quadrature to voltage converter

[Jamie Morken]

Hi,

I have a rotary encoder outputting a quadrature signal that I would like
to convert to a voltage that indicates the encoders rotation direction
and rotation rate. ie. when the rotary encoder is rotating one
direction the voltage will be less than 2.5Volts and when rotating the
other direction the voltage will be greater than 2.5Volts, and at rest
the voltage should be close to 2.5Volts. Any ideas for a circuit that
can do this? I am running out of microcontroller inputs, so I would
like to use a single ADC channel instead of two digital inputs to decode
the quadrature signal.

cheers,
Jamie Morken

 

[Adam S]

Jamie Morken wrote:

Hi,

I have a rotary encoder outputting a quadrature signal that I would like
to convert to a voltage that indicates the encoders rotation direction
and rotation rate. ie. when the rotary encoder is rotating one
direction the voltage will be less than 2.5Volts and when rotating the
other direction the voltage will be greater than 2.5Volts, and at rest
the voltage should be close to 2.5Volts. Any ideas for a circuit that
can do this? I am running out of microcontroller inputs, so I would
like to use a single ADC channel instead of two digital inputs to decode
the quadrature signal.

cheers,
Jamie Morken

I don't think there is an easy way, especially since you want frequency
to voltage conversion. If you absolutely cannot spare another input pin
then maybe its possible to add a 8 channel mux/demux,
74HC151/153/259/138/4051 to free 5 pins.

Adam Seychell

 

[Tim Wescott]

Jamie Morken wrote:

Hi,

I have a rotary encoder outputting a quadrature signal that I would like
to convert to a voltage that indicates the encoders rotation direction
and rotation rate. ie. when the rotary encoder is rotating one
direction the voltage will be less than 2.5Volts and when rotating the
other direction the voltage will be greater than 2.5Volts, and at rest
the voltage should be close to 2.5Volts. Any ideas for a circuit that
can do this? I am running out of microcontroller inputs, so I would
like to use a single ADC channel instead of two digital inputs to decode
the quadrature signal.

cheers,
Jamie Morken

Blech. You want to take a nice clean encoder signal and crap it up in

the analog world just so you can crap it up more with an A to D
conversion? Blech. I'd just get a bigger microprocessor.

Do this as an exercise in digital design:

.----------.
A .----. .----. | | ___ To ADC
------|D Q|---o--|D Q|-----| up|---|___|---o---------
| | | | | | | |
.--|>ck | .----|>ck | | | |
| '----' | | '----' | | ___ |
| | '-------------| ~down|---|___|---o
B | .----. | .----. | | |
---|--|D Q|-|-o--|D Q|-----| some | |
| | | | | | | | logic | ---
o--|>ck | o----|>ck | | | ---
| '----' | | '----' | | |
| | '-------------| | |
| | | | ===
---o---------' '----------' GND
clock
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

The first pair of D flip-flops cleans up the signal (if you're clocking
this fast then use a chain of two or three here). The second just
delays it by one clock. Run the four signals into some combinatorial
logic -- the idea here is to have the "up" line go high whenever the
encoder advances, and the "~down" line go low whenever the encoder
retreats. I am _not_ going to detail the combinatorial logic here --
just get out a pencil and paper, make some Karnaugh maps, and be happy.
You'll end up using lots of xor gates.

Size the resistors and cap for your desired level of anti-aliasing, or
put a real filter there, and have fun.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

 

[Joerg]

Hello Tim,

Blech. You want to take a nice clean encoder signal and crap it up in
the analog world just so you can crap it up more with an A to D
conversion? Blech. I'd just get a bigger microprocessor.

Hey, nothing against the anlog world, please...

