Stereo Cutting Head - based on Grooveguys design

[Estrada]

Hello,

Just thought I would share about a stereo cutting head I have been building over the last six months or so. The design is based on Grooveguys design in this forum

https://www.lathetrolls.com/viewtopic.php?f=15&t=8187&p=52504&hilit=grooveguy#p52504

I've used the same audio exciters as Grooveguy but I have modified the design of the chassis to incorporate 3D printed parts. I found some proprietary T-Slot brackets online which I thought had potential for use as a chassis for the head. I bought some and they size seemed just right. I then designed a set of 3D printed interior parts that fit into this chassis and house the exciters, wiring etc. I could have printed this as one single piece but I thought that there would be an advantage in being able to remove one piece in case I blew a driver, that way I could replace the affected piece rather than the whole lot. It does also provide flexibility and working room when it comes to fitting the torque tube. There is also a 3D printed out shell which tidies everything up. The torque tube is made from 3.2mm diameter aluminium. It is designed to fit a presto style cutting stylus held in place with a 2mm grub screw. Making the tube required a couple of 3D printed jigs to act as a guide for drilling holes in the right place. It worked but the accuracy would be a lot better if I had better equipment (all drilling is done with a hand drill). I have modified the stylus geometry as I am cutting plastic and not lacquer so there is no spring back to allow for. The head is mounted at 8° to the record surface with a 7° angle in the stylus hole so that the stylus is just off 90° to the disc surface (about 1° plough angle).

So I put everything together and it seemed to work reasonably well. The only thing that wasn't so good initially was the stereo separation. Initially the torque tube design was similar to a Caruso head, I eventually worked out that I needed a V spring to improve the stereo separation. I mocked up something out of plastic and it seemed to improve things a lot, last I checked there was 6 - 10db of separation. I haven't made a decent V spring yet but I am thinking of building a second version of the head so I might do so at that point.

I'm using a Myshank diamond and cutting into polycarbonate. Thus far I am quite happy with the sound.

Here are some photo's showing the final head and the basic parts.

Head assembled and mounted to lathe

IMG_3973.JPG

Underside

IMG_3974.JPG

3D printed parts and aluminium T-slot bracket

IMG_3976.JPG

Assembly

IMG_3977.JPG

Top cover

IMG_3978.JPG

I have also attached a file with a sample cut. There is a little bit of flutter which is an issue with the lathe platter, I've been working on this and I think I have it sorted.

[farmersplow]

You must be a hand drill virtuoso!
It is unbelievable what you have achieved with simple tools. You can be very proud of yourself. What would you be able to do with the right tools?
The quality of the cut is excellent (except for a small crackle). Many (including myself) would like to achieve such good results in this phase of development.
(Which power amplifier do you cut with? Because the high notes also sound quite decent).
Nice greetings from Vienna
Thomas

[Estrada]

Thanks for the kind words Thomas. I was happy with the sample except for the slight speed issue which I have fixed. I think the crackle might have just been because I didn't clean the record first, I think it cleared up after a good clean. To drive the head I am using this very simple and inexpensive power amplifier;

https://surplustronics.co.nz/products/10408-amplifier-board-dual-channel-stereo-high-power-digital-

I'm running it on 12v so it has an output of 50w per channel which seems to be plenty to run the drivers. I'm planning to make a second version of the head, for this version I would like to have the 3d printed parts fabricated from aluminium and would like to improve the accuracy of the build, particularly with respect to the torque tube. I know someone who could possibly fabricate a V-spring from spring steel which would also be great.

Thanks
Craig

[grooveguy]

Hello, Estrada, thanks for sharing this. Beautiful work, especially given the restriction of hand tools, and the audio quality seems remarkably good. Do you use as much EQ as in the write-up to achieve these results? Sounds quite smooth, although the stereo 'soundstage' does not appear very wide. As I found, you may have to crossfeed -L into R, and -R into L to optimize the stereo. I have the feeling that the V-spring may well hold the key to separation, as it defines the planes in which the stylus can move most freely. It appears that you have indeed anchored the tail of the torque tube, rather than burying it in rubber as I did. I am going to try this; again, this part may be an important factor. Good work!

[Dogtemple]

Well done and good idea on the bracket.

