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

[EpicenterBryan]

I call this project the "Groove Scribe".

It's a DYI stereo cutting head you can build yourself with off the shelf drivers and lots of work on your part. Right now, this is a non feedback project.

This is a new thread, using things learned from a previous thread ... http://lathetrolls.com/viewtopic.php?f=15&t=5249. The old thread is not dead for ever. Eventually things learned on this project may be applied to the old project with the goal of having a feedback system.

IMG_2900.JPG

Since a few things have been discussed about this in the other thread, I'm going to start this new thread with a test video first. A new test video will be this weekend and as I have time in between, I'll fill in some details about the components that were made for this head.

Let's get right to the first audio test and be sure to watch the whole video for lessons learned and what's next!

[opcode66]

You don't know that it is actually stereo until you do test cut with audio delivered to the head in just one channel. If on playback you have audio in only one channel then you have a true stereo head. If on playback you have audio in two channels, then you have crosstalk issues. I am pretty much going to guarantee you have crosstalk given your design.

[EpicenterBryan]

You don't know that it is actually stereo until you do test cut with audio delivered to the head in just one channel.

Todd, I love you like a brother but it's stereo man. Listen again.

I will do Right only and Left only noise cuts this weekend. That will not only give channel separation data, but will be used to generate separate R and L EQ curves for a new cut.

Todd, I am not trying to detract from the efforts and discoveries you are making and implementing on Blade Runner. Discoveries made on my last project are based on ideas of yours that you were willing to share with us all. You are the man. We all appreciate you.

[ROLANDJAYS]

wow, I like it !
cool man cave you got there !

[opcode66]

Todd, I love you like a brother but it's stereo man. Listen again.

Not trying to be a detractor or anything. I understand that there is audio in both channels. That was not what my point was. If there is crosstalk or not determines if you have true channel separation and therefore a true stereo experience and not simply a stereo expanded experience. Dig?

You can play a mono cut record with a stereo cartridge and have audio in both channels. Is that truly stereo????

[EpicenterBryan]

All the sudden, I may have to ditch the entire video / first cut test because YouTube (owned by Google) recognized the song in the test and sent me this nasty-gram:

"Your video may include music that is owned by a third party.
To hear the matched music please play the video on the right. The video will play from the point where the matched content was identified.
Your video is available and playable.
Copyright details:
"Donald Fagen-Slinky Thing", sound recording administered by: 3:06
WMG
If you don't want this claim on your video, you may be able to remove the song. Learn more"

Nice. Never whistle while you work either or you will be fucked.

So I'll need to delete the video I just posted, start the thread over, and use original content for frequency sweeps.

Fuck Google and Fuck You - YouTube!

[opcode66]

It doesn't matter. As long as the initial copyright owner doesn't care.

[Ciuens]

Hello Bryan, congratulations on developments in its cutting head. I'm working on a using the same transducers. In my tests the sound has a very high resonance, due to the rigidity of metal springs. just print new springs in ABS to test the results, and seem much more exciting. following images of printed pieces and a video of the working transducers.

http://youtu.be/SeMUjsGP8IA

[EpicenterBryan]

[quote="Ciuens"]just print new springs in ABS to test the results, and seem much more exciting.

AWESOME!!!!!
Ciuens, your video is a must view for anyone who is having a hard time visualizing how this 45/45 system works! I know Todd has a good handle on this but I had a really hard time getting my head (pardon the pun) around this because the movements involved seem so small. It's kind of like a non electronics guy trying to imagine a microsecond. With the proper tools like a scope, it's no big deal but for a first year student (or a newbee like me with a cutter) it's a steep learning curve to climb.

I am really impressed with your mods to the driver! I asked Bennett if he had access to an EDM machine at his school thinking he might be able to make new less springy springs but your solution is just too good.

I have given some thought to changing the mounting orientation of the drivers and cutting two of the spring connections to encourage the driver to move in the correct direction with less effort but your solution is a far better one. It's pretty cool that you have used Todds idea incorporating the the push rod into the driver and the same connection method to the torque tube. It solves many of the the difficulties hooking everything together.

Bravo, and thanks so much for sharing that awesome video!!!!
More videos!!!!!

