Searching \ for '[EE] Linear potentiometer, noise suppression' in subject line. ()
Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=linear+potentiometer
Search entire site for: 'Linear potentiometer, noise suppression'.

Exact match. Not showing close matches.
'[EE] Linear potentiometer, noise suppression'
2008\11\17@212014 by

Hi, I have a linear potentiometer that gives out about 3v p-p noise when
moved. This noise must be suppressed so the signal can be read by
an 12bit 0-10v A/D input. Ideal would be if the noise could be suppressed
below 0.01v That might not be possible in real world? If not, I can live
with 0.03v

This is how its connected:
24vDC --- 12k resistor --- 5k lin. pot. --- 0vDC
Output from the 5k lin. pot goes into the A/D input (100kohm)

What would be the simplest way to smooth out the noise on the pot. output?
A cap (size?) in parallel with the pot's output?

TIA.

If your potentiometer is worn, it may be impossible.  Well, it'll smooth
out the "noise" but you could leave your pot at the 50% position and it
reads as 80% just because the static resistance is bad not just the
moving resistance.

Danny

Info wrote:
{Quote hidden}

Hi,
Its brand new. 100 million cycles is what the datasheet says.

-----Ursprungligt meddelande-----
If your potentiometer is worn, it may be impossible.  Well, it'll smooth
out the "noise" but you could leave your pot at the 50% position and it
reads as 80% just because the static resistance is bad not just the
moving resistance.

Danny

part 1 1015 bytes content-type:text/plain; charset="us-ascii" (decoded quoted-printable)

WHAT is is used for?
Servo position?
What is the travel time from one end to the other?
Double wiper contact or single?
The higher input impedance of A/D, the less change the series contact resistance will make when moving.
But if motorized linear pot, noise from motor?
Ray

On Tue, 18 Nov 2008 03:21:47 +0100, Info wrote:
{Quote hidden}

part 2 35 bytes content-type:text/plain; charset="us-ascii"
(decoded 7bit)

It measures the position of a pneumatic cylinder (or acually the
parts that the cyl. moves) I need to know where it stops.
Time for full stroke is about 0.1 sek

Wipers, no idea. It uses conductive plastic is about all i know
about the inner workings of it.

No motor, noise is only present when the pot. moves.
When the movement stops I get a nice crisp line on the scope.

-----Ursprungligt meddelande-----
WHAT is is used for?
Servo position?
What is the travel time from one end to the other?
Double wiper contact or single?
The higher input impedance of A/D, the less change the series contact
resistance will make when moving.
But if motorized linear pot, noise from motor?
Ray

> What would be the simplest way to smooth out the noise on the pot output?

Try a squirt of lubricant

If you don't need the measurement while it is moving, only stopped, treat it
like contact bounce and resample with delay until you get repeated values.
Perhaps a .11 sec delay, so you're sure to be post-movement.

-Skip

Info wrote:
{Quote hidden}

> No motor, noise is only present when the pot. moves.
> When the movement stops I get a nice crisp line on the scope.
>

I'd start with a 10uf or so cap and see how it behaves, increasing the
cap if it doesn't smooth it enough, decreasing the cap if it slows the
response too much.

I would also place some clamp diodes to make damn sure the output did
not go outside the range the ADC input can safely handle.

> No motor, noise is only present when the pot. moves.
> When the movement stops I get a nice crisp line on the scope

If you were doing this with a PIC to detect the end of an event of
indeterminate length -

Say the timer is set to rollover or reach a certain count after 0.5s. Any
noise (ie change) on an input resets the timer. Only when there's been
no noise for 0.5s will the timer complete its count

Alternative is to continuously poll the ADC looking for stability

This may be useable depending on the movement and timing patterns
of the cylinder

Thanxs guys, I'll give 'wait until it comes to a stop' a go
and also see what happens with a 10uF cap. to start with.

-----Ursprungligt meddelande-----
Från: piclist-bouncesmit.edu [piclist-bouncesmit.edu] För peter
green
> No motor, noise is only present when the pot. moves.
> When the movement stops I get a nice crisp line on the scope.
>

I'd start with a 10uf or so cap and see how it behaves, increasing the
cap if it doesn't smooth it enough, decreasing the cap if it slows the
response too much.

I would also place some clamp diodes to make damn sure the output did
not go outside the range the ADC input can safely handle.

