Corian Turntable Fun

[valveitude]

Hi Sheldon,

Not as long as you might think. Air slowly diffuses out of tires (or tyres) and your construct has a much higher surface to volume ratio. Also, silicone sealant has a particularly high permeability (diffusion rate x gas solubility). You could lift the unit off the feet when you are not using it to maintain the air volume in your pads.

You may end up being correct, BUT, its's not just the silicon holding the pressure. At the top, other than a 1/8" wide silicon bead a 1/2" deep, its sealed with a 1/4" copper disc. At the bottom, other than a thin layer of silicon sanwitched between the foot plate and the plinth, a 1/2" thick piece of arcylic ( Corian) is sealing it. The walls of the pod cavities are a 1/8" layer of silicon over laminated Corian and metal.

I am not saying the pressure won't diffuse...eventually. I do believe however it will take a l-o-n-g time. Exactly how long remains to be seen (I haven't found any studies regarding bubble-wrap pneumatic springs). If I understand the physics (and I am not at all sure that I do), their needs to be a pressure differential in order for diffusion to take place. The pressure in the bubble-wrap is being equalized by the pressure of the silicon pushing back. The pressure of the silicon is being equalized by the Corian pushing back. To hedge my bet, I plan on coating the silicon in a layer of Plasti-Dip rubber before final assembly.

The patter & top plenth together weigh approx. 45 lbs.. The discs have around 5 sq. ins. of surface area each, for a total of 15 sq. in.. If the weight was distributed equally, that would be 3 psi. of pressure. I'm guessing the distribution is more like 4 psi. on the front feet and 2 psi. on the back foot. I don't think this will contribute greatly to diffusion.

From Sheldon,

edit: or you could tap a hole into the gas chamber and fit an airtight screw. When your pads flatten too much, you can lift the table and open the screw to let a new charge in.

From maxro (hello)

Or, you could glue in some valves from old bicycle inner tubes and inflate them with a very low volume pump, such as that used for mountain bike supension forks.

These suggestions go under the heading of "Plan B". If I turn out I am full of it regarding the pods longevity, I will plumb it. In fact, this was my first plan. It was while I was laying out how best to do this that I saw the bubble stuff laying in the corner.

Casey

[valveitude]

First things first.

HFGuy and Sheldon…Mea Culpa, Mea Culpa Mea-Maxima Culpa…bubble wrap was one of my less stellar ideas. You guys were right on with the diffusion problem…in retrospect, I think I was trying to “will” it to work. I will get back to the suspension later ( hint: I have acquired a 12V compressor, some small pneumatic solenoid valves , and a pressure sensor IC), but right now , I’ve been working on speed control.

My first task was to reduce the radius of the platter about .025” with the sanding drum made for the job. While profiling the plinths, I managed to undersize them slightly, causing the edge of platter to overhang. Once this was corrected I moved on to the strobe pattern.

My PSN (Platter Speed Neurosis) dictates that I must be able confirm speed at any time…a permanent strobe pattern is in order. In the “real” manufacturing world, a few lines of G-code fed into a 4 axis CNC mill would whip out a pattern in a couple of minutes. My world has a lathe and a modified drill press.

I spent the first day cobbling together a “dividing head”…



…actually this is the fourth revision. Originally the shaft was a piece of wood dowling...didn’t work out. Then the collar on the dividing wheel cracked, causing the spacing to shift…it was then glued. Then the set screw in the collar holding the platter to the shaft vibrated loose, causing the spacing to shift…it was glued. After each failed attempt, holes were filled, and drilling became harder.

Spacing was set by printing a pattern drawn with AutoCad, cutting it into strips, and taping end to end…



…if your incredibly lucky, the last piece will land on the line…I wasn’t. Getting it right entails shifting the “print scale” in .01% increments up and down until it does. Once I got the pattern to line up, I drilled a hole .04” deep after each graduation is lined up…



…and the break is set…



…this process is repeated for each hole (actually multiple times for some holes due to the failures). I intend to fill the holes with reflective paint on the final finish, but for now I painted the bottom of the platter white, then blackened it with a Sharpie felt pen for contrast…



Using the shop florescent lights as a “strobe” confirmed that sanding a profile on the drive pulley lowered the speed slightly. Not to worry, The speed control would fix it.