Do this as an exercise in digital design:

.----------.
A .----. .----. | | ___ To ADC
------|D Q|---o--|D Q|-----| up|---|___|---o---------
| | | | | | | |
.--|>ck | .----|>ck | | | |
| '----' | | '----' | | ___ |
| | '-------------| ~down|---|___|---o
B | .----. | .----. | | |
---|--|D Q|-|-o--|D Q|-----| some | |
| | | | | | | | logic | ---
o--|>ck | o----|>ck | | | ---
| '----' | | '----' | | |
| | '-------------| | |
| | | | ===
---o---------' '----------' GND
clock
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

Ok, can't do ASCII art so I'll describe an alternative:

Replace the cap with a resistor, delete the digital stuff and run A and
B directly into the two horizontal resistors but make these of different
value. Now there will be four distinct voltage levels coming out. That
allows to resolve from the ADC value the status of A as well as B at any
time. Basically you now have a 2bit poor man's DAC that feeds the ADC input.

Regards, Joerg

http://www.analogconsultants.com

 

[Frank Bemelman]

"Joerg" <notthisjoergsch@removethispacbell.net> schreef in bericht
newslxIe.9977$_%4.1312@newssvr14.news.prodigy.com...

Ok, can't do ASCII art so I'll describe an alternative:

Replace the cap with a resistor, delete the digital stuff and run A and
B directly into the two horizontal resistors but make these of different
value. Now there will be four distinct voltage levels coming out. That
allows to resolve from the ADC value the status of A as well as B at any
time. Basically you now have a 2bit poor man's DAC that feeds the ADC
input.

Very nice Have to try that once, just for the heck of it...


--
Thanks, Frank.
(remove 'q' and 'invalid' when replying by email)

 

[Tim Wescott]

Joerg wrote:

Hello Tim,

Blech. You want to take a nice clean encoder signal and crap it up in
the analog world just so you can crap it up more with an A to D
conversion? Blech. I'd just get a bigger microprocessor.


Hey, nothing against the analog world, please...

I don't have anything against either analog or digital. It's just
jumping from one to the other like a hyperactive dolphin that gives me
the heebee jeebies.

-snip-

Ok, can't do ASCII art

It's a cool little program, and free.

so I'll describe an alternative:

Replace the cap with a resistor, delete the digital stuff and run A and
B directly into the two horizontal resistors but make these of different
value. Now there will be four distinct voltage levels coming out. That
allows to resolve from the ADC value the status of A as well as B at any
time. Basically you now have a 2bit poor man's DAC that feeds the ADC
input.

I've done similar before. Yes it would work, you'd have to be careful

of transitional values and you'd have to sample faster than the encoder
would ever go, but you'd keep the digital digital.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

 

[Joerg]

Hello Tim,

I don't have anything against either analog or digital. It's just
jumping from one to the other like a hyperactive dolphin that gives me
the heebee jeebies.

I have done that sometimes and it also gave the other (client's)
engineers the heebee jeebies. But if the accuracy is good enough and it
is cheaper, why not? Once a huge hotrod DSP became more or less
unemployed in the wake of some anadigital mods. They decided to leave it
on there anyway but from a MIPS point of view it could have been
replaced by a 4bit Chinese uC.

Ok, can't do ASCII art

It's a cool little program, and free.

Yes, I'll have to get it. Wish Eagle had a routine for converting a
schematic to ASCII. Shouldn't be too tough but I am not good at writing
these Eagle user language programs.

so I'll describe an alternative:

Replace the cap with a resistor, delete the digital stuff and run A
and B directly into the two horizontal resistors but make these of
different value. Now there will be four distinct voltage levels coming
out. That allows to resolve from the ADC value the status of A as well
as B at any time. Basically you now have a 2bit poor man's DAC that
feeds the ADC input.

I've done similar before. Yes it would work, you'd have to be careful
of transitional values and you'd have to sample faster than the encoder
would ever go, but you'd keep the digital digital.

I hope Jamie forgives me but I was just waiting for the other shoe to
drop. Like that he's also running a bit too tight on MIPS to do this or
almost out of ROM space ;-)

Regards, Joerg

http://www.analogconsultants.com

 

[Joerg]

Hello Frank,

Replace the cap with a resistor, delete the digital stuff and run A and
B directly into the two horizontal resistors but make these of different
value. Now there will be four distinct voltage levels coming out. That
allows to resolve from the ADC value the status of A as well as B at any
time. Basically you now have a 2bit poor man's DAC that feeds the ADC

input.