Jinx wrote:
>> What would be the simplest way to smooth out the noise on the pot output?
>
> Try a squirt of lubricant
>

Or glue

-
M
> >> What would be the simplest way to smooth out the noise on the pot output?
> >
> > Try a squirt of lubricant
> >
>
> Or glue

Hehe - "Why, that's the most stable it's ever been. Thank you chemistry, thank you !"

I've heard Jamie Hyneman say the M5 shop motto is "With lubrication, we can do anything"

Would that glue be conductive, say with silver filler? :) :)

Jinx wrote:
>>>> What would be the simplest way to smooth out the noise on the pot output?
>>>>
>>> Try a squirt of lubricant
>>>
>>>
>> Or glue
>>
>
> Hehe - "Why, that's the most stable it's ever been. Thank you chemistry, thank you !"
>
> I've heard Jamie Hyneman say the M5 shop motto is "With lubrication, we can do anything"
>
>
Hehe, an 1uF capacitor did the trick nicely. From what I can tell with
the 'scope all the noise is gone.

Tho it still doesnt work as it should, the voltage across it jumps up/dn
even when it is still. Wonder if it is a crap quality linear pot.

-----Ursprungligt meddelande-----
Would that glue be conductive, say with silver filler? :) :)

Jinx wrote:
>>>> What would be the simplest way to smooth out the noise on the pot
output?
>>>>
>>> Try a squirt of lubricant
>>>
>> Or glue
>>
>
> Hehe - "Why, that's the most stable it's ever been. Thank you chemistry,
thank you !"
>
> I've heard Jamie Hyneman say the M5 shop motto is "With lubrication, we
can do anything"

> Tho it still doesnt work as it should, the voltage across it jumps up/dn
> even when it is still. Wonder if it is a crap quality linear pot.

ISTR you mentioned it's a new pot with a conductive plastic track. You
might expect poor performance like that from an old carbon pot. I wonder
if there's some contamination. Perhaps clean it out with meths / compressed
air. It seems presently it's unuseable so you've more to gain than lose

Is this a high value resistance that might be picking up electrical noise ?

Yes, its a brand new pot with a conductive plastic track
Warranty, and it's supposed to last 100 000 000 cycles.
I have three of those.

5 kohm, but I had to put a 12kohm resistor in series to
do a voltage divide to get it down from 0-24v to 0-10v.
Unfortunately it came more to 0-6v so perhaps it could
help to change that resistor to something lesser to get
more voltage over the lin-pot.

-----Ursprungligt meddelande-----
> Tho it still doesnt work as it should, the voltage across it jumps up/dn
> even when it is still. Wonder if it is a crap quality linear pot.

ISTR you mentioned it's a new pot with a conductive plastic track. You
might expect poor performance like that from an old carbon pot. I wonder
if there's some contamination. Perhaps clean it out with meths / compressed
air. It seems presently it's unuseable so you've more to gain than lose

Is this a high value resistance that might be picking up electrical noise ?

Info wrote:
> Yes, its a brand new pot with a conductive plastic track
> Warranty, and it's supposed to last 100 000 000 cycles.

Once again, 100M cycles to what state?  Just guaranteed that there's still a
knob to grab onto or some minimum linearity or what?  I'm rather skeptical
that this pot has 100M cycles of useful life as required by your circuit.

> 5 kohm, but I had to put a 12kohm resistor in series to
> do a voltage divide to get it down from 0-24v to 0-10v.

So you aren't keeping the wiper unloaded as I explained in a earlier post.
What you are doing is just asking for a scratchy signal.

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
>> Yes, its a brand new pot with a conductive plastic track
>> Warranty, and it's supposed to last 100 000 000 cycles.

>Once again, 100M cycles to what state?  Just guaranteed that there's still
a
>knob to grab onto or some minimum linearity or what?  I'm rather skeptical
>that this pot has 100M cycles of useful life as required by your circuit.

100 M cycles usefull life. Meaning it should be in good shape,
not fall apart. This is a lin.pot. for tool-machines.

>> 5 kohm, but I had to put a 12kohm resistor in series to
>> do a voltage divide to get it down from 0-24v to 0-10v.

>So you aren't keeping the wiper unloaded as I explained in a earlier post.
>What you are doing is just asking for a scratchy signal.

