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

[EpicenterBryan]

I'll chime in just so others have some background.
The iPads I'm using and the app, can only run at a CD sampling rate of 44.1Khz. My old school Protools gear is only up to 48khz. Modern Protools gear running with software and such can go to 192khz sample rates. DSP's are also involved. The sample rate is mainly limited to stay compatible with 48khz sample rate gear.

Todd, you and I still think in a linear fashion, like things have to be done in sequence. They don't have to be in sequence. You know... Do this, then that, calculate the other thing and do something else. DSP's don't work like that. The tasks can be done in parallel and the results are obtained literally in clock cycles.

Even in my old school Protools TDM system, with all the complicated tasks like processing real-time reverb is not done via the processor and operating system, it's done on DSP cards with many specialized processors doing their own thing. The processor and operating system just puts things on the screen. That's it.

Let's look at a crazy application like mining bitcoins. A few years ago, they used the parallel pipeline processors on high end video rendering cards to do the calculations. Later, specialized chips that were tailored to do the calculations in parallel came out - all in an effort to do it faster and have more processors running in parallel at the same time thus producing more theoretical coins. It was an arms race since as more bitcoins were mined, it took more effort (by design) to mine aditional coins until the theoretical limit is reached. It's kind of like "peak oil" is supposed to work. After the easy "dig a hole and oils spurts up" is over, it takes more effort and costs more to produce the same amount.

Hummmmmm. Did I loose you here?

B

[opcode66]

Parallel processing is not particularly useful for a linear stream of real-time data. Having multiple thread or processes on separate processors would be great if you had chunks of feedback data at a time. So, each could be doing part of the process like analyzing eq, detecting resonance points and controlling gain. However, you are getting the feedback data linearly and in real-time. Not in blocks. So, paralleling in this particular cases saves little. Did I lose you?

Parallel processing is good for splitting up the tasks on an enormous chunk of data. Or, for increasing the perceived responsiveness of an application. If I wanted to make bots that analyze the market (and have), each one is going through different tickers. Or, the same ticker looking for different things. This is a great example of paralleling.

I've been a developer for multi-threaded and parallel processing business applications for years. I've been a developer for 16 years.

[dubcutter89]

The take away is being 1 sample off can mess up the phase and your feedback at high frequencies! Wow.

We had the same experience with a cheapo DAC which we used for generating complex signals which had to be accurate in amplitude and phase. I found it by using and analog scope in XY mode showing the typical circles/ellipses. Our programming guy then just put a 1 sample delay into the "faster" channel and the problem was gone.

We also checked some other Interfaces and found some really weird stuff that was going on. Loosing samples, Flipping Channels... This is what is left in modern music electronics since the glory analog era. Everything has to be small and cheap, nobody cares about quality and most people will never recognize....

BTW, what cutter preamp did you use? is it your selfmade "phono" pre with caruso?

Lukas

[markrob]

Hi Todd,

Total nonsense. You may know PIC assembly, but you have no experience with DSP or feedback control systems. I'm done debating this with you.

Mark

[opcode66]

Prove me wrong. You can certainly make this system. But, it wont be as accurate as analog. I maintain this assertion until you build a system to prove otherwise. What you have here is a real-time problem.

[EpicenterBryan]

We also checked some other Interfaces and found some really weird stuff that was going on. Loosing samples, Flipping Channels... This is what is left in modern music electronics since the glory analog era. Everything has to be small and cheap, nobody cares about quality and most people will never recognize....

That is really interesting. I ran into a similar "Channel Flipping" issue, not only with this Transfer App, but also with a free RTA app. When I saw that with the RTA app, I assumed the free version did that to poke at you to buy the full version.

When I ran into that with the Transfer app and I had seen it before with the RTA app, it seamed to happen after I took a screen capture, then E-mailed it to myself. Then I found a note that the iPad should be in "airplane mode" and then I made sure I had that set when I used the app. I thought I was safe to then be able to switch apps to "cue" an e-mail of a screen shot but also ran into an occasional channel flip. Now I think it's actually related to switching apps so I never brought it up with the developer. It seams to work fine as long as it's the only app running and you use it until done and close it out.

I have had one or two things happen that were so strange I rebooted the iPad and all was fine. Crazy things like a meter showing a signal that's not there. That makes you question the measurement system for sure. Since I have had only the one app running and have not switched screens or apps everything seams to work so that's what I do now.

Bryan

[EpicenterBryan]

Another interesting weekend.

I decided to try a few things in an attempt to tame the drivers. I picked up another phono cartridge (actually two Sure M95ED's) for use on my driver test bed since I still have a few extra stylus for that model.