Very good results there and I’m surprised on the amp, didn’t think something like that would do the job but it sounds great

[Estrada]

Hello, Estrada, thanks for sharing this. Beautiful work, especially given the restriction of hand tools, and the audio quality seems remarkably good. Do you use as much EQ as in the write-up to achieve these results? Sounds quite smooth, although the stereo 'soundstage' does not appear very wide. As I found, you may have to crossfeed -L into R, and -R into L to optimize the stereo. I have the feeling that the V-spring may well hold the key to separation, as it defines the planes in which the stylus can move most freely. It appears that you have indeed anchored the tail of the torque tube, rather than burying it in rubber as I did. I am going to try this; again, this part may be an important factor. Good work!

Thanks Grooveguy. Yes, there is a fair amount of eq involved to get a flat response. Here is a screenshot of a pink noise test fed straight into the head with no corrective eq

Pink Noise RIP - no eq.png

The response looks a bit different to yours which I would assume is due to differences in construction. My main resonance seems to be a bit lower than yours, I'm not sure whether this is a good or bad thing but I seem to have been able to get it under control. I'm not sure that my eq setup is perfect yet but it is acceptable.

The V-spring definitely seems to be the key to getting separation. I initially started out with a shorter torque tube with a longer piece of piano wire as you can see in the photo below. This was based on a photo of the Caruso head.

240016015_542700610115176_4379079199960814918_n.jpg

Unlike the Caruso head I don't have a feedback system so I don't think this design is right for a dynamic head. What I observed was that there was quite a bit of vibration in the piano wire. My theory is that when the directional movement from one channel is applied, it creates a vibration in the piano wire which behaves a bit like a guitar string in that it doesn't just vibrate in the direction that it is struck but turns into a rotational vibration which transfers that vibration in all directions including back into the opposite channel. I think that the V-spring stops this rotational movement from happening.

I did try some -L and -R cancellation at one point and it did seem to help. I think this was before I installed the V-Spring. I might do some more experimenting with this as I would like to improve the separation. I do recall getting some strange effects in the audio, probably from overdoing it. I'm thinking it might be worthwhile turning down the centre information from the out of phase material. It is quite likely that my construction lacks some accuracy which could be contributing to the reduced stereo field, possibly with regard to the angle of the pushrods. Hopefully this is something I can improve on in the next iteration

[Estrada]

Well done and good idea on the bracket.

Very good results there and I’m surprised on the amp, didn’t think something like that would do the job but it sounds great

Thanks Dogtemple. Yes, the amp seems to do a good job. I thought it would be a good starting point while I got the head operational, I have already destroyed an old stereo receiver that I was using while working on my old mono head so I was keen to get something cheap so it wouldn't be a disaster if I blew it up. It seems to be doing the job though so I'll stick with it until I find the need for something better.

[grooveguy]

I think you hit the nail on the head, insofar as the piano wire is concerned. I didn't use piano wire, but a small diameter rod, which (judging by a pink noise test) has its own resonances... and in directions other than the driver is aiming it. I especially like your coupling to the voice coils. The lightweight aluminum funnels were an attempt to keep the mass low, but your adapters appear to be 3D printed and probably don't amount to much mass. Anyway, even with more mass, just increase the stiffness and pour a few more watts into the coils to overcome the attendant loss. That really is a very nice looking head.

I also notice that your peaks and valleys in the raw response are not nearly the heights and depths I encountered, and I'll bet you have a good deal less 'holdover' at the resonant frequencies. Obtaining a flat swept or pink noise response looks good on paper, but if there is ringing at the resonances, it muddies-up the music. I know I hear a bit of that with my head.

[dmills]

Tip for the V spring, PCB manufacturers offer a laser cut and electropolished stainless steel 'stencil' intended for applying solder paste...

If you draw the geometry you want on the paste layer of a PCB (Remember it is negative you get a hole where you draw the paste) you can get a great many 0.1mm thick highly accurate stainless steel springs out of a small stencil, just remember to leave tabs so the whole thing does not fall to bits before you get it (And ask the PCB shop to NOT modify the stencil design).

This is usually cheaper then wire EDM.