I was trying to decide which approach to take for signal generation for the AC supply when I ran across the Maxim MAX29x series of switched capacitor filter. This slick chip is a 8-pole low pass filter with the corner frequency determined by its clock frequency. You feed it a clock 100X the corner frequency, 6000 Hz for a corner of 60Hz for example. By feeding it a square wave of the desired frequency, you get a low distortion sine wave of stable amplitude. It also has a built in op amp that can be set up for an additional 2-poles of filtering.

Here is the basic logic configuration feeding it, it is an early draft and doesn’t show the op amp filter…



A signal source 400X the desired frequency (more on that in a bit) feeds both a 4013 flip-flop, and U2, a 4017 decade counter. The 4013 is configured as a divide-by-4 providing the 100X clock for the filter. U2 is set up as a divide-by-ten, and feeds U3 set up as a divide-by-5. This composite divide-by-50 pair provides a clock 8X the target frequency. Up to this point could have been simplified if all I wanted was one signal…but I want a two phase supply with a 90 deg. shift between them to drive a synchronous AC motor without a run capacitor. I needed the 8X for this part, the final 4017 is set up as a divide-by-4. Since each of the four outputs are high 25% of the time..in other words 90 deg. apart…all I need to do to get my 2 phases, is to take the signal from 2 neighboring outputs. I now have 2 clocks 2X times the target. These feed a final flip-flop to get my final frequency into the MAX291, and two clean sine waves 90 deg. apart are ready to feed the power amp. Besides eliminating the need for different run caps for the different speeds, the “cleaning” effect of the dividers means that the only thing you have to worry about with the oscillator is frequency stability. Wave shape , distortion etc. doesn’t matter. This broadens the oscillator options significantly.

For my oscillator, I’m using a Advanced Linear Devices ALD555 set up as a astable multivibrator. The ALD555 only needs one resistor and one capacitor for this, making changing the frequency easy.

For the strobe, I built The Vinyl Engine's strobe circuit (is this the same as dice45’s?). The construction is perfboard and hot glue point-to-point…



…Everything to the right of the reed relays is the divider/filter block. The wiring was very simple up to the 2-pole analog filter built around the MAX291 op amp…it got pretty “busy” there. In the middle is the oscillator, the relays pull in different values of timing resistors, changing the speed. Not seen is the trim pots on the end of the pigtail plugged in below the relays. The little circuit to the left is the strobe.

I played around with a couple of options for the power amp before settling on the TA8215 , a 18W Chn. BTL amp designed for car stereos. I built the amp straight out of the datasheet. This drives two 6.3V transformer primaries to get the line levels needed for the motor…



…This isn’t the final configuration. I set it up all together for testing, when the circuit(s) are finalized it will be “chunked” up and different sections will be different places.

All of that for this…



BOO-YAH !!!

Unfortunately the picture doesn’t do it justice . The dots are bright and clear. Is it perfect? No. The strobe pattern “walks” back and forth about 1mm per revolution, a product of the printer used for the pattern. Both my bride and my brother had to strain to see what I was talking about..it’s minor..but after the Herculean effort I wouldn’t mind perfect. I’ll wait to see if it’s bugging me in a couple of weeks before I investigate a re-do.

I’ll be experimenting with drive voltages, and oscillator configurations for the next couple of days…I’ll keep ya posted.