First off, I took a chart of the unloaded driver response with a carbon fiber rod with the phono cartridge as pickup at about 6" away from the driver. This is a "No Load" chart. I added two support guides next to the pickup. The guides have rubber bushes that allow the rod to move back and forth. I also printed an adjustable cartridge mount and shims. The carbon fiber rod has a nick cut on the bottom for the stylus to fit into so it does not need to be glued. That is important, so I can actually remove the cartridge and test the response... but for now I want to share what I have.

The reason I'm going into this, is I want to go over the mods I made to the driver to compare.

IMG_3542.JPG

And here is the frequency response. Again, this is un-loaded. It looks like what we have seen before except the HF is much higher. Also, the resonant frequency is about 1/2 what we have seen before.

IMG_0043.JPG

Now for the interesting stuff. I decided to try to damp the voice coil and spring. My thought was to try to place the spring and voice coil into damped compression in each direction of movement. To do that on the inside, I made two doughnuts of 1/8" Sorbathane glued together and glued them to the internal magnet. That forced the spring out about 1.5mm.

IMG_3534.JPG

IMG_3535.JPG

CONTINUED ON NEXT POST:
.
.

[EpicenterBryan]

To put the outside of the driver in an equal but opposite compression using the same thickness (2 layers of 1/8" Sorbathane) I made a plastic spacer ring, and an aluminum bulkhead plate. The Sorbathane was glued to the inside of the plate and once installed it compressed the outer Sorbathane by the same 1.5mm amount to keep the coil and spring centered. So now it's damped in both directions.

IMG_3536.JPG

IMG_3537.JPG

And here is the same set up with the damped driver:

IMG_3547.JPG

And here is the result:

IMG_0045.JPG

At this point I'm not sure what I'm looking at.
The phase plot is still messed up, and I'm shocked that the HF (even though it is out of phase) is that much different than as measured in the head. I guess it's possible that the Audio Technica cartridge was damaged when I broke the stylus loose eary on to do other testing... and the HF is actually higher. I just don't know. I'll pull this cartridge and do a sweep with a test disc to make sure the HF is real...

Comments anyone (Especially Mark).

B

[dimi751]

Hi Bryan

I'm enjoying your post I'm learning a lot. I will like to build a cutter head similer to yours
But I don't know what ampli I need to run the surface mount speaker your are using
In the picture your posted, could you recommend a ampli for these speakers I'm guessing it does not need anything too much power.

Dimi

[EpicenterBryan]

Well, there may not be any updates for 2 weeks.
I'll be doing a bit of light reading while on vacation including some stuff on opamps and feedback among other things.

So far I've packed the important stuff - snorkel gear and the AES Disc Recording Volume One! I haven't read anything yet but in flipping through it does look really interesting.

IMG_3549.JPG

There are a few circuits I've started looking at to try to tame the phase issues. And I've been following an interesting thread having to do with phase issues with phono cartridges. I think some of what I'm seeing relates to use of a phono cartridge for phase measurement (and feedback).

I'll post if I learn something, but I won't be able to test a darn thing for 2 full weeks. Argh!

Bryan

[dimi751]

Enjoy your break Bryan nude snorkling is popular this time of year

[opcode66]

I have two copies of the red book. I bring one with me on all my travels. I've re-read a number of sections. There is always something I've missed... A fantastic resource.

[EpicenterBryan]

Enjoy your break Bryan nude snorkling is popular this time of year

Too funny! Shorts and shirts have since been packed. I'm too old and modest to go "au naturel". That is best left for the young and well ... (a word here that rhymes with young that starts with an "h").

B

[EpicenterBryan]

I have two copies of the red book. I bring one with me on all my travels. I've re-read a number of sections. There is always something I've missed... A fantastic resource.

Todd, you have mentioned this book several times and it was because of your recommendation that I bought it. Thanks again! It will be fun to actually have time to go through it. Life you know....

B

[dimi751]

Ha Bryan lol your funny

[dimi751]

image.jpg

Enjoy swimming Bryan !

[EpicenterBryan]

I'm back, well rested, and have no shark bites.
I had zero time to read the AES book I brought but learned quite a bit from this other book from 1974-1979 (this edition was 1979). I'll have to share a few things in another thread... That's my wife doing the "peace" photo bomb.

IMG_3573.JPG

It was nice to get away, and to clear my head... Now that I'm back I've had 2 days to try / measure a few things that I thought about for the last 2 weeks and couldn't try or check out.

1). I came up with a cool idea to be able to make phase changes without effecting amplitude. It's a separate feedback loop that is useful in the overall frequency correcting feedback loop by injecting phase changes as needed to help the main feedback loop maintain stability over a wider range. My initial test is promising. No more details yet. Other issues exist that I need to look at and relate to the issues below...