Very nice Have to try that once, just for the heck of it...

Well, yes, but as Tim mentioned keep in mind that you have to run the
corresponding ADC channel at a good enough clip. You don't have an
interrupt capability from these signals like you would if there were
enough uC pins. Maybe you could finagle something with the comparator
but then it gets ugly. Plus Jamie doesn't seem to have that extra pin.

Regards, Joerg

http://www.analogconsultants.com

 

[Jonathan Kirwan]

On Fri, 05 Aug 2005 00:47:09 GMT, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

It's a cool little program, and free.

Yes, I'll have to get it. Wish Eagle had a routine for converting a
schematic to ASCII. Shouldn't be too tough but I am not good at writing
these Eagle user language programs.

If you want, you can also try using this program of mine to convert
LTSpice schematics to ascii:

http://users.easystreet.com/jkirwan/new/LTSpice.html

Lets you use a decent schematic capture. However, it is pretty
limited in the ASCII models I've hooked into it. Easily added to the
library file with any editor, but then you still have to add them.

An example output is shown on the web page above.

Jon

 

[Ben Bradley]

On Thu, 04 Aug 2005 22:06:28 GMT, Jamie Morken <jmorken@shaw.ca>
wrote:

Hi,

I have a rotary encoder outputting a quadrature signal that I would like
to convert to a voltage that indicates the encoders rotation direction
and rotation rate. ie. when the rotary encoder is rotating one
direction the voltage will be less than 2.5Volts and when rotating the
other direction the voltage will be greater than 2.5Volts, and at rest
the voltage should be close to 2.5Volts. Any ideas for a circuit that
can do this? I am running out of microcontroller inputs, so I would
like to use a single ADC channel instead of two digital inputs to decode
the quadrature signal.

How about a few resistors making up an R-2R ladder from the
quadrature signals (this assumes active outputs, both pulled up and
pulled down - passive switches is only a little bit different),
creating four discrete voltages for the four quadrature states, then
the controller can do the quadrature decoding with the values from the
ADC. You would want three comparison values, each halfway in between
the adjacent voltages, for reliable operation.
This of course requires more programming, but a little code space
and a few resistors is cheaper than extra active components, isn't it?
If I had to do EXACTLY what you asked, I would make the 'circuit'
out of a small microcontroller that has two digital inputs for the
quadrature device and one DAC output. I suspect any design to do it
with "discrete" components (no micro, no FPGA) would cost more, and
certainly would take more components.

cheers,
Jamie Morken

May I send you a consulting bill?

-----
http://www.mindspring.com/~benbradley

 

[Bob Monsen]

Jamie Morken wrote:

Hi,

I have a rotary encoder outputting a quadrature signal that I would like
to convert to a voltage that indicates the encoders rotation direction
and rotation rate. ie. when the rotary encoder is rotating one
direction the voltage will be less than 2.5Volts and when rotating the
other direction the voltage will be greater than 2.5Volts, and at rest
the voltage should be close to 2.5Volts. Any ideas for a circuit that
can do this? I am running out of microcontroller inputs, so I would
like to use a single ADC channel instead of two digital inputs to decode
the quadrature signal.

cheers,
Jamie Morken

Here is a trick that may work. Build negative pulse detectors for the A
and B outputs, and use an RS flipflop. Connect B to set, A to reset.
Then, when the encoder is going forward, the duty cycle is long, whereas
when the encoder is going backwards, the duty cycle is short. This way,
you can use 'interrupt on change' and TIMER0 to determine the frequency
and direction...

You probably know this, but a simple negative pulse detector is

Vcc ----------o---.
| |
- \
^ /
| \
| |
input ---||---o---o--- output

RC << your highest expected input frequency, but make sure it's longer
than the hold time for the reset and set inputs.