[opcode66]

The video points out a problem. If you watch, when one transducer moves in and out, the other teeters. If you have teetering, you don't have proper vertical movement in the stylus.... You need another spring to keep the transducer from teetering.

[Ciuens]

Hi Todd, like this?

[opcode66]

That works. My transducer springs have 3 contact points, not 2. That is the only real different. But, yes. That will now torque the tube and give vertical movement.

[EpicenterBryan]

Here is a quick update.
Major progress last weekend. I've made a few changes to the head and had better response as shown below. I'll be posing a new test cut / video and then start discussing the parts and how I made them.

I've added some Sorbothane in several key locations, including behind the V spring, around the piano wire connection to the body, and at all 4 spring points on both drivers. Those pieces and not glued in, so I could get them out depending on how it went. I've also glued some on the inside of the plastic body in several locations. I also added a segment of carbon fiber on each push rod in the center. I'll go into this more and post pictures.

I have discovered there are many variables involved in this project for sure. It turns out the frequency response changes with cutting force, and how much signal you drive with as well.

I checked the channel separation (cross talk) at it looks like it's at around 15dB. I'm pretty happy with that.
For that check, I cut white noise right only, then left only. For each measurement I took the difference between the peak reading for the driven channel and the non driven channel.

One of my other problems was since the head weighs so much more than my 1D I had to add a spring. But the spring needs to connect to the Z carriage or the spring tension changes when the head is moved up or down. So I revamped my scheme and I built a mechanism to counter balance the head with a pulley arrangement and a weight attached to a shuttle that rides on a vertical track behind the overhead. I'll show it in the video. It's pretty slick.

Some of you may know that my lathe was not cutting at the correct speed and that I installed a variable frequency drive so I could adjust the synchronous motor RPM. If you saw the last video I mentioned I was going to use a frequency counter and make sure the 1K reference was dead nuts on when playing back the frequency test cut this time and I did. Although the center hole on the disc was just a hair bigger than it should have been so the cut was slightly off center, the HP counter said it was +/-1hz in real time while I was recording playback from disc. And the software agreed with the actual recorded data file playback registering a peak at 1001hz so I'm confident in the measurement this time!!!!

For tonight, here are some new curves. One thing to note is that the right and left channel responses are not identical. I know I did not get the carbon fiber centered exactly on both driver push rods, and I know that the 4 pieces of Sorbothane were not the exact same thickness on each driver since I cut a strip then cut the 4 small segments and installed on one driver, then did the same thing to the other. So even if the drivers had the exact same response to start I doubt they would ever respond exactly the same in anything hand built like this gizmo.

For this testing I am not doing anything about the low frequency issues. I still have a bit of 60hz noise I have to chase down and a bit of rumble. And for these curves I'm not doing anything about 10K and above since I'm still not using a heated stylus nor am I heating the disc.

So here is what you have been waiting for:

Left Channel Response 1W peak - No EQ:

GS response Left Channel 12-14-14.JPG

Right Channel Response 1W peak - No EQ

GS response Right Channel 12-14-14.JPG

Left Channel EQ. Top line is no EQ, bottom line is after parametric EQ:

GS Left EQ 12-14-14.JPG

Right Channel EQ. Top line is no EQ, bottom line is after parametric EQ:

GS Right EQ 12-14-14.JPG

[opcode66]

I want to hear rips. The pictures you are posting of your grooves seem like they are very low volume. Not a lot of movement. Which means you must be amplifying the rips of your cuts a lot. Either analog or digitally. The problem is that gives you incorrect results because of background noise in your grooves that is over amplified.

I can cut a noisy unmodulated groove. Rip it. Raise the volume to max levels digitally. Then do a frequency analysis. Guess what? You have a curve. And, you have a lot of high frequencies. Just from bg noise and hiss. Do the same for the original low volume audio and what looked like a response curve before now essentially goes flat.

Food for thought. Until you have good sounding and loud audio, frequency sweeps are only giving you a ballpark idea of what your response is.

[Bahndahn]

I see no specification of volume in the images.