2). I was puzzled by the phase issues I plotted at high frequency and wondered if it was real. It is real in that it exists with the wiring and op-amps I used. It is not real in that there are issues I did not realize needed to be addressed...and those issues can be fixed. It all goes back to using a phono cartridge as a feedback device. The phono cartridge coils have an inductance, and a series resistance. The wires from it to the preamp also have a capacitance. After measuring the actual capacitance of the wires from the feedback phono cartridge to the preamp and doing a SPICE simulation on the data - magically there is 167 degrees of phase shift from about 1Khz to 20Khz which explains why the loop could not be closed at high frequencies. More on this later, as I try out some other stuff.

3). The next thing I'll try is to use twisted pair wire from the cartridge to a differential amplifier. I'll prototype something similar to what Flo does on the Caruso board. Basically, it will be a differential microphone style preamp so noise common to the twisted pair is rejected and issues related to capacitance of the wires used is less relevant.

4). I will also play with modifying the drivers but only on my test bed, and if things work out will be used on the next version of the head. This head can not be modified...

Below is a shot of a driver that I drilled the rivets out to remove the stainless spring. Threaded into the spring is a very heavy adapter that makes it easy to attach different driver links. Next to it is a plastic driver spring with a built in collete that I can glue a carbon fiber rod into for testing. This plastic spring will be attached to the coil initially with nylon screws and nuts. Later it can be glued without the extra mass.

IMG_3687.JPG

That's it for tonight.

Bryan

[EpicenterBryan]

Here is a quick update.

First off, I mentioned some issues and things I was going to try. I have tried several things, and worked out a few issues.

1). the Cool idea I had for changing phase without changing amplitude is more complicated than I thought. I'll touch on that later in another post.
2). The phase shift is real even after addressing what I thought was the root cause. But it is better now. Let's talk about what was discovered....

As I eluded to, the cables to the phono preamp have a capacitance. I had no idea what I was using was so high. Take a look at this... The long cable (about 2 meters) from the phono cartridge to the preamp turned out to be 940PF, and the cables internal to the box (that I was too lazy to shorten) was an additional 140pf for a grand total of 1080pf.

Phono Preamp in cables HIGH C.jpg

The ones I replaced them with are 105PF external (same 2 meter length), and shorter 22pf segments internal - For a grand total of 127pf. That's a big difference! That's all about HF.

Phono Preamp Cables REPLACED.jpg

There are also issues related to the inductance and DC resistance of cartridges and their ideal loading that I have not totally worked out. It turns out the Sure M95ED cart I have for use on the test bed has specs of of 650 mlllihenrres, and 1550 ohms resistance. They specify a load resistance of 47K ohm, and 400 to 500PF of load capacitance. The Audio Technica AT3600 used for feedback on my head, (and I also purchased one for the test bed) is a bit more of a mystery. They specify the same 47K ohm load, but do not specify the capacitance load nor the inductance nor resistance numbers. I called Audio Technica last week and explained my situation and they were very interested in the application and said they would get back to me but never did....

I can only say that the preamp I have has switchable load capacitance and switchable load resistance. I have played with all the settings with both cartridges, and saw little difference but some of that was before making the wiring changes. So I'll need to try things again fresh now that the wiring has changed...

3). I mentioned I would play with twisted pair and something like Flo's FB input and I did. I got the total wiring capacitance for the twisted pair down to 102pf using Cat6 twisted pair network cable. I realized I had a tap I installed on his board for VU's that I wasn't using but didn't get what I expected. Looking at the circuit, the pseudo differential resistors were 1K, and knowing that the recommended load for both of the phono cartridges I am playing with is 47K, I looked at the spec sheet for that chip and found the maximum recommended resistor value is 10k. So I figure it may not be a good match for either of my test phono cartridges. But I have not spend any time looking into it.

4). I mentioned modifying the driver. Below is a plot. This is after replacing the cables to the preamp and internal to the preamp. Again, this is with no load or damping on the driver - using an Audio Technica AT3600 cart.
The plastic springs are 0.75mm thick ABS plastic.

GS plastic spring AT3600 revised cables.JPG

IMG_3709.JPG

That all the time I have for tonight.

Bryan

[opcode66]

Good progress as usual, Bryan. I can't wait to get myself back up and running.

[EpicenterBryan]

I wanted to try dampening the plastic spring with the same idea from the metal spring attempt.

One thing I discovered when I took the driver apart was my spring was slightly too large near the mounting hole and was actually touching the mounting screw. I trimmed it this time to clear.

IMG_3731.JPG

Here is what the dampening looked like:

IMG_3735.JPG

And here is what it looked like with the outer pressure plate attached:

IMG_3748.JPG

And what did the frequency response look like?

IMG_0017.JPG

Better but still sucks.

Next, I'm going to replace the carbon fiber rod with non magnetic Stainless steel and I'm going to open the tolerances on the bushes and see if there is a difference.