Casey Brown

[Sheldon]

Originally posted by valveitude
HFGuy and Sheldon…Mea Culpa, Mea Culpa Mea-Maxima Culpa…bubble wrap was one of my less stellar ideas. You guys were right on with the diffusion problem…in retrospect, I think I was trying to “will” it to work. I will get back to the suspension later ( hint: I have acquired a 12V compressor, some small pneumatic solenoid valves , and a pressure sensor IC),Casey Brown

I think I know where you are going with that, but maybe something simpler will do, and with your fabrication skills will be easy. Just use a piston with a trapped air charge. If you use essentially an oil bearing with a small guage O-ring seal to retain the oil, (obviously needs reasonably tight clearance between piston and chamber), the actual surface area of the O-ring can be quite small and the diffusion rate low enough to be practical. You could install a valve and maybe just use a syringe to periodically charge the chambers. It's not pressure you care about, as that's a funtion of piston area, but volume. The limitation is that this will give good vertical isolation (where I guess you need it most), but not much in the lateral plane.

As for your motor control, now that's some serious DIY for sure. But would we expect any less at this point?

Sheldon

[Nordic]

Hey, instead of printing all those little squares and cutting them out, why not make a long string on the PC, print it out and use that.. even if you just have A4 paper you would still have far fewer segments, and less errors...

[valveitude]

Hi Sheldon,

That’s an interesting idea. Sort of, but not quite, what I was trying to do earlier. Instead of a piston proper, my rubber mounted discs acted like a low travel piston, and the bubble wrap was my (admittedly lousy) trapped air. I'll have to give it some more thought.
The scheme I'm considering now is also similar. In this case the trapped air would be adjusted automatically. The only “hard” part will the control circuitry.

Hi Nordic,

Hey, instead of printing all those little squares and cutting them out, why not make a long string on the PC, print it out and use that.. even if you just have A4 paper you would still have far fewer segments, and less errors...

Look at the pictures a little closer…this is exactly what I have done. Looking back, it’s actually pretty amazing the pattern is as accurate as it is. Consider this, if each segment is off only .001”, then the whole thing would be off .216, or just shy of ¼”, for the 33.33 rpm band (216 marks). The fact that I was able to jigger it around with the “print scale” to fit at all using a cheapo Cannon printer is pretty incredible.

I have concluded that what I got with this method is the best I could hope for. If I decide I can’t take the very minor pattern wobble, I will either a) complete the project I started several years ago, but never finished, of building a “real” dividing head, b) buy a “real” dividing head, or c)farm out the job to a guy with a 4-axis CNC mill. The last option would yield the best results, but I would have to forever live with the fact that I “choked”.

Casey

[Sheldon]

Originally posted by valveitude
Hi Sheldon,

That’s an interesting idea. Sort of, but not quite, what I was trying to do earlier. Instead of a piston proper, my rubber mounted discs acted like a low travel piston, and the bubble wrap was my (admittedly lousy) trapped air. I'll have to give it some more thought.
The scheme I'm considering now is also similar. In this case the trapped air would be adjusted automatically. The only “hard” part will the control circuitry.

Hi Nordic,



Look at the pictures a little closer…this is exactly what I have done. Looking back, it’s actually pretty amazing the pattern is as accurate as it is. Consider this, if each segment is off only .001”, then the whole thing would be off .216, or just shy of ¼”, for the 33.33 rpm band (216 marks). The fact that I was able to jigger it around with the “print scale” to fit at all using a cheapo Cannon printer is pretty incredible.

I have concluded that what I got with this method is the best I could hope for. If I decide I can’t take the very minor pattern wobble, I will either a) complete the project I started several years ago, but never finished, of building a “real” dividing head, b) buy a “real” dividing head, or c)farm out the job to a guy with a 4-axis CNC mill. The last option would yield the best results, but I would have to forever live with the fact that I “choked”.

Casey

I think the air springs might work fine. As I said, it's volume you need to control, not pressure. You could have position sensors and add compressed air to reach a preset level, but I don't think you need to do that. With the above method, leakage will be slow - think nitrogen filled shock absorbers. In fact, the junk yard may have the parts you need. The air springs will be just like linear mechanical springs. The spring rate will depend on the length, i.e., longer springs - softer, shorter springs - harder.