Its pretty much unloaded. The wiper sees a 100k ohm load from the A/D
with a 1uF cap in parallel to 0v to supress the noise.
24v --- 12k resistor ---- linear pot --- 0v

Info wrote:
>>> 5 kohm, but I had to put a 12kohm resistor in series to
>>> do a voltage divide to get it down from 0-24v to 0-10v.
>
>> So you aren't keeping the wiper unloaded as I explained in a earlier
>> post. What you are doing is just asking for a scratchy signal.
>
> Its pretty much unloaded. The wiper sees a 100k ohm load from the A/D
> with a 1uF cap in parallel to 0v to supress the noise.
> 24v --- 12k resistor ---- linear pot --- 0v

Ah, the 12Kohm resistor is in series with the *input* of the pot, not the
wiper.  Have you tried what I suggested in my first reply?

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
Info wrote:
>>> 5 kohm, but I had to put a 12kohm resistor in series to
>>> do a voltage divide to get it down from 0-24v to 0-10v.

> Its pretty much unloaded. The wiper sees a 100k ohm load from the A/D
> with a 1uF cap in parallel to 0v to supress the noise.
> 24v --- 12k resistor ---- linear pot --- 0v

Ah, the 12Kohm resistor is in series with the *input* of the pot, not the
wiper.  Have you tried what I suggested in my first reply?

Exactly. I wanted to get down to max 10v out of the pot.

No, I havent seen that reply. Dont know what have happend to it.
Can you copy-paste it?

Info wrote:
> No, I havent seen that reply. Dont know what have happend to it.
> Can you copy-paste it?

I was going to say something like "look in the archive, moron", but decided
to check first.  My original post isn't in there.  Now I remember getting
two bounces from the server a day or two ago, something about disk space.  I
guess my detailed response has been forever lost in the laborinth of the
internet.

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
> > Have you tried what I suggested in my first reply?
>
> No, I havent seen that reply. Dont know what have happend to it.

I didn't dee a reply from Olin either. Lost in transit somewhere I guess

> the voltage across it jumps up/dn even when it is still

How much does it jump ?

Is the jump the same wherever the wiper is ? ie should be nothing
at the earthy end. The pot is earthed properly ?

How clean is the 24V ?

What happens if you substitute a normal 5k rotary ?

> I was going to say something like "look in the archive, moron",

In this case anyway :-)

R

I think the problem is that the pot is flimsy inside and the slightest
vibration makes the wiper 'jump'. I tapped the cover lightly and the output

jumps an impossible amount, corresponding to several mm which it clearly
isnt moving.
So something else is moving/changing and the only thing it can be is
the wiper inside the pot. from what I can tell.

It jumps from the A/D conversion, as much as 16 deciamal. Translates to
about 0.1mm, thats expected. Problem comes if there is any mechanical
vibrations which makes it jump all over the place. Yesterday I noted
a 5mm jump.

The 24v 10A switched powersupply has 0.2v noise. I divide the voltage down
to 10v and the noise is 0.1v. The powersupply noise on the output (wiper)
goes down linear towards zero when the wiper gets closer to 0v side of the
pot.

When the 0.1uF filtercap is added, the 0.1v-powersupply noise is
practically gone on the pot's output. It also takes care of the
terrible noise the pot makes when it moves.

>jump the same wherever the wiper is
Hmm, if I am right about the flimsy inside it should be about the same
everywhere. I'll try to test that.

-----Ursprungligt meddelande-----
> the voltage across it jumps up/dn even when it is still

How much does it jump ?

Is the jump the same wherever the wiper is ? ie should be nothing
at the earthy end. The pot is earthed properly ?

How clean is the 24V ?

What happens if you substitute a normal 5k rotary ?

Info wrote:
> the only thing it can be is
> the wiper inside the pot.

LoL.  Because it's the only thing you can think of, it's the only thing it
has to be!!?

Think about the mechanism of a pot being scratchy.  What's more likely, that
a little vibration makes the wiper move sideways, or that it goes to high
impedence momentarily?

As I said before, keep the wiper unloaded and put a small cap on it to
"remember" the voltage when the wiper goes momentarily open.  As I also said
before, I would add a single R-C low pass filter after that.

This isn't one of those industrial position sensing pots with 360deg
rotation and two wipers, is it?

