Position Encoding.

Mostly it means that you can save on the "appropriate start bits and parity".

I don't think you could do this with BCH -- I'm pretty sure that the end of that road is more or less what I'm proposing.

--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
 Click to see the full signature
Reply to
Tim Wescott
Loading thread data ...

can someone explain more clearly to me what centration error is? does it have to do that the "pickups" are not placed at exactly 90 degrees (or 120 degrees for three-phase sensors)? or is it because the hole in the disk was not precisely in/at the center? are these two conjectured causes roughly equivalent?

if so, or for whatever source, ain't the errors totally self-canceling in the long term (sorta like our 50 or 60 Hz time reference we get from our handy nearby power socket)?

i'm trying to understand the nature of the error and what kind of error signal it adds to position or any derivative w.r.t. time thereof. i'm wondering how bad can it be if the apparent position signal is smoothed a little. that should kill most of it. i haven't followed every subthread of this thread (Tim, your thread exploded). so maybe it's old news.

tnx,

r b-j

Reply to
robert bristow-johnson

If I'm not completely out to lunch, centration error is due to the runout of the tracks with respect to the rotational axis of the disk. Points

180 degrees apart will always be 180 degrees apart, but within that there'll be a sinusoidal position error that is worse the more off-center the tracks are.
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
 Click to see the full signature
Reply to
Tim Wescott

Well, I mentioned BCH because it will correct code words in any rotation. We've used it that way recently.

It would be interesting to come up with a set of 2^N such codewords, correcting an N-bit payload (though the BCH data word probably needs to be slightly larger), that share the maximum bit-overlap in a cycle, consistent with the error correction plan you decide.

Overlapping Gray codes is another avenue to consider, just need to figure out how to make the result self-clocking.

Clifford Heath.

Reply to
Clifford Heath

What do you mean by "passive"?

It's an absolute encoder. It doesn't seem to be to make any difference what the code is. Reading a single position, you can still get the wrong code at the edges. You've added in a linear error of the read "head" to your encoder error (in addition to the run-out error).

But a wobbly shaft adds even more error.

Reply to
krw

(snip, someone wrote)

I believe the latter.

I believe that they are very different. In DSP terms, one is more noticable at high frequencies, one at low frequencies.

In many cases, one or the other can likely mostly be ignored.

With the two sensors appropriately phased, you get four transitions for each spoke on the wheel. It might be that you only need one or two of those, the others being required to get the direction right.

While you usually have the ability to move the sensors to get close to the 90 degree difference, the 180 degree depends on the width of the spokes and the beam size of the sensor, which might not be easy to adjust.

Off-center is mostly low frequency, and will be most noticable for larger rotations especially rotations of near 90 and 270 degrees.

-- glen

Reply to
glen herrmannsfeldt

(snip, someone wrote)

(snip)

I am not sure how you are counting your bits, but the proposed system avoids the need for start bits and finding word boundaries. You find the absolute position sooner, though that may or may not matter.

Using the usual asynchronous communications protocols with start and stop bits, and if you allow starting out of frame (plug in the connector while data is being transmitted) it takes some time to synch up.

One that I suggested previously has N stop bits before the start bit, followed by the N data bits. On the average, you go about 1.5*N bit positions to get the position.

-- glen

Reply to
glen herrmannsfeldt

t
n
.
e

A similar search on "John Larkin" pulls up nothing that looks as if it's yours.

It would be an unremarkable performance for an academic, but I'm not an academic, and I publish in peer-reviewed journals for my own amusement.

My father - "A R Sloman"- did a bit better, with one patent that has ben cited 19 times since it was granted in 1963, and two that have been cited nine times.

The median author publishes one paper, which is never cited, so I'm doing better than average - if not much better than average.

-- Bill Sloman, Nijmegen

Reply to
Bill Sloman

st

an

e

d.

he

And what "nail" would you think that that might be?

-- Bill Sloman, Nijmegen

Reply to
Bill Sloman

The idea is to not need to rotate at all, absolute position. The position is known before the rotor is even moved.

Same idea has having a pot fixed to the shaft but done at a digital level to capture position in precise indexes. You can take a reading at any time.