As for the encoder, why not just buy a disc type and slap it on the bottom? Then you don't have to worry about the precise diameter of your platter. Or solve it in "software" - longer time constant? Won't make the visible wobble go away, but actual platter speed could be correct.

Sheldon

[valveitude]

Sheldon,

I think the air springs might work fine. As I said, it's volume you need to control, not pressure. You could have position sensors and add compressed air to reach a preset level, but I don't think you need to do that. With the above method, leakage will be slow - think nitrogen filled shock absorbers. In fact, the junk yard may have the parts you need. The air springs will be just like linear mechanical springs. The spring rate will depend on the length, i.e., longer springs - softer, shorter springs - harder.

The real advantage to me with the compressor/valve route is the easy tunability. I still am considering your approach though.

As for the encoder, why not just buy a disc type and slap it on the bottom? Then you don't have to worry about the precise diameter of your platter. Or solve it in "software" - longer time constant? Won't make the visible wobble go away, but actual platter speed could be correct.

At this point I am not using any encoding. The speed control is "open-loop". My experience with closed loop systems has been constant "hunting" of the operating point.

The "wobble" I've got is purely cosmetic...it's so slight that it doesn't interfere with adjusting the speed.

Casey

[AuroraB]

The original "Platine Verdier" used leaky air filled chambers, with small loudspeaker surrounds as gaskets. There was a spring inside with a screw adjustments for levelling. Then there was a small valve for adjusting the damping of the spring......
definately a possible route to follow for those so inclined

[Sheldon]

Originally posted by valveitude
The real advantage to me with the compressor/valve route is the easy tunability. I still am considering your approach though.

I'm not so sure about how easy that's going to be. Of course, I may be imagining something completely different from what you have planned. I'm thinking along the lines of three separate and unconnected pistons. You'd use an amount of air (or argon, with likely a lower diffusion rate) in each to level the table and to give the spring rates you want. The friction of the o-ring and the viscosity of the oil would control the dampening rate. Again the pressure is constant (function of weight and piston cross section area), so I don't see where a pressure sensor fits in.

Originally posted by valveitude
The "wobble" I've got is purely cosmetic...it's so slight that it doesn't interfere with adjusting the speed.

Ok, non functional, very slight wobble. So we're just gilding the lily eh? Easier to put the strobe led back out of the normal viewing range and just slap anyone who looks too closely.

Sheldon

[valveitude]

Hello AuroraB,

The original "Platine Verdier" used leaky air filled chambers, with small loudspeaker surrounds as gaskets. There was a spring inside with a screw adjustments for levelling. Then there was a small valve for adjusting the damping of the spring......

Hmm..interesting. I'll do a Google search and see what other info is out there...thanks for the tip.

Sheldon,

I'm not so sure about how easy that's going to be. Of course, I may be imagining something completely different from what you have planned.

I'm not planning anything too fancy, basically a pressure sensor controlling a valve(s) and a small compressor that "tops off" the pressure in the cavities when it drops to a set point. It will be connected to the motor control so that it can't run while playing a album. The finer points aren't worked out yet (my speed control has been monopolizing my two remaining active synapsis).

Again the pressure is constant (function of weight and piston cross section area), so I don't see where a pressure sensor fits in.

Compliance...the compliance of an air spring is determined by volume and pressure. The rough tuning will be by varying the volume, fine tuning by varying the pressure. Once I get it right, the sensor/valve setup should be able to maintain the desired pressure within a couple tenths of a psi.

Ok, non functional, very slight wobble. So we're just gilding the lily eh? Easier to put the strobe led back out of the normal viewing range and just slap anyone who looks too closely.

Yup..if that lily gets much more gilding there won't be much of a lily left. I'll have the strobe front and center...those with a keen eye might notice...I'll just slap 'em if they say anything.

Casey