Need a bit of design work done

[Mike Rocket J. Squirrel E]

On 4/3/2005 8:45 PM Mark Borgerson wrote:

In article <Ab2dnZGatI8yBM3fRVn-uw@adelphia.com>,
j.michael.elliottAT@REMOVETHEOBVIOUSgmailDOT.com says...

On 4/3/2005 12:28 PM Mark Borgerson wrote:


In article <4d55b0d8fetonyw@ledelec.demon.co.uk>,
tonyw@ledelec.demon.co.uk says...


In article <4d5430173dtonyw@ledelec.demon.co.uk>,
Tony Williams <tonyw@ledelec.demon.co.uk> wrote:



Send the 6-bit output (1 of 64) into 2x 3-8 line
decoders. Arrange the 48 relays into a 6x8 matrix,
with 6x high-side relay drivers and 8x low-side.

Get the make-before-break by slugging the OFF of
each relay...... perhaps with 14x R+C gated gates
before the row and column relay drivers.

Followup.....

There's a problem when trying to do make-before-break
with a matrix. Overlap of two relays is ok until the
selection of the next relay requires a change to both
a new row and column. Two rows and two columns being
active will result in 4 relays being energised.

The only solution I can see is to have an R+C across
each relay. Around about 100 ohms and 47uF will do a
5-10mS holdup.



I presume that you mean to place the R+C across the input
to the relay driver. Placing substantial resistance
and capacitance near the actual relay coils, with their
low resistance and high inductance, sounds like a problem
looking for a place to happen.


It all seems a bit of overkill when you could simply buy
a 46-position stepped attenuator from Marchand Electronics

http://www.marchandelec.com/att.html

Or are you trying to compete with them by offering an
extra two steps?


No, no competing. I've tried those Shallco/Shallcross rotary switches
before and they don't offer a very good "feel."

However, your system makes the electrical connections, the 'feel' is
going to be an element of the mechanical design of the switches and
detents. Perhaps this is an issue for a mechanical engineer rather
than an electrical engineer.

That's why I want to use a good-feeling potentiometer for the control.
They are easily found in small quantities, whereas multiple-position
rotary switches with a good "feel" -- once you get past 20 or so
positions -- are not. Designing a custom switch for this application is
not feasible due to the very small quantities of units to be made.

It may definitely be easier to get a good 'feel' if you divide your 48
steps up into 3 revolutions of 16 steps with an electronic indicator.
That isn't easily done with a simple rotary switch.

Well . . . I see what you are saying . . . but, um. . . . I dunno. Let's
see: A 16-step rotary switch that goes from fully CCW to CW, and a
"low," "middle" "high" switch. To ramp the volume up you might start in
"first gear," crank the knob all the way up, then shift to "second
gear," while cranking the revs . . . I mean the knob back down . . . I
can see my wife facing such a thing: "I don't drive a stick."

And what if the average range you like is in between step 15 of the
middle range and step 2 of the high range. That could prove to be a
hassle what with knob twisting and range switching all the time.

You glance at a volume control knob and note it is at 12 o' clock. You
know how loud that is (generally -- after a bit of time with any system
you get a feel for how the knob pointer relates to sound level).
Automatic transmission operation.

--
--
Mike "Rocket J Squirrel" Elliott
71 Type 2: the Wonderbus
84 Westfalia: "Mellow Yellow (The Electrical Banana)"
KG6RCR

 

Mike Rocket J. Squirrel Elliott wrote:

On 3/29/2005 2:50 PM Helmut Sennewald wrote:

"Mike Rocket J. Squirrel Elliott"
j.michael.elliottAT@REMOVETHEOBVIOUSgmailDOT.com> schrieb im
Newsbeitrag
newsPmdnd2iT_QmKdTfRVn-pw@adelphia.com...