You can also do this with a simple NPN flipflop. Use two NPN
transistors, emitters to ground, collectors to 5V through 10k resistors.
Connect the base of each transistor to the other's collector through a
100k resistor. Now, connect A to one side of a 10nF cap. Connect the
other side of the cap to Vcc through a 1k resistor. Then, connect a
diode anode to the base of one of the transistors, and the cathode to
the junction of cap and resistor. Same for the B side.

Use the collector of the transistor which is connected to B as your PIC
input. Then, forward gives a 75% duty cycle, backwards a 25% duty cycle.
Determine the frequency using the PIC clock.

--
Regards,
Bob Monsen

If a little knowledge is dangerous, where is the man who has
so much as to be out of danger?
Thomas Henry Huxley, 1877

 

[Joerg]

Hello Jon,

If you want, you can also try using this program of mine to convert
LTSpice schematics to ascii:

http://users.easystreet.com/jkirwan/new/LTSpice.html

Thanks, pretty cool. However, I draw schematics with Cadsoft Eagle. With
Spice I often fall into the old habit of just typing up an idea as text.

Regards, Joerg

http://www.analogconsultants.com

 

[Jonathan Kirwan]

On Fri, 05 Aug 2005 16:45:08 GMT, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

Hello Jon,

If you want, you can also try using this program of mine to convert
LTSpice schematics to ascii:

http://users.easystreet.com/jkirwan/new/LTSpice.html

Thanks, pretty cool. However, I draw schematics with Cadsoft Eagle. With
Spice I often fall into the old habit of just typing up an idea as text.

Regards, Joerg

No problem. It was more for my own use. I thought it may help some
others who were already using LTSpice or could consider it for
schematic capture, as it is free.

When I want to put out a schematic on a newsgroup, there is an
advantage to doing it in ASCII since it is a light load on the web and
will be archived for the long haul on google. I also tried to use
Andy's program for a bit, but I really found myself not liking some
aspects I couldn't customize and, more fundamentally, I just don't
like editing and modifying schematics done in ASCII. I often go
through several iterations and ASCII editing does NOT allow me to
stretch wires and so on -- too little intelligence. I also prefer,
because I'm a hobbyist and not a professional in this field, to do
some measure of validation before I post. LTSpice helps me do some
basic checks to help catch certain mistakes I may make or to point me
to thinking more on an issue before I write about it.

So, for me, LTSpice provides a one-stop shop where I can quickly hack
up an idea, verify I didn't completely get it wrong, modify and try
out some different approaches (not necessarily simulating them, just
putting down various concepts to see which might be easier for me to
explain), and then to spin off an ASCII image from it (including the
descriptions and equations used.)

For my use, it's just right.

Jon

 

[Joerg]

Hello Jon,

When I want to put out a schematic on a newsgroup, there is an
advantage to doing it in ASCII since it is a light load on the web and
will be archived for the long haul on google. I also tried to use
Andy's program for a bit, but I really found myself not liking some
aspects I couldn't customize and, more fundamentally, I just don't
like editing and modifying schematics done in ASCII. I often go
through several iterations and ASCII editing does NOT allow me to
stretch wires and so on -- too little intelligence. I also prefer,
because I'm a hobbyist and not a professional in this field, to do
some measure of validation before I post. LTSpice helps me do some
basic checks to help catch certain mistakes I may make or to point me
to thinking more on an issue before I write about it.

That is great. Probably you should promote it some more or at least
place a post with "LTSpice to ASCII schematic" or something in the
subject line to a few NGs. I'd volunteer to post in the German NGs if
you like. There are a lot of people who use LTSpice for schematics and
they could really use your program but most likely just do not know
about it.