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
> tapped the cover lightly and the output jumps an impossible amount,
> corresponding to several mm which it clearly isnt moving

The other possibility is that one of the other terminals is damaged where
it connects to the track. Any vibration would cause electrical noise, just
like a dry joint does

True, but I have serveral and it woule be strange if all were broken
straight out of the box =)

-----Ursprungligt meddelande-----
> tapped the cover lightly and the output jumps an impossible amount,
> corresponding to several mm which it clearly isnt moving

The other possibility is that one of the other terminals is damaged where
it connects to the track. Any vibration would cause electrical noise, just
like a dry joint does

> Info wrote:
>> the only thing it can be is
>> the wiper inside the pot.

> LoL.  Because it's the only thing you can think of, it's the only thing it
> has to be!!?

That's puzzling.
Why would you think that?
It's not what he said.

You trimmed his post and thus are misquoting him.
Viz (just 5 more words)

>> So something else is moving/changing and the only thing it can be is
>> the wiper inside the pot. from what I can tell.

> Think about the mechanism of a pot being scratchy.  What's more likely,
> that
> a little vibration makes the wiper move sideways, or that it goes to high
> impedence momentarily?

Indeed. That comes under "something else is moving/changing" which he
couldn't tell :-). Having you and Jinx 'tell' what else it may be is
(hopefully) useful.

> As I said before, keep the wiper unloaded and put a small cap on it to
> "remember" the voltage when the wiper goes momentarily open.  As I also
> said
> before, I would add a single R-C low pass filter after that.

Sounds good. May even work :-).

_________

If you have more than one of these pots then trying a second to see if it
produces the same result may be a very useful idea. If the result is
substantially different the first may befaulty. If the result is much the
same yiu have a problem with its design that you need to compensate for.

Russell

Could be part of a production run, and all had same defect that wasn't
caught. I saw where engine starter gears were machined 10 on a mandrel
at a time. one was found with defective machining, it was assumed there
were another 9 pieces somewhere that had gotten on to assembled engines.
That was about the only part that isn't tested during the hot test (run
the engine for 6 minutes on gasoline) since an air motor on the front of
the crankshaft was. Within an hour the plant manager had dispatched a
engineer with a test gauge to each of the customers (auto assembly
plants) to intercept those gears before the plant had back charges for
swapping engines at the assembly plant, plus the negative publicity and
statistics. This defect could not have gotten to the consumer, since the
engine would not start at the assembly plant. This plant built 1400
engines per 8 hour shift.

Info wrote:
{Quote hidden}

Yup, I think its either a bad batch or a bad design.
Ouch, catching it early is always good =)

-----Ursprungligt meddelande-----
Could be part of a production run, and all had same defect that wasn't
caught. I saw where engine starter gears were machined 10 on a mandrel
at a time. one was found with defective machining, it was assumed there
were another 9 pieces somewhere that had gotten on to assembled engines.
That was about the only part that isn't tested during the hot test (run
the engine for 6 minutes on gasoline) since an air motor on the front of
the crankshaft was. Within an hour the plant manager had dispatched a
engineer with a test gauge to each of the customers (auto assembly
plants) to intercept those gears before the plant had back charges for
swapping engines at the assembly plant, plus the negative publicity and
statistics. This defect could not have gotten to the consumer, since the
engine would not start at the assembly plant. This plant built 1400
engines per 8 hour shift.

I have the same problem with all of them.
Will compensate by throwing them in the bin and replacing them with a
new properly working model, possibly other brand.

Thanxs for the help all =)

-----Ursprungligt meddelande-----
If you have more than one of these pots then trying a second to see if it
produces the same result may be a very useful idea. If the result is
substantially different the first may befaulty. If the result is much the
same yiu have a problem with its design that you need to compensate for.

Russell

>LoL.  Because it's the only thing you can think of...
Thanxs for the misquote. I didn't say that.

>high impedence momentarily?
If the wiper or its mechanics is flimsy/weak that would be likely.
If it isnt, and keeps a high contact-pressure the other option is more
likely. (Not sideways, inline) There probably is a magnectic coupling
between in-outside which has spring-properties that could be the reason
for overshoot (jump) during fast stops or vibration.

>wiper unloaded and put a small cap
>single R-C low pass filter
It is practically unloaded (100k)and there is a 1uF cap to 'remember'
W/o that the noise from the pot is terrible. (2v p-p)
It isn't the norm to connect these pots that way. (cap and R-C filter)
But the cap helped, and I bet a R-C filter would do a little too. But
its not fixing the real problem.

>industrial position sensing pots
It is, but straight 100mm. Don't know how many wipers. This kind
is supposed to have no noise from the wiper. Be linear to 0.05%
(laser trimmed) and last many cycles (100M+) And they are not very
cheap.