[opcode66]

We have seen in the video Bryan posted the grooves that were cut with his head. I can read grooves as can any experienced cutter. I know how much excursion you must have to achieve specific volume levels. But, other than that, frequency response charts, the left axis is in db...... So, there is an indication of level. But, I have no idea if it has been doctored. Is that representative of the level on disk, or has it been raised more than your average phono pre would raise it? And, has it been raised again in the computer? That is the question. How much the gain was added after ripping is what I'm asking about.

Again, if I cut a noisy groove with no audio (like one with no heat), rip it, amplify it to max volume in the computer, and performed a frequency analysis on it, I would get a curve with data in the high end. But, that is obviously erroneous data since I did not cut any sound. Without actually hearing the source audio and knowing what transformations it went through, seeing the frequency response is rather meaningless. Which is exactly why I've been posting the audio for my rips for all to hear as well as the frequency analysis. Dig?

Also, curious why a block size of 8192 is being used for your analysis. Seems awfully large.

[Bahndahn]

This excursion stuff you speak of I don't know, I have never cut a record! What I can share is: we must be cautious of our terminology if we want to be specific and critical.

-There is nothing called volume in digital or analog audio. Its called amplitude.

-dB is a relative measurement. Take a look at the various types and always keep cognizance of the various 'gain stages' present in the signal path [i.e. the phono preamp used, the analog to digital converter (and its settings) used, and any other possible gain stage in the chain.

-look at the 'relativity' factor in his cuts - there is a profound rolloff in the high frequency register suggesting the noise floor is lower than 'measured'.

-also [somewhat unrelated], look into the 'function' of dithering... a dither is the use of noise to 'increase the dynamic range'---just what does that mean? Food for thought.

I totally agree with you in your quest for disambiguation. I value your questions and am interested in the answers! If we shoot towards a more 'fundamentalist' vocabulary when talking about this stuff, we will get to the goods faster.

[opcode66]

You are nitpicking the wrong direction... We have well covered what db is. What level is on disc. How that is measured. 0 VU = +4dbm which is = 7cm/s of 1Khz tone on disc. Etc. Search the board. I'm part of that discussion. I absolutely don't need a lecture from you. Especially since that has nothing to do really with what I'm saying.

The amount of excursion, which is the proper terminology for record cutting, is really the only way to measure level on disc. There is no other way that is not relative. I don't see a lot of movement in the video that shows us his grooves.

You can absolutely show better frequency response that you are actually getting. That is the point I'm trying to make. The rolloff could simply be background noise or hiss itself not existing in the entire frequency band.

I know from experience here that you can show more frequency range in analysis due to artificially raising the volume and then doing analysis. I'm trying to help and offer advice from my personal experience. Take it or leave it.

Try it for yourself. You don't have to be able to cut records to do this. You only need to be able to record. Take a rip of an old record with noise in the lead out groove. The lead out will be nice and long. Record it at the highest gain your phono pre-amp offers. In the computer, normalize to 99%. Now, run a frequency analysis on it and tell me what your results are. I think then you'll get what I'm saying.

[Bahndahn]

Yes indeed, amplification of any noise to a 'normal' audio level will allow you to measure its frequency components, it most often resembles standard pale-color [white, pink] noise. There is noise in every analog signal, this includes all various media, cables, preamps, amps, and transducers. The same test could be done with nothing plugged into your soundcard's preamps, with their level potentiometers turned to anything above their minimum.

This inherent noise in the medium is commonly called noise floor.

Im not looking to single you out or 'lecture', I'm just sharing what I have been taught for the general well-being of our communal knowledge! Thats why we come here to speak. There is something to learn from everyone; people are learning things as they read this forum that we may already know and assume as common knowledge.

[EpicenterBryan]

Wow, some heated discussion around here in my absence.
How about if you guys check out test cut #2 using the response curves from my previous post.

I'm pretty happy with this round of testing.

Oh and by the way, that Alesis DIO2 was advertised as being 24 bit, up to 48K sample rate. But buried in the manual for instillation on an XP system (my ProTools TDM Mix system is XP) it says to use the standard USB sound card driver which is 16 bit at 44.1K period. Real men do not read instructions, we read the sales lit. But I guess saying it was compatible with XP wasn't misleading had I read the manual. Argh.