Regards, Joerg

http://www.analogconsultants.com

 

[Jim Thompson]

On Fri, 05 Aug 2005 17:26:25 GMT, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

Hello Jon,

When I want to put out a schematic on a newsgroup, there is an
advantage to doing it in ASCII since it is a light load on the web and
will be archived for the long haul on google. I also tried to use
Andy's program for a bit, but I really found myself not liking some
aspects I couldn't customize and, more fundamentally, I just don't
like editing and modifying schematics done in ASCII. I often go
through several iterations and ASCII editing does NOT allow me to
stretch wires and so on -- too little intelligence. I also prefer,
because I'm a hobbyist and not a professional in this field, to do
some measure of validation before I post. LTSpice helps me do some
basic checks to help catch certain mistakes I may make or to point me
to thinking more on an issue before I write about it.

That is great. Probably you should promote it some more or at least
place a post with "LTSpice to ASCII schematic" or something in the
subject line to a few NGs. I'd volunteer to post in the German NGs if
you like. There are a lot of people who use LTSpice for schematics and
they could really use your program but most likely just do not know
about it.

Regards, Joerg

http://www.analogconsultants.com

Interesting. LTspice can actually read-in a PSpice Schematics. Now
all I need to figure out is how to import my PSpice symbols into
LTspice. I have tons that I've made up that don't exist in the
standard libraries.

Any simple way to do that Mike?

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[Joerg]

Hello Jim,

Interesting. LTspice can actually read-in a PSpice Schematics. Now
all I need to figure out is how to import my PSpice symbols into
LTspice. I have tons that I've made up that don't exist in the
standard libraries.

Maybe Jon's program could also read PSpice schematics then. But even if
you could translate it all to LTSpice I guess you'd be restricted to
whatever symbols he's got in his conversion program or you would have to
create new ones.

Any simple way to do that Mike?

Which Mike?

Regards, Joerg

http://www.analogconsultants.com

 

[Jim Thompson]

On Fri, 05 Aug 2005 18:30:50 GMT, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

Hello Jim,

Interesting. LTspice can actually read-in a PSpice Schematics. Now
all I need to figure out is how to import my PSpice symbols into
LTspice. I have tons that I've made up that don't exist in the
standard libraries.

Maybe Jon's program could also read PSpice schematics then. But even if
you could translate it all to LTSpice I guess you'd be restricted to
whatever symbols he's got in his conversion program or you would have to
create new ones.

Any simple way to do that Mike?

Which Mike?

Regards, Joerg

http://www.analogconsultants.com

The one and only Supreme Mikey, creator of LTspice ;-)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[John Larkin]

On Thu, 04 Aug 2005 16:36:17 -0700, Tim Wescott <tim@seemywebsite.com>
wrote:

I don't have anything against either analog or digital. It's just
jumping from one to the other like a hyperactive dolphin that gives me
the heebee jeebies.

Heck, mixing them all up (analog, digital, software) is the fun part.

John

 

[Bob Monsen]

Jonathan Kirwan wrote:

On Fri, 05 Aug 2005 00:47:09 GMT, Joerg
notthisjoergsch@removethispacbell.net> wrote:


It's a cool little program, and free.

Yes, I'll have to get it. Wish Eagle had a routine for converting a
schematic to ASCII. Shouldn't be too tough but I am not good at writing
these Eagle user language programs.


If you want, you can also try using this program of mine to convert
LTSpice schematics to ascii:

http://users.easystreet.com/jkirwan/new/LTSpice.html

Lets you use a decent schematic capture. However, it is pretty
limited in the ASCII models I've hooked into it. Easily added to the
library file with any editor, but then you still have to add them.

An example output is shown on the web page above.

Jon

That is nice. I have not looked at the C, but it must use the location
information in the .asc file, right?

Having an ascii schematic generator that works off of the netlist would
be pretty cool, but placement would be hard. Having an ascii editor that
does the rubber band connections would be even mo better.

However, what I would really like is for Andy's Ascii to work properly
under linux wine (the fonts are wrong, and he appears to be drawing in a
mode that screws up wine, so it leaves trash all over the screen). I
used it all the time under XP until I switched over to using linux most
of the time, so it's a pain to boot into windows to do a circuit for
somebody.

Alternately, an emacs mode for drawing circuits would be a ueful hack.