> True, but I have serveral and it woule be strange if all were broken
> straight out of the box =)

You did open the box carefully ? Didn't use explosives or a forklift ?

A dodgy batch is not impossible. You could've just got ..... lucky

Broken or frayed wire ?

Info wrote:
> True, but I have serveral and it woule be strange if all were broken
> straight out of the box

Yes, but have you tried the others?

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
lol, nah no explosives this time. I did put them on the forklift tho! :P
Yup, cant rule out I got ... lucky  :P
Will be even luckier with another model =)

-----Ursprungligt meddelande-----
> True, but I have serveral and it woule be strange if all were broken
> straight out of the box =)

You did open the box carefully ? Didn't use explosives or a forklift ?

A dodgy batch is not impossible. You could've just got ..... lucky

Broken or frayed wire ?

I have,  they are all up and running. And they work equally bad.

Info wrote:
> True, but I have serveral and it woule be strange if all were broken
> straight out of the box

Yes, but have you tried the others?

Info wrote:
> Thanxs for the misquote. I didn't say that.

You stated that some assumption or other was "the only thing it can be".
You did later say "from what I can tell", but that sort of blanket and
authorotative statement causes a interrupt in my mental parser.  You are in
no position to proclaim what the only possible explanation for the stated
symptoms are.  Then trying to add the qualifier "from what I can tell"
contradicts the original statement.  If you meant "the only thing I can
think of", then you should have said that.

{Quote hidden}

You came here asking for help with a problem you were unable to solve
yourself.  Then when people suggest things, you argue why they're wrong
instead of trying the suggestions and maybe stopping to understand the
mechanisms.  If you're so sure what is and isn't wrong, then don't waste
everyone's time asking.  You're making some incorrect assumptions above, but
I don't feel like arguing back and forth with you.  Go believe whatever you
please.  Wave a dead fish over it if you want, but then leave us out of it.

> There probably is a magnectic coupling
> between in-outside which has spring-properties that could be the
> reason for overshoot (jump) during fast stops or vibration.

Huh?  This is the first I remember hearing of any magnetic coupling.  Is
this just another wild guess you've come up with or is this not a normal pot
with the shaft mechanically tied to the wiper?  If the latter, you might
have mentioned it sooner.  If the former, stop speculating since you clearly
don't have the expertise to do so.

> (cap and R-C filter) But the cap helped,

This is meaningful.  But as explanations and advice only result in arguments
with faulty assumptions trying to explain why I'm wrong, I'll leave it to
your wild imagination to decide what it means.

> and I bet a R-C filter would do a little too.

You "bet", meaning you simply assume it won't work and therefore haven't
tried it, even though you don't have the expertise to judge nor is there any
evidence you've done the math to decide one way or the other?

> But its not fixing the real problem.

I thought the problem was to get a position signal from the pot with
something like a 95% settling time in 100mS.  There is a good chance a R-C
filter would solve this problem, so you need to explain what the real
problem is.

>> industrial position sensing pots
> It is, but straight 100mm.

By "straight" in this context you mean it has a linear slider instead of a
turning shaft?

> Don't know how many wipers.

How could you not know this!?  First, pins - 2 is a good starting point.
Second, if it's really a slide pot instead of a circular one, then it's only
got one wiper.

The reason I asked was if you had a two-wiper pot and tied the wipers
together, you'd not only get strange readings but also would be asking for
extra scratchiness.

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
>If you meant "the only thing I can think of",
ok ok ok, agreed that would have been a good way to put it.
But for heavens sake cut me some slack, English isnt my first
language and its hard to find the exact right words. I'd like to
see you try to do better in my language.

>people suggest things, you argue why they're wrong instead of trying
>the suggestions
No, I implemented the filtercap suggestion and also tried almost
every suggestion made by Jinx so far. His suggestions helped me
conclude that these pots are crap and I'll get a new ones. Thanxs
again btw =)

> If you're so sure what is and isn't wrong,
Thats how we work here. Someone suggests can it be this or that?
We check, and come back: No, it cant because... or yes thats it!
If someone is sure he is right he explains why he is right and if
the others cant explain why they think he is wrong his point stands.
You call it arguee, we call it working together troubleshooting.