[Estrada]

Tip for the V spring, PCB manufacturers offer a laser cut and electropolished stainless steel 'stencil' intended for applying solder paste...

If you draw the geometry you want on the paste layer of a PCB (Remember it is negative you get a hole where you draw the paste) you can get a great many 0.1mm thick highly accurate stainless steel springs out of a small stencil, just remember to leave tabs so the whole thing does not fall to bits before you get it (And ask the PCB shop to NOT modify the stencil design).

This is usually cheaper then wire EDM.

That's a great tip, thank you! Will definitely look into this.

[Estrada]

I think you hit the nail on the head, insofar as the piano wire is concerned. I didn't use piano wire, but a small diameter rod, which (judging by a pink noise test) has its own resonances... and in directions other than the driver is aiming it. I especially like your coupling to the voice coils. The lightweight aluminum funnels were an attempt to keep the mass low, but your adapters appear to be 3D printed and probably don't amount to much mass. Anyway, even with more mass, just increase the stiffness and pour a few more watts into the coils to overcome the attendant loss. That really is a very nice looking head.

I also notice that your peaks and valleys in the raw response are not nearly the heights and depths I encountered, and I'll bet you have a good deal less 'holdover' at the resonant frequencies. Obtaining a flat swept or pink noise response looks good on paper, but if there is ringing at the resonances, it muddies-up the music. I know I hear a bit of that with my head.

I'm not entirely sure what you mean by 'Holdover' but I agree with you on the flat pink noise response. I have found that I have needed to lower my eq around the main resonance more than the pink noise test would suggest to get rid of some ringing at those frequencies, it has yielded better results though.

[grooveguy]

Yes, exactly. And that's what I mean by 'holdover.' Actually a sound at that frequency that remains (ringing) after the legitimate sound ends. Generally a term used for bass frequencies where an actual 'tone' can be heard after a drum beat, for example; nevertheless, happens at any frequency and just serves to increase the energy there.

[dmills]

This is one of the major differences between an open loop (non feedback) head and a closed loop (feedback design).

In the feedback head you design the thing to be very under damped (At least for most of the resonances) and then do the damping in the electronics, this improves the efficiency and lowers the power required at high frequency. There is a slight proviso in that some of the high frequency resonances need damping because the acoustic delay is too long to allow the feedback to do it.

In a non feedback head you pretty much cannot do this as you have no way to compensate for errors in the response, and the Q value is not particularly consistent, so you use mechanical damping to ensure no high Q resonances can exist. It costs you efficiency and is why most such heads do badly in the top octave.

[grooveguy]

Right you are about that! I tried putting feedback around a Presto 1D once, but most of the feedback came from magnetic coupling with the drive coils. Something like a capacitive pickup would have done better, but the 1D was developed for open-loop service and did pretty well for itself as it came.

[dmills]

It is one of the reasons I am never quite convinced by the open loop copies of the SX68 and friends that everyone builds, there surely has to be a better way of doing an open loop head that gets the damping closer to reasonable for that operating mode, damned if I can see what it should look like however.

[grooveguy]

Perhaps we should take a cue from the various mono cutterheads that gave reasonable and predictable results all those years ago. Most had damping very close to the stylus tip; that is, the damping was tightly-coupled (or a low-impedance relationship) with the stylus. Most homebrew stereo cutters utilize the traditional Westrex design, which I think is the same as your reference to SX68. There may be other and better ways for the amateur to skin the cat; my next test will be the Ortofon 'rocking bridge' design. Again, without feedback, but with fingers crossed and hope springing eternal.

[dmills]

probably going to be tricky to get the compliance identical for vertical and lateral motion with that, but worth a shot, maybe do something like install the damping between the front of the inner pole piece and the flexture that couples to the bridge assembly.
I would note that the length of the cutting diamond is fairly critical to the geometry in that head design.

I am ruminating on doing something with damped flextures (And **really** wanting a wire EDM rig).