--
Regards,
Bob Monsen

If a little knowledge is dangerous, where is the man who has
so much as to be out of danger?
Thomas Henry Huxley, 1877

 

[Jonathan Kirwan]

On Fri, 05 Aug 2005 13:42:50 -0700, Bob Monsen <rcsurname@comcast.net>
wrote:

Jonathan Kirwan wrote:
On Fri, 05 Aug 2005 00:47:09 GMT, Joerg
notthisjoergsch@removethispacbell.net> wrote:


It's a cool little program, and free.

Yes, I'll have to get it. Wish Eagle had a routine for converting a
schematic to ASCII. Shouldn't be too tough but I am not good at writing
these Eagle user language programs.


If you want, you can also try using this program of mine to convert
LTSpice schematics to ascii:

http://users.easystreet.com/jkirwan/new/LTSpice.html

Lets you use a decent schematic capture. However, it is pretty
limited in the ASCII models I've hooked into it. Easily added to the
library file with any editor, but then you still have to add them.

An example output is shown on the web page above.

Jon

That is nice. I have not looked at the C, but it must use the location
information in the .asc file, right?

Yup. Just examines the graphics x,y locations and uses GCD to extract
the minimum possible arrangement for ASCII positions. Nothing fancy.

Having an ascii schematic generator that works off of the netlist would
be pretty cool, but placement would be hard.

That's a real pain, I think. If someone had come up with a good way
of communicating schematics through an automated generation from
netlists, it would have already been done. But there is a lot of art
in communication and whether you should just stub off a voltage rail
connection without actually showing the wire or otherwise show it can
be, in some cases, a decision of style. I don't like it, myself, when
schematics are laid out with all the supply wires criss-crossing
around. I suppose for someone with a soldering iron in hand, or
wire-wrapping the darned thing, it's nice to see the actual wiring.
But in trying to understand a schematic, it distracts the heck out of
me to see all that wire when it is the signal lines that really are
important and are being hidden a but by busing voltage all over the
place.

I've no idea how I'd go from netlist to a reasonable schematic for
study purposes, except in highly trivial cases.

Having an ascii editor that
does the rubber band connections would be even mo better.

I'd like that. And that is probably the one thing that kept me from
making use of Andy's program. I frequently just want to "get it down
on paper" and then move things around, adjust, add, delete, and then
re-position stuff so that it "looks right" for explanatory meaning. I
wind up erasing everything and starting over with Andy's program,
because I really can't stretch, shrink, grab and reverse or rotate,
etc. with his ASCII editor. He really needs to handle the objects as
objects and not only as ASCII text once it is placed. But that would
bring it to a whole new level and I don't expect it.

The schematic editors, like what is in LTSpice, on the other hand,
*have* to do all that. It's essential. So those features are built
in and handy. But then, it doesn't spit out ASCII. So that's the
part I tried to add. This way, I get the typical features of a decent
schematic entry program, complete with the concept of objects and
wires, plus I get some built-in ability to validate things. And then,
immediately upon some spice testing I can emit the ASCII directly from
the schematic that was just verified. No possibility of transcription
errors. Which is nice, too.

However, what I would really like is for Andy's Ascii to work properly
under linux wine (the fonts are wrong, and he appears to be drawing in a
mode that screws up wine, so it leaves trash all over the screen). I
used it all the time under XP until I switched over to using linux most
of the time, so it's a pain to boot into windows to do a circuit for
somebody.

Alternately, an emacs mode for drawing circuits would be a ueful hack.

Well, that's another possibility. I think I'll probably stick with
(and try and improve upon) the LTSpice-->ASCII method, though.
LTSpice is free and it has a decent schematic entry and editing
capability.

Even if you don't want to use the simulation feature (and there is no
need for that, many times, because you know what you want to get down
on paper and don't want to bother with exact values, at all), LTSpice
remains a very quick way to slap down a conceptual idea in schematic
form and get it quickly into a form that communicates well. So you
can go directly from schematic entry to ASCII, this way. Just don't
use the spice part. Or, if you want, spice is there, too. Doesn't
get much slicker than that.

Jon