>> There probably is a magnectic coupling
>is this not a normal pot with the shaft mechanically tied to the wiper?
No, its no normal pot. Ive said its a special expensive one, laser trimmed
for industrial linear measuring use. Again English isnt my first language
so I have problems to describe that part of it but here goes.
It has got a plastic part outside a alu. housing that moves ontop
of a stainless plate, no mechanical connection through to the inside
so when the plastic part moves the wiper inside must move by mag-force
or woodoo. (I am wildly guessing magnetic, but thats just me and I
am barely qualified to sit on a chair so dont take my word for it)

>You "bet", meaning you simply assume it won't work
No, I dont. Its you that simply assume that I assume that.
Actually I meant alot, but didnt want to go into detail about it since
it is a slow 'painfull' process for me to write in English. And thers a
good chance I get yelled at :P

I am sure you could suggest an R-C filter that would help smooth out the
jumps to 95% within 0.1s and it would be very interesting to try how much
of the sympthoms it actually could be fixed just for the fun of =)

But there isnt supposed to be any noise or jump issues to begin with. So my
thought was why try to fix/reduce this faulty pot's output with first a cap,

then a RC filter and might end up not good enough in the end. So it is
better
to get a new working pot w/o these issues.

These pots sits in a place that vibrates enough for them to give out random
signals which the RC-filter would haveto smooth out all the time. Also when
the pot stops efter movement it overshoots 250% of the actual value it
measures.
This means the sensors would be giving out false readings all the time, not
the best of conditions for accurate measurements.

>you need to explain what the real problem is.
There should be no noise issues, nor jumps to begin with! Thats the
real problem. Adding caps and filters is hiding the problem, not solving
the ground problem.

When I first tested the pot it was impossible to get a good reading due to
the severe noise. Thats not normal, but hey there is a switched powersupply
so it could be due to that and not a faulty pot so fix that. Filtecap added,
and the signal looks good.
Now the tests, not drowned in movement noise or anything from the PS,
shows the output 'jumps' and it gets worse as speed increases. Thats bad,

>it has a linear slider instead of a turning shaft?
Correct

>> Don't know how many wipers.
>How could you not know this!?  First, pins - 2 is a good starting point.
>Second, if it's really a slide pot instead of a circular one, then it's
only
>got one wiper.
Yes, its a linear slide pot. (didnt know it was called that) I thought you
meant if it had more than one wiper inside on the plastic track.

I didnt want to make the assumtion it can't have more than one wiper inside
the housing w/o bringing it out to the outside. Possibly more than one wiper
is what they use to make the no-noise-pots. Not all pots are equal I
thought.
Hence I wrote I dont know, since I dont know for sure if there absolutely
can
not be more than one wiper inside.

>The reason I asked was if you had a two-wiper pot and tied the wipers
>together, you'd not only get strange readings but also would be asking for
>extra scratchiness.
Interesting, didnt know that. Good info.

> You stated that some assumption or other was "the only thing it can be".
> You did later say "from what I can tell", but that sort of blanket and
> authorotative statement causes a interrupt in my mental parser.

His original said:

>> So something else is moving/changing and the only thing it can be is
>> the wiper inside the pot. from what I can tell.

As Occam indicates that the probability that "." is more likely a typo,
perhaps intended to be the symbol usually found just to the left of "." on
most keybaords, than a cessation of sentence or line of thought is to be
prefered (but may or may not be true), it may be wise to increase the scope
for your current value of "later" and / or desensitise the interupt trigger

> If you meant "the only thing I can
> think of", then you should have said that.

Nah. That's letting him off far too lightly. There are dozens of other more
explicit ways of saying it than he did. He should be taken to task, at
length, and preferably in triplicate, for not having used wording that
somebody else preferred to convey a concept, even if his way of saying it
was clear enough. In fact, despite his way of saying it having been clear
enough.

>> It is practically unloaded (100k)and there is a 1uF cap to 'remember'

> ... Then when people suggest things, you argue why they're wrong
> instead of trying the suggestions and maybe stopping to understand the
> mechanisms.

I could have sworn that adding a capacitor "to remember" the reading was
exactly what somebody or other here was suggesting. Mind grows dim though,
so maybe that wasn't what was suggested.

>... You're making some incorrect assumptions above,

:-)

> ... Go believe whatever you please.

:-)

>> and I bet a R-C filter would do a little too.

> You "bet", meaning you simply assume it won't work and therefore haven't
> tried it,

Interesting. My parser would have produced from that something like

You are probably correct that ...

rather than something like

I don't believe you ...

_________

>> Don't know how many wipers.

