmills]

That's about the truth of it.

'Sort of works' is easy, for shits I tried cutting with a disk held in a mandrill in my Hardinge HLV-H and a cutting head mounted on the cross slide... It sort of worked.
Reliably making a commercial sounding record is MUCH harder.

Realise that the noise floor implies cutting way down below the wavelength of visible light.

[orlandoinsane]

Super interesting project, would you mind sharing your CAD for others to play with?

[Estrada]

Super interesting project, would you mind sharing your CAD for others to play with?

Yes, happy to share files. Send me a PM if you are interested.

[B_Lach00]

Hi, I'm curious about those two thin wires attached to the stylus. What's their purpose? Also I would be very grateful for sending me CAD files, I'm designing my own cutting head currently but seeing files of already made and working head would help me a lot. Cheers!

[Estrada]

The wires coming from the stylus are heating wires. Not totally necessary, in fact I don't use them most of the time. If you sent me a PM I can send you the files for 3D Printing.

[Estrada]

Greetings to anyone following this thread. I thought I would post an update to share some of the things that I have been working on with this design to try and improve on the sound quality and response of the cutting head. One of the main issues with this head has been unwanted distortion caused by resonance and/or harmonics. At first I was controlling this by limiting the volume and applying quite a lot of de-essing. This worked to some extent but limited the amount of I could get out of the head before getting really noticeable distortion. I tried a couple of things to sort this out. My first thought was that the 3D printed pieces might be contributing to the problem so I milled some aluminium parts to replace the pieces which the drivers are mounted onto

IMG_5551.JPG

This didn't really make any difference. I do like the aluminium parts, they feel more robust, but they didn't make much difference.

So I decided that my next option would be to try adding some kind of damping material to the drivers in order to reduce resonance. During some previous experiments with my old mono head, I made a new dampening membrane with scotch mounting tape which you can get at almost any hardware store

Screenshot 2025-03-08 at 11.33.55 AM.png

I found that this had very similar physical properties to the original material in the mono head. There may be better material out there for the purpose but this is what I have used to date. It is a bit hard to work with to any degree of accuracy because it is so sticky. I rather crudely stuffed some of this material into the diaphragm of the driver. I don't have photos from the original head but here is a photo of a second head that I have built. In this image the installation is a bit neater, I have wrapped 3 layers of the tape around a piece of plastic tube. Although this is neater, I think it may need a bit more depth as it may not be damping quite as well as the first head.

IMG_7278.JPG

The main resonance seems to be sitting at around 400hz. Below is the response of the head. this is showing the playback of a pink noise test cut with no corrective eq. The main resonance is clearly visible with a peak around 400hz.

230302 Pink Noise Flat.png

With the damping fitted to the driver as described above, the main resonance is peak is pretty much eliminated by the damping alone. The image below again shows the playback of a pink noise test cut with no corrective eq

241014 Pink Noise Test Stereo Head.png

As I mentioned above I have now built a second head. Both the original head and the second head have been fitted with dampening material. The original head has the dampening material packed in more tightly and so I have noticed that this head takes more power to drive and has a lower volume than the second head. Overall the volume on the original head is a bit lower than I would like, it is not far off a pressed record but it is lower. The second head in which the dampening is not so thick (see image above) can achieve higher volume but I have had more issues with distortion on this head. Perhaps with more experimentation I will find a balance somewhere between the two. Possibly another material may work better. Perhaps a pourable urethane which can be cast to form a more consistent shape.
I hope this is helpful to anyone who has attempted to build this head.
Here is a photo of the second head. It is mostly build out of left over parts from the first head. You'll see I've now got funnel cones fitted to the drivers which simplifies things a bit. I think I have improved on my assembly this time around as the stereo separation seems quite good.

IMG_7430.JPG

[grooveguy]

Good sleuthing and application, Estrada! Removing or at least lowering those resonance peaks and valleys through mechanical means is definitely necessary in a non-feedback cutterhead. Your second spectrum plot certainly looks so much better. I hope you are making some test cuts so we can hear the progress as you refine the damping.

It appears that head efficiency and response flatness are tradeoff partners (or adversaries) in all this, and as more damping reduces resonances, and increased stiffness raises the cutoff frequency, more power is going to be required from the cutting amp. And, once more, this means a tradeoff between what the amp can deliver to the head and how much the head can take before the voice coil vaporizes. The little 3-5 watt drivers you and I have been using may prove inadequate as you make strides in the audio quality, but even the next size up on that same variety will increase the size and weight of the head. Keep trying and be sure to post the results for the rest of us. Good work!

[Thelatheofus]

Really nice build estrada ! That torque tube and v spring looks really nicely crafted.

My take on this head that I also built is that the resonnances come from the drivers not being adapted to this job. I had decent results at low levels, but when getting louder there was a lot of distortion.

I took one driver apart, and I think the issue is a combination of : the fact that there is little clearance between the voice coil and the inner magnet (so when the whole assembly moves from left to right the voice coils touch the magnets), and the spider suspension that is not stiff enough.
As Konstantin Tokarev explained to me on a fb post (he builds cutting heads from start to finish, making the voice coils, suspensions etc..) there is a fine balance to find between the power of the voice coil and the stiffness of the suspension (which souldn't be a spider suspension, but something stiffer, like the ones used in neumann heads).

