New Technics SP10 motor controller specification

[Steerpike]

On the unit as supplied, if you try to slow it down using your hand, you can see the strobe marking move slightly, still in hard lock, but displaying a phase shift. Also the current goes up to nearly 1A from the 32V

Any servo system will display that behaviour; the limiting factor is how much torque you can get out of the motor to correct the disturbance, and by what degree. The motor is what it is - no amount of electronics is going to make it more powerful.

This is only a drawing - not a real oscillograph, but it makes me think I should see sine waves in real life.

[Dave Cawley]

"Any servo system will display that behaviour;" Yes I'm aware of that, just thought I would pass it on that it really did work as you would expect. I have seen the waveforms myself, it could be all my 3 decks are faulty?

To be honest, I think I'll give up on this one.

Regards

Dave

[Steerpike]

To be honest, I think I'll give up on this one.

We can't let you get away so easily.
Even if you give up on construction - keep posting your ideas!

That circuit fragment I copied up last post... it has two electrolytic capacitors C16 and C17. These are well known for drying up and becoming leaky, even in a handful of years. That may be a common fault in all SP10s after such a long time in service.

re: C for PICS.
Are there any free compilers that produce *good* hex code though? My limited experience so far was that the free compilers generate poor code - inefficient use of memory, many redundant op-codes, and slow or unpredicable timing. The code optimiser costs big money.

I kinda like ASM, because I like to know I have complete control over exactly what is going on... it can be so tedious to create though - and madening to modify a few weeks later when you (I) have forgotten how it was suppossed to work!

[Shaun]

Originally posted by Steerpike
re: C for PICS.
Are there any free compilers that produce *good* hex code though? My limited experience so far was that the free compilers generate poor code - inefficient use of memory, many redundant op-codes, and slow or unpredicable timing. The code optimiser costs big money.

I can ask about this. My suspicion is that currently memory space is less of a constraint than it was, say, 10 years ago. Even when I was writing in C, I found it useful to check the generated code (the assembler version), especially when the timing yielded unexpected results. I believe that the secret is to keep the interrupt routines simple.

Originally posted by Steerpike I kinda like ASM, because I like to know I have complete control over exactly what is going on... it can be so tedious to create though - and madening to modify a few weeks later when you (I) have forgotten how it was suppossed to work!

Then you have been naughty!!!

Did you neglect to document your code properly? ....adding comments and making flow charts? .... I think most of us are guilty of that....

[Steerpike]

Memory space is a bit of an irrational thing for me - something the young people of today know nothing about.
I suppose it's like people who grew up in 'The War' cannot waste food; I still think it's a sin to waste RAM, remembering a time when you had to get your whole program and OS into 16kB.

Am I showng my age? Bah that's nothing.. my first 'record player' had to be cranked every two sides, and the needle replaced just as often.

[Steerpike]

Things are progressing...
I've decided on a Wein bridge for the phase excitation oscillator, since it requires no inductors, and it gives a low distortion sinus. Don't be confused though - this sine wave is not the one that actually drives the motor, so low distortion maybe is not all that important.
Anyhoo.. it's an easy oscillator to make work, with everyday parts (like an NE5534, as I used here - a good use for the noisier ones that have no place in pre-amps!)

With that done, and an active full-wave peak detector to recover the signal from the phase pickup coils, this is the waveform that I get: oscillograph.

(now that's one of the three 6Hz or so signals that will actually drive the motor)

There is some distortion, which does NOT come from the detector circuit - it's a very linear detector topology. So I went searching for the source....
It turns out to be in the construction of the motor itself:
[It seems I can't put two picture in one post, so see next post for this picture]
take a close look at the copper-coloured reluctance ring: the 'top' and 'bottom' of the diamond shaped holes are flattened off, which puts a plateau on the lower limit of the magnetic coupling between the exciter coil and the phase pickup coil.
After AM demodulation, that shows up at a slight flattening off of the lower peak of the sine wave.

I suppose purist might even want to remove this ring and file it to a better shape???

I don't believe it will have any notable effect on the final system; the motor/platter inertia will damp most of it, and the negative feedback servo will reduce any residual non-linearity to imperceptible levels.

Next step, get 3 L165V power op amps. These will become the motor drive circuits. WAY simpler than actually building a discrete transitor power driver, if slightly more expensive (the L165V is us$3.50 at my supplier).

[Steerpike]

The SP10 motor reluctance ring, referred to in the previous message.

While staring at it I had another thought.... for the masochistic purists (puritans???) it would be possible to laser-print a more sinusoidally accurate black/white image, much like this string of diamonds (or even cut it out of paper, like those party festoons of figures holding hands!), and stick this printout/cutout on the outter periphery of the motor - where the brake band is normally. Then read it with three photo-diodes.

[Steerpike]

After more amplification & filtering. (3.5v p-p, 5.555Hz)

On my very fuzzy SP10 circuit diagram, it appears the motor drive voltage is 2.8V. Has anyone any idea what this goes up to when the motor is under heavy load - i.e., just short of the PLL losing lock?

[Dave Cawley]

.
I'm back!

I would just stop the bottom edge clipping, perfection and all that!.

Under near stall, but locked condition, the voltage stays the same but the drive current goes from about 100mA to nearly 1A.

The voltage is actually 28 volts. If I have managed to do this, here is a image of the drive waveform under load. How to put a thumbnail in would be good?



Regards

Dave
.

[Steerpike]

>>I would just stop the bottom edge clipping, perfection and all that!.

It isn't clipping - well not in the sence that amplifiers clip, due to transistor saturation, or running out of headroom.
You have a way to 'peak' the bottom of the wave?

What I'm getting does look rather cleaner than the 'real' SP10 signal you're measuring.
The broadening of the trace at the peaks is because the waveform fluctuates very slowly, and the pic is a very slow exposure.

28V ? Ok.. we can do that.

{Thanks for the number 28!}