> How could you not know this!?

I'd assumed he meant that he thought you meant that there may be more than
one contact finger running on the track, but that he didn't know how many
fingers there were. I thought you may have too, not having initailly read
the original with as much attention as some may have. It now seems you may
have meant (and Occan says it's to be preferred) that there may have been
two pots which are being commoned, or a pot with wipers at two differnt
locations on the same track. Or something. But maybe not. What would Occam
know anyway?

RM

>>> the wiper inside the pot. from what I can tell.
>As Occam indicates that the probability that "." is more likely a typo,

Actually the dot is because I belive the word pot to be short for
potentiometer and I thought that if you chop it off you add the dot
to indicate you have cut it short. Maybe thats wrong to do in English.

>  You are probably correct that ...
Right again =)

>I'd assumed he meant that he thought you meant that there may be more than
>one contact finger running on the track, but that he didn't know how many
>fingers there were.
Right again =)
Contact finger... ok, that is what I thought wiper was.

>two pots which are being commoned,
Its three pots. They share power (24v) and ground (0v). Their outsignals
are separate. Eletrically they are ordinary, its mechanically they differ
from plain rotary pots.

Info wrote:

It would be nice if in the future you put a blank line between your reply
even deliberately deleted blank lines I put in there for readability.

> No, I implemented the filtercap suggestion and also tried almost
> every suggestion made by Jinx so far.

I don't think Jinx suggested a R-C low pass filter, I did.  I even derived
the values and gave a long explanation, but that apparently got swallowed by
the server.

> Thats how we work here. Someone suggests can it be this or that?
> We check, and come back: No, it cant because...

But your reasons why were incorrect.

> If someone is sure he is right he explains why he is right and if
> the others cant explain why they think he is wrong his point stands.
> You call it arguee, we call it working together troubleshooting.

That's fine when you're working together like in a design review.  It's a
bit different when you're asking someone else for a favor.

>>> There probably is a magnectic coupling
>>
>> is this not a normal pot with the shaft mechanically tied to the
>> wiper?
>
> No, its no normal pot. Ive said its a special expensive one, laser
> trimmed for industrial linear measuring use.

(blank lines restored for clariy)  Yes, you said it was laser trimmed, but
this is the first time you said anything about magnetic coupling.  That
makes a huge difference.

> It has got a plastic part outside a alu.

What's a "ALU".  To me that means "arithmetic logic unit", which doesn't
apply to pots.  You say you have problems with english, so don't make things
worse by using abbreviations unless you are really really sure they will be
universally understood.

> housing that moves ontop
> of a stainless plate, no mechanical connection through to the inside
> so when the plastic part moves the wiper inside must move by mag-force
> or woodoo. (I am wildly guessing magnetic, but thats just me and I
> am barely qualified to sit on a chair so dont take my word for it)

It never occurred to you that this is really important information!!?  This
puts a totally different light on things.  My first reaction is that this is
not a appropriate pot for the job.  Of course there are going to be
overshoots depending on speed since the slider is not ridgedly coupled to
the wiper.  This also explains some of the noise.  Pots can scratch when
moving, and this wiper moves whenever there is mechanical vibration.  You
can probably even get it to resonate at the right frequency.

> These pots sits in a place that vibrates enough for them to give out
> random signals

As expected, now that you've explained the slider is magnetically coupled to
the wiper.

>> you need to explain what the real problem is.
>
> There should be no noise issues, nor jumps to begin with!

If your spec is "no" noise, then go home, quit, pack it in.  There will
always be noise.  Specifying none is useless and only undermines your
credibility.

> When I first tested the pot it was impossible to get a good reading
> due to the severe noise. Thats not normal,

It probably is for this type of pot.  I think the real problem is the wrong
pot was specified.

Step back a few layers and look at the real problem.  Apparently you want to
sense the position of something.  There are lots of ways to do that.  Slide
pots are only one, and not usually the first choice.  Look at the whole
range of possibilities, including rotary shaft encoders, rotary pots, LVDTs,
and various more "elaberate" schemes.

> Yes, its a linear slide pot. (didnt know it was called that) I
> thought you meant if it had more than one wiper inside on the plastic
> track.

I've seen pots intended for full circular position feedback that have two
wipers 180deg apart.  These pots have 4 terminals, two for the ends of the
resistor and one for each wiper.  With the right firmware you can recover
full 360deg rotational position.  I was vaguely thinking that maybe you had
one of these and connected the two wipers together.  Clearly that's not the
case though.