On 3/29/2005 10:43 AM Mike Rocket J. Squirrel Elliott wrote:


On 3/29/2005 3:08 AM Robert Baer wrote:


Mike Rocket J. Squirrel Elliott wrote:


I'm an analog guy and I need for a bit of logic/adc design work
done,
will pay. The design is essentially to sample the voltage on the
wiper
of a potentiometer, and use it to drive one of 48 relays,
depending on
the wiper voltage. The relays need to be driven in a
make-before-break
style. I don't want to run any clocks in the product if at all
possible. Low-speed stuff. I reckon this is digital design 101
as
taught in 1980.

Please private mail me for more information if this sounds like
no-brainer stuff to you and you have the time and are
interested.

I'm located in Carlsbad, CA.

Oops -- Thanks! (Wrong chip, right family. The LM3914 is the linear
one.)

I don't know how to write code, so programming a uC would require
paying
someone for that. And buying hardware to program the uC. I can't
debug
code, either, so I would need to keep bothering the programmer. I've
had
difficulty in the past with code that needed debugging a couple years

later, only to find that the programmer guy was no longer available.
A
hardware implementation, like this simple IC approach, is something I

can understand and debug myself. I guess I prefer to do things myself
if
possible using technology even I can understand.

Clocks and potential interference. The product will be a vacuum-tube
high end audio phono preamplifier. I personally don't think that
having
an oscillator in this product will be a problem with halfway decent
shielding. But the marketplace will have more trust in the product if
it
is dead quiet (EMI-wise) inside. High-end audio is odd that way. Far
easier to not build in a perceived problem then try to defend
challenges
later.

Have you thought about using a rotary encoder?

Hewlett-Packard do incremental encoders from 200 to 2000 increments per
rotation, and while this isn't entirely clock free, it would probably
generate a lot less noise than switching your relays on and off.
Farnell list a respectable range of parts, from about $30

Farnell also list the Bourns EAW0J-B24-AE0128 7-bit absolute rotary
encoder which encodes 128 different positions as an 8-bit Gray code
onto 8 parallel outputs, which is as clock-free as you are going to
get, and only costs about $12 in small quantities, but you won't have
as much freedom to play with the "feel"of the knob as you would with
the HP kit.

The rest of the circuit could probably be a Xilinx CoolRunner PLD -
these are available with enough output pins to drive each of your 48
relays separately, though I'd go for a 2-D matrix which could select
any one of the 48 with just 14 outputs.

----------
Bill Sloman, Nijmegen

 

Then you want an absolute rotary encoder, like the Bourns part I
mentioned.

http://www.bourns.com/pdfs/ACE.pdf

Put a collet knob with a pointer or a dot on the shaft of the encoder,
and make sure that the shaft is sitting in the right place before you
lock the knob onto the shaft by tightening the collet, and you've got
your display and memory.

The running torque is specified as between 0.5 to 1.5 newton-cm (0.75
to 2.5 ounce-inches) so it should stay put if you don't explicitly turn
it. You might want to put a sector plate on the shaft with a pin on the
panel (under the knob) to prevent the knob rotating through 360
degrees, but you could probably sell the continuous rotation to your
customers as a feature of a professional servo-quality rotation sensor
.......

------------------
Bill Sloman, Nijmegen

 

[Mark Borgerson]

In article <Ab2dnZGatI8yBM3fRVn-uw@adelphia.com>,
j.michael.elliottAT@REMOVETHEOBVIOUSgmailDOT.com says...

On 4/3/2005 12:28 PM Mark Borgerson wrote:

In article <4d55b0d8fetonyw@ledelec.demon.co.uk>,
tonyw@ledelec.demon.co.uk says...

In article <4d5430173dtonyw@ledelec.demon.co.uk>,
Tony Williams <tonyw@ledelec.demon.co.uk> wrote:


Send the 6-bit output (1 of 64) into 2x 3-8 line
decoders. Arrange the 48 relays into a 6x8 matrix,
with 6x high-side relay drivers and 8x low-side.