We have 2 different vender of digital string pots we use, one uses an analog photo disc and has very smooth transitions.

The other is a gray code type encode disc that is 10 bit, not very good but it gets the job done. That one has what looks like a processor encoder with a PWM output into a unity amplifier. This unit will produce steps in your DC Out, but not very big ones. It does have smoothing compensation when the string is on the move to give you a clean looking ramp.

Both units have a zero/offset and a spanned.. Not something you'd want to use in precise movements since it only outputs analog in the end and most likely drifts a bit.

Jamie

Reply to
Jamie

What's all this "citing" nonsense? Who cares if a bunch of doofusses are citing one another and keeping score? It's about as important as Twitter. Pathetic.

I must have about 10,000 purchase orders by now, from serious scientific and aerospace companies. Stuff is being built, tested, and shipped. It works. That's real.

John

Reply to
John Larkin

--
Because a 555 wouldn't be suited for that application.
Apples and oranges.
Reply to
John Fields

The error due to runout arises in part from the radius being longer than intended during half the revolution and shorter on the other half. The total error can be expressed as a Taylor series. The method, which unfortunately won't work with the coded system you want, effectively eliminates the first (and by far the dominant) term.

Projecting a part of the track onto the other part 180 degrees away through a simple lens provides a mask that moves counter to the disk itself, moving in the opposite direction. If the disk is eccentric, the mask will move slowly when the disk moves quickly and vice versa.

There are (or were) commercial products made this way. A wedge in the optical path moves part of the image to provide the second phase.

Jerry

--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply to
Jerry Avins

Imagine a record (platter) positioned off center on the turntable. The result is "wow". If a constant frequency is recorded --as on an encoder disk) the frequency out will vary. Counting cycles will not be the same as counting angles. I won't belabor the description, but let me know if I was too brief.

Jerry

--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply to
Jerry Avins

...

That's what my patents are about. However you implement it -- there are several ways -- the up count must occur at exactly the same angular location as the down count. The usual arrangement results in up counts that occur half way between down counts. That almost guarantees trouble.

Jerry

--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply to
Jerry Avins

known before the rotor is even moved.

capture position in precise indexes. You can take a reading at any time.

gets the job done. That one has what looks like a processor

in your DC Out, but not very big ones. It does have smoothing compensation when the string is on the move to give you a clean looking ramp.

in precise movements since it only outputs analog in the end and

Jamie,

I don't know that thread you're replying to, but apparently it's not the one that starts with this:

Clifford Heath.

Reply to
Clifford Heath

What's the usual arrangement?

I always use an incremental decoder that counts up (or down) on every transition of the input, such that the bottom two bits of my decoded count match the gray-decoded state of the encoder.

That way the shaft can vibrate all day across one bit transition, and never lose count.

00 0000 01 0001 11 0010 10 0011 00 0100 01 0101 11 0110 01 0101 00 0100

etc.

--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
 Click to see the full signature
Reply to
Tim Wescott

most

e an

ute

ded.

a

the

g

It's how academics count coup.

In universities, getting jobs and getting promoted depends on your academic reputation, and the number of times your best papers get cited is generally seens as a pretty good guide to how good the papers are. The people involved aren't often doofusses, and it matters to them a lot more than Twitter does to its addicts.

Or - to be more precise - it is a reality that John Larkin understands. There's a lot more reality out there that you don't understand.

-- Bill Sloman, Nijmegen

Reply to
Bill Sloman

Because the 555 isn't suited to all that many applications any more. Apples and a lemon.

Not to hand.

up

It was back in 1974, and the specification proved to be unreasonably stringent. The 555 dropped out very early in the search.

Otherwise I would have used it ...

In the same way that you affection for the chip precludes any rational appreciation of its defects?

You may want to think so.

Your perception of my status is about as realiable as your perception of your own, and you're still posting.

-- Bill sloman, Nijmegen

Reply to
Bill Sloman

Well that explains why we have a lot of mentally incapacitated free loading students, in school now. Thank you for pointing that out.

A perfect example.

Jamie

Reply to
Jamie

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.