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
>I don't think Jinx suggested a R-C low pass filter, I did.  I even derived
>the values and gave a long explanation, but that apparently got swallowed
by
>the server.

He suggested a cap, you went further and suggested R-C filter.
That might have been productive and interesting to read. No copy

>That's fine when you're working together like in a design review.  It's a
>bit different when you're asking someone else for a favor.

So if I ask a favour I cant question the suggestions?

>What's a "ALU".

Aluminium housing.

>If your spec is "no" noise,

With none, I meant not so much noise that the signal itself
drowns in it 200-fold. Less than 5mV noise is ok, and can be suppressed
to even lower levels. Perhaps with a cap or an R-C filter... =)

>Slide pots are only one, and not usually the first choice.  Look at the
whole
>range of possibilities, including rotary shaft encoders, rotary pots,
LVDTs,

For now I have another solution in place. Talked to the sales rep. today and
they are looking into it. He said the noise nor jumps is normal by any means

and will get back to me after he has talked to the manufacturer.

> Info wrote:
>> the only thing it can be is
>> the wiper inside the pot.

Perhaps an experiment to see if the pot is actually going to high
impedence....

1) Create a stiff voltage divider off the 24V supply to output say 4V.

2) Connect the output of the voltage divider to the wiper of the pot via a
100K ohm resistor.

3) Disconnect the A/D and use a high-impendence (1M or better) scope to
watch the signal.

4) Now if the pot (wiper) is apprpropriately low impedence the pot should be
able to pull the scope input well away from 4V. If it goes transiently high
impedence the voltage should jump toward the 4V created by the voltage
divider. The closer it gets to 4V the higher the wiper impedence must be.

-- Bob Ammerman
RAm Systems

Info wrote:
> This is how its connected:
> 24vDC --- 12k resistor --- 5k lin. pot. --- 0vDC
> Output from the 5k lin. pot goes into the A/D input (100kohm)

Most probably your potentiometer is KTC / KTF Linear sensor 75-1000 mm

http://www.regal.se/eng/pdf/Keen-engelsk.pdf

The specs says:
"The transducer must be used as voltage divider with a maximum current
in the cursorloop of 1ìA, else use an operation amplifier as shown."

Since your A/D input is 100kohm then the A/D input voltage should not
exceed 0.000001 A * 100000 Ohm = 0.1V. So, to amplify the input signal
use Op Amp as recommended. And also use integrating A/D converter to
integrate the noise.

Good Luck.
Info wrote:
> So if I ask a favour I cant question the suggestions?

It's more that you argue why it would not work without having tried it and
not following the directions in your analysis.  I said several times that
the best defense against scratchy pots is to keep the wiper unloaded.
Instead of buffering the wiper signal you explain that it goes into a A/D
with 100Kohm input impedence, which is definitely loaded in this context.

For normal pots, scratchiness comes from intermittent contact between the
wiper and the pot body, not from jitter sliding in the normal direction.
Electrically this looks like a varying resistance in series with the wiper.
Therefore you don't want current thru the wiper since that times the noisy
resistor is the noise voltage.  This is also why a cap helps.  If the
voltage coming from the pot is steady but the resistance to this voltage
varying, the cap voltage will stay constant once it's been charged to the
pot output voltage.  That takes care of the steady state case.  There will
be unavoidable scratching as the pot is moved, which is what the low pass
filter after the cap is for.  There is always some lag time you can tolerate
during motion, so you adjust the filter to be as aggressive as possible
without exceeding this lag time to some settling fraction.

I would have gone into all this detail and shown real numbers if you hadn't
kept trying to tell me it wouldn't work without having tried it and without
understanding the problem.

However, this was all before we knew that the mechanical coupling to the
slider was magnetic.  That adds a whole host of problems, including
overshoot and more scratchiness since the contact force is probably less to
keep the moving resistance low.

Slide pots are well known to suffer from these problems because they are
difficult to seal compared to rotary pots.  The magnetic coupling achieves
good sealing at high cost in dollars, ringing, overshoot, scratchiness, and
vibration tolerance.  None of what you are observing should be a big
surprise.  Basically, you're using the wrong tool for the job.

Why not convert the linear motion into rotary motion and use a rotary pot or

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

More... (looser matching)
- Last day of these posts
- In 2008 , 2009 only
- Today
- New search...