Get the make-before-break by slugging the OFF of
each relay...... perhaps with 14x R+C gated gates
before the row and column relay drivers.

Followup.....

There's a problem when trying to do make-before-break
with a matrix. Overlap of two relays is ok until the
selection of the next relay requires a change to both
a new row and column. Two rows and two columns being
active will result in 4 relays being energised.

The only solution I can see is to have an R+C across
each relay. Around about 100 ohms and 47uF will do a
5-10mS holdup.



I presume that you mean to place the R+C across the input
to the relay driver. Placing substantial resistance
and capacitance near the actual relay coils, with their
low resistance and high inductance, sounds like a problem
looking for a place to happen.


It all seems a bit of overkill when you could simply buy
a 46-position stepped attenuator from Marchand Electronics

http://www.marchandelec.com/att.html

Or are you trying to compete with them by offering an
extra two steps?


No, no competing. I've tried those Shallco/Shallcross rotary switches
before and they don't offer a very good "feel."
However, your system makes the electrical connections, the 'feel' is

going to be an element of the mechanical design of the switches and
detents. Perhaps this is an issue for a mechanical engineer rather
than an electrical engineer.

It may definitely be easier to get a good 'feel' if you divide your 48
steps up into 3 revolutions of 16 steps with an electronic indicator.
That isn't easily done with a simple rotary switch.

But I think I owe an apology to all here on sci.electronics.cad -- my
original post was just a solicitation for assistance on this design, to
be taken off-line and done privately. I had not intended to chew up
bandwidth with a non-CAD thread. So to anyone who's gotten a little
tired of this, I apologize for the inconvenience.

It's still interesting in that it reminds us that electronic does still
include switches and not every user interface requires a graphical
screen.

Mark Borgerson

 

[rickman]

Tony Williams wrote:

In article <_46dnR4qh5l7js3fRVn-sQ@rcn.net>,
Chuck Harris <cf-NO-SPAM-harris@erols.com> wrote:

Only works if you can predefine the rate the pot is turned.

There is a requirement for make before break on the
relays which inherently sets the maximum change rate.

A little comms relay will take not more than about
2 to 3 mS to close. Assume that 48 relays represent
1 full turn of the pot. So that would be about 100
to 140 mS end-end, or equivalent to about 500 rpm.

The better solution is to use some logic, perhaps a small
microprocessor that has an A-to-D converter built in, and
has a clock that is entirely contained within the wafer of
the chip.

A 40 pin PIC with ADC would certainly do the job,
probably allowing Bill's out-of-sequence delayed
matrix driver to be implemented in software.

Software has the advantage of being able to do
extra performance enhancers, without additional
(and expensive) hardware. For example, putting
in a small amount of hysteresis, so that there
is no relay chatter at the switchover points.

The OP said no clocks, but I just realized that a small MCU can use an
entirely internal clock which would produce very, very little in the
way of EMI which is what the "No Clock" requirement is about. Once you
go with an MCU with a clock rate 1000x the relay rate, you can do
pretty much anything you want in software.

BTW, the EMI issue is normally only when the pot is not being turned.
My experience with high end audio is that adjustments like this are
only used when no recording is being done. Clearly the relays make
more EMI than a dozen MCUs with external clocks. But the relays only
change when you twirl the knobs. The MCU clock runs all the time.

 

Mike Rocket J. Squirrel Elliott wrote:

On 4/2/2005 4:17 AM bill.sloman@ieee.org wrote:

Then you want an absolute rotary encoder, like the Bourns part I
mentioned.

http://www.bourns.com/pdfs/ACE.pdf

Put a collet knob with a pointer or a dot on the shaft of the
encoder,
and make sure that the shaft is sitting in the right place before
you
lock the knob onto the shaft by tightening the collet, and you've
got
your display and memory.

The running torque is specified as between 0.5 to 1.5 newton-cm
(0.75
to 2.5 ounce-inches) so it should stay put if you don't explicitly
turn
it.