A suspension
- too stiff would require too big (and heavy) voice coils (which will lower the main resonance frequency too much, or would require too much power to get highs),
- too loose will cause distortions because the speaker doesn't come back to its original position quick enough

Most of the homemade heads that I saw that worked well that used off the shelf drivers used more powerful ones than the DAEX19CT, usually around 30-40W if I'm not mistaking. There are a lot of heads that work nicely with off the shelf drivers, so I'm pretty sure it would be possible to adapt others to this cutting head design (which I really like), but I'm tempted to try to build drivers from start to finish someday (or re-using some parts from pre existing drivers and adapting them to this task).

[grooveguy]

Many thanks, Thelatheofus, for those observations... good information. Please keep us all apprised of any successes and breakthroughs in your continuing attention to our mutual endeavor.

[Snowmaan]

This is beyond my scope of understanding but still looks beautiful

[grooveguy]

Guys, here's the transducer I used in the first version of my stereo cutterhead. Diameter is about an inch, and talk about stiffness, it's very stiff. What's more, it has a good deal of metal in the magnet structure, which might help dissipate heat better than the shallow transducers. No need for little funnels either, the voicecoil terminates in a threaded stud, about the metric equivalent of a 12-24 screw that can be drilled axially for a pushrod, and laterally for setscrews. Trouble is, I haven't seen these around much anymore, not even on Ali Express, where you can find most anything of an 'offshore' nature. If anyone runs across these, let me know, please.

[Thelatheofus]

I can't edit my last message, I'll do it here :

"My take on this head that I also built is that the distorsions (not resonnances) come from the drivers not being adapted to this job. I had decent results at low levels, but when getting louder there was a lot of distortion."

Thanks for the heads up groove guy, iirc you moved away from those drivers because the suspension was not ideal either ?

[grooveguy]

Thelatheofus: Are you cutting or embossing? I've never experienced level-dependent distortion with homebrew cutterheads, but almost all my experience has been cutting lacquers. My own attempts to emboss have met with only marginal audio-quality results, and I have noticed that at high levels there seems to be more of what sounds like the playback stylus rattling around in the bottom of the groove. I have heard some surprisingly good embossing results by others on this forum and would like to get to that point myself.

[Estrada]

Thanks for the various replies to this.

"My take on this head that I also built is that the distorsions (not resonnances) come from the drivers not being adapted to this job. I had decent results at low levels, but when getting louder there was a lot of distortion."

Thelatheofus, I agree with your observation, particularly with the first iteration of the head which did not utilise any damping. I too could get nice results provided the volume was not too high, above a certain level the distortions would creep in and they could be quite nasty. It hadn't occurred to me that this might be the result of the coils rubbing against the side of the magnet, but this makes sense. I've had a look at an old driver and you are right, there is not a lot of clearance between the coil and the magnet. I am wondering if it might be possible to correct this by removing a small amount of material on the lathe, then remounting the coils with a steel spring.

I've actually built another head using some larger drivers, I think they are the next size up from the ones Grooveguy originally used. They are much bigger and heavier than the little coin exciters so the resulting head is much larger. I did notice that there is a lot more clearance between the coil and magnet on these drivers.

IMG_7288.JPG

I haven't progressed this head very far, other than to do a quick test. As I suspected, the resonant frequency is quite low and the HF response is poor due to large moving mass. I think from memory the springs on these drivers are 0.6mm thick. The next step will be to replace the springs with something thinner and lighter and probably make up some lightweight cones in place of the bulky aluminium attachments in the middle, which in this case are just the big flat aluminium pads that came with the drivers with the flat part cut off. These drivers have been used successfully by others to build good cutting heads, including a couple that utilise feedback so I think it will be worth pursuing.

It appears that head efficiency and response flatness are tradeoff partners (or adversaries) in all this, and as more damping reduces resonances, and increased stiffness raises the cutoff frequency, more power is going to be required from the cutting amp.

Grooveguy, there is definitely a relationship between the head efficiency and flatness response. The damping material certainly increases the amount of energy required to drive the coil, but it does cut down on the resonance a huge amount which means very little corrective eq to get a flat response. Perhaps a good solution would be to use it more sparingly to reduce resonance while dealing with the distortion from the rubbing coils in another way - perhaps by replacing the springs or possibly using a different driver with better clearance between the coil and magnet.

After writing the previous post, I accidentally tested the first head to destruction as I had the wrong fuses in line. It took a lot of power to get that head cutting at a volume similar to a pressed record and the coils eventually fried. I think I was pushing around 6-7v into them on average and it took that for quite a while before they gave up. The head was actually quite hot on the outside, if it had been the head with all plastic parts I think some of the plastic would have warped. They do seem quite resilient.

[Thelatheofus]

That last head looks really nice !
They look indeed like the drivers used initially by "the farmer". My bet is that most of the improvements will come from changing the springs more than getting more clearance. You have big margins of improvement to reduce the weight of moving parts here. Building a proper 1:20 drill bit to use a 320 stylus would be a big improvement also that could save precious mg.

Good luck with your build!