[Mr.Inventor]

First thank you for saving me a bunch of time. I was in the process of copying Neumanns old mic designs updated for modern voltages (45B+ on tube and 250vdc in, post war Germany, it was what they had) and more easily available tubes, 6AS6 subbing in for a vf14, when I found his lathe designs and everything else was dropped for a moment. I'm a computer engineer who had a father who was big into tubes, I KNEW with modern magnets the drive heads wouldn't need anywhere NEAR the massive power they had in previous cutters, a pair of el82 tubes could easily drive a NIB magnet Through a vinyl record. Over the years I've seen hard drive heads (that undergo higher g forces than a recording stylus) plow themselves through the sides of aluminium drives, perfectly cut the platters into rings, etc.


My question/comment is this. While searching and finding all the schematics and mechanical explanations and etc for the various lathes. ALL of them seemed to indicate that one of the cutter head coils was in reverse as this apparently dealt with an EQ issue that wasn't fixable with then technology... Here on this forum people have found the inverted coil on the westrex system. On the Westrex schematic the channels are phase split and my understanding was one of the channels was purposely backwards on the recording head. This apparently forced the head to resonate differently. I apologize if this has been addressed and I'm just new.


I too am starting the build on such a device from a printer/computer background. The audio path needs to remain analog and even the power supply will be a basic transformer (I'm wrapping mic sized ones by hand right now. The spindle and the advance will be controlled digitally. An arduino can very easily examine a song ahead of time, find the various pitch/bass levels and adjust the track advance from a very simple program. In this way the pitch control is removed entirely as the modifications can be done digitally to a metal blank in my case that is recording an analog signal.

I built my own embroidery machine from scrap and amazon parts, I couldn't find the time to market it correctly as I was in the process of moving and I was way nervous.
https://www.indiegogo.com/projects/an-open-source-embroidery-machine#/
There was also already a market for what I created. Apparently NO ONE makes new record lathes, not with proper audio paths and whatnot. There is a crowdfunding version that uses.. transistors *spit*, however it IS a lathe but that is the only one I know of.

FYI you need to lighten the transducer assemblies, I'm going to hand wrap coils to fit inside high end NIB magnet cylinders and use titanium wire pushers with actual (synthetic)diamond anvils, stuff Neumann could have only dreamed of. I'm planning a magnetically floated drag head rather than a pusher head, better to match the rake angles of the read and write together.

Also how big do you want the EDM rig ? I can guide you to make a small one easily.


Building something that nobody makes anymore, quite looking forward to it. I need to post a second thread asking the question, what people would pay for a modern lathe?, I have no idea of what such a thing would even cost?

[dmills]

The reason for the reversed polarity to one of the drive coils is that it causes the mechanical summation to work correctly to give lateral movement for mono input and vertical movement for the side input.
Sure you can do the same thing electrically, but the SAL74 is two sets of identical electronics and the thing is unbalanced so it was easiest to do it at the driver (Just note that the feedback polarity must match or the loop will oscillate).

There is a major trap on the power required thing, unlike a classical power amplifier, this one needs to make most of its power at high frequency (Beefing up the snubber is a good idea), almost no power is required in the mid band, and worse the drive coils are rather reactive up where the action is, consider SOA carefully. You don't need to be pushing 8A into the coils very much of the time, but for the 100us at a time you need it, you do need to be able to do it. Power required generally rises at 12dB/Octave from about the main resonance on up.

You are going to want something a fair bit more skookum then an Arduino for doing pitch computer, I suggest a snickerdoodle because it lets you build whatever logic interfaces you need in the FPGA fabric, but a raspberry pi compute module would work, as would an IMX.6 or such, it would just need more external interfacing logic. Having the FPU is a major win, as is having enough memory to do the delay line.

I wish you luck with the head, they are tricky things.

[Mr.Inventor]

You are overthinking the audio works, you don't need magic just trim the frequencies you want and have the arduino sample multiple input pins to "hear" the various frequencies. You could have it read a summed voltage to detect the places you need to worry about.

I don't understand it, the proof of concept has been done, why has nobody built a new lathe in 30 years ? A brand new functioning lathe is easily worth $10,000 It's like my previous invention, people don't realize how simple it actually is. Yes I'll more or less have to make the actual power amp block just because of balance issues and whatnot, but still, meh, tube amp tied to a pair of magnetic actuators. Before your work I thought there Must be more to the head, you did the hard part, advancing the head along, adapting the speed for sonic grove size, thats programming crap I'm good at.

You did the proof of concept hard stuff.