Does anyone know how to translate a torque spec like this into how a
knob "feels"? On a proper bit of high-end audio, the knobs want to
feel
silky with a bit of resistance, like they are damped with heavy
grease.
A knob that is too hard to turn, feels gritty or rough, or spins like
a
pinwheel with little or no resistance is likely to raise eyebrows.
Raised eyebrows: bad!

Buy one and find out.

--------
Bill Sloman, Nijmegen

 

Mike Rocket J. Squirrel Elliott wrote:

On 4/3/2005 8:45 PM Mark Borgerson wrote:

In article <Ab2dnZGatI8yBM3fRVn-uw@adelphia.com>,
j.michael.elliottAT@REMOVETHEOBVIOUSgmailDOT.com says...

On 4/3/2005 12:28 PM Mark Borgerson wrote:


In article <4d55b0d8fetonyw@ledelec.demon.co.uk>,
tonyw@ledelec.demon.co.uk says...


In article <4d5430173dtonyw@ledelec.demon.co.uk>,
Tony Williams <tonyw@ledelec.demon.co.uk> wrote:



Send the 6-bit output (1 of 64) into 2x 3-8 line
decoders. Arrange the 48 relays into a 6x8 matrix,
with 6x high-side relay drivers and 8x low-side.

Get the make-before-break by slugging the OFF of
each relay...... perhaps with 14x R+C gated gates
before the row and column relay drivers.

Followup.....

There's a problem when trying to do make-before-break
with a matrix. Overlap of two relays is ok until the
selection of the next relay requires a change to both
a new row and column. Two rows and two columns being
active will result in 4 relays being energised.

The only solution I can see is to have an R+C across
each relay. Around about 100 ohms and 47uF will do a
5-10mS holdup.



I presume that you mean to place the R+C across the input
to the relay driver. Placing substantial resistance
and capacitance near the actual relay coils, with their
low resistance and high inductance, sounds like a problem
looking for a place to happen.


It all seems a bit of overkill when you could simply buy
a 46-position stepped attenuator from Marchand Electronics

http://www.marchandelec.com/att.html

Or are you trying to compete with them by offering an
extra two steps?


No, no competing. I've tried those Shallco/Shallcross rotary
switches
before and they don't offer a very good "feel."

However, your system makes the electrical connections, the 'feel'
is
going to be an element of the mechanical design of the switches and
detents. Perhaps this is an issue for a mechanical engineer rather
than an electrical engineer.

That's why I want to use a good-feeling potentiometer for the
control.
They are easily found in small quantities, whereas multiple-position
rotary switches with a good "feel" -- once you get past 20 or so
positions -- are not. Designing a custom switch for this application
is
not feasible due to the very small quantities of units to be made.

It may definitely be easier to get a good 'feel' if you divide your
48
steps up into 3 revolutions of 16 steps with an electronic
indicator.
That isn't easily done with a simple rotary switch.

Well . . . I see what you are saying . . . but, um. . . . I dunno.
Let's
see: A 16-step rotary switch that goes from fully CCW to CW, and a
"low," "middle" "high" switch. To ramp the volume up you might start
in
"first gear," crank the knob all the way up, then shift to "second
gear," while cranking the revs . . . I mean the knob back down . . .
I
can see my wife facing such a thing: "I don't drive a stick."

And what if the average range you like is in between step 15 of the
middle range and step 2 of the high range. That could prove to be a
hassle what with knob twisting and range switching all the time.

You glance at a volume control knob and note it is at 12 o' clock.
You
know how loud that is (generally -- after a bit of time with any
system
you get a feel for how the knob pointer relates to sound level).
Automatic transmission operation.

A four postion coarse switch - "quiet", "clear","emphatic" and "loud"
-would allow you to overlap your ranges - say

1-16
11-27
22-37
33-48

which should eliminate any need for frequent range switching.

-------
Bill Sloman, Nijmegen