AVR-Chat

I have been looking at using a 2313 for an expermental pulse motor, but by
the time I add enough address decoder parts to do what I want, the circuit
starts getting a bit complex. I started think about maybe using a controller
per coil plus 1 master controller. The problem I see with the little 8 pin
atmel chips is, while they have a lot of stuff in them, you still end up
needing a number of external parts to demultiplex and get at them. The coil
controller needs at least 2 output bits, one input and one interupt. It
would also need a serial port of some kind to receive commands form the
master control, and and ADC that can be used to monitor the voltage across
the coil. I like the Atmel stuf because the chips are cheap, many being
under $5, and they have free development software.

On Sun, 12 Mar 2006, Timothy D. Lenz wrote:

> I have been looking at using a 2313 for an expermental pulse motor, but by
> the time I add enough address decoder parts to do what I want, the circuit
> starts getting a bit complex. I started think about maybe using a controller
> per coil plus 1 master controller. The problem I see with the little 8 pin
> atmel chips is, while they have a lot of stuff in them, you still end up
> needing a number of external parts to demultiplex and get at them. The coil
> controller needs at least 2 output bits, one input and one interupt. It
> would also need a serial port of some kind to receive commands form the
> master control, and and ADC that can be used to monitor the voltage across
> the coil. I like the Atmel stuf because the chips are cheap, many being
> under $5, and they have free development software.
> 

If you're going to do that, why not just go with one chip with more pins?   
Or do the individual controllers have to be in different locations?

Zack

For a task like this, take a look at the Mega48. It is one
of the newer, low-cost processors that still has a DIP
version. It has more I/O pins so that an app like this
might take less external hardware.

Jim


On Sun, 12 Mar 2006 14:43:28 -0700
 "Timothy D. Lenz" <tlenz1@cox.net> wrote:
> I have been looking at using a 2313 for an expermental
> pulse motor, but by
> the time I add enough address decoder parts to do what I
> want, the circuit
> starts getting a bit complex. I started think about maybe
> using a controller
> per coil plus 1 master controller. The problem I see with
> the little 8 pin
> atmel chips is, while they have a lot of stuff in them,
> you still end up
> needing a number of external parts to demultiplex and get
> at them. The coil
> controller needs at least 2 output bits, one input and
> one interupt. It
> would also need a serial port of some kind to receive
> commands form the
> master control, and and ADC that can be used to monitor
> the voltage across
> the coil. I like the Atmel stuf because the chips are
> cheap, many being
> under $5, and they have free development software.
> 
> 
> 
>  
> Yahoo! Groups Links
> 
> 
> 
>  
> 
> 

---------------------------------------------------------------
The Think Different Store
http://www.thinkdifferentstore.com/
For All Your Mac Gear
---------------------------------------------------------------

I've looked at the higher pin count parts some but there are a couple of
problems with them. This will be just a hand wired on perf board circuit to
test and those higher pin count chips tend to be in the newer high density
packaging. I have found a few sockets that seem to be for doing that which
will likly end up using at some point anyway. The other is, Im not sure the
ADCs will be fast enough to do what I want even with one chip per coil. It's
not that the cpu won't be fast enough, but I'm not sure I'd get the sampling
rate I need. The goal is to try and monitor when the coil becomes saterated
by monitoring the voltage drop across it and then cutting the power and
allowing a FWBR capture the back EMF. The turn the coil on again repeatin
until the rotor has passed a certain point at which time the coil would be
left off until time to fire again. each coil cpu would get 2 rotor signals,
one for polarity passing magenets and when they are in line and the other a
pulse train to provide postion between pole shifts. the cpu would provide 2
signals used for controling the coil. The master control cpu would then
decide how many coils need to be used at the time.

The first test motor will only have 6 magnets and 5 coil pairs. But latter
versions will have more rings of coils and magnets with each ring increasing
number of magents and coil pairs by 4. It would get to be a lot of cpus, so
moer centralized would be prefered. I may be able to do compairs against set
voltages insted, since I'm mainly looking for the voltage to reach a certain
high or low. I haven't seen any min compair speed yet lookin over the pdfs,
but havn't had time to look much for that.


----- Original Message ----- 

From: "Zack Widup" <w9sz@prairienet.org>
To: <AVR-Chat@yahoogroups.com>
Sent: Monday, March 13, 2006 8:37 AM
Subject: Re: [AVR-Chat] selection for coil control


> If you're going to do that, why not just go with one chip with more pins?
> Or do the individual controllers have to be in different locations?
>
> Zack

As long as you aren't going to need more than 127 controlled by the same
master I2C sounds just about perfect for the application.  You specified:
2 - outputs
1 - input
1 - interrupt
1 - ADC
Serial Communication ( 2 pins for I2C )

With the above requirements I'd say that the 2313 is about perfect at the
moment.  If you could wait for the new ATTiny24/44/84 series of 14 pin chips
you can save yourself some space (14 vs. 20 pin, not bad).

Either way you have built in hardware support for TWI (I2C) serial, a proper
external interrupt, and the ADC's you require.  The extra pins, 4 on the
24/44/84 and 10 on the 2313 can be used as hardware address selection lines
that would allow 16 and 127 ( actually 1024 with all 10 pins but I2C doesn't
support that size address space ) possible addresses respectively.

If timing is vitally important you'll want to run an external crystal and
that'll take two of your pins, in that case I'd recommend the 2313 since the
24/44/84 would only allow 4 addresses at that point.  If all you needed was
4 slave controllers there's no problem but...

Martin Jay McKee

I have a similar problem/project, controlling a large number of
solenoids, which are spaced somewhat close together, but I
would like all the electronics for each unit to be easily
removable, and only have a simple communications path to
a master controller.
I am starting by using the Tiny26, which I already know well,
and it seems a good value in pins and dollars and functions.
I will have more pins than I need; I need
a PWM output, 2 analog sensor pins, and serial communication.
Also, 2 aux digital I/O. One of the analog sensor inputs may end
up being a diffential pair.

Ken

> On Sun, 12 Mar 2006, Timothy D. Lenz wrote:
>
>> I have been looking at using a 2313 for an expermental pulse motor, but
>> by
>> the time I add enough address decoder parts to do what I want, the
>> circuit
>> starts getting a bit complex. I started think about maybe using a
>> controller
>> per coil plus 1 master controller. The problem I see with the little 8
>> pin
>> atmel chips is, while they have a lot of stuff in them, you still end up
>> needing a number of external parts to demultiplex and get at them. The
>> coil
>> controller needs at least 2 output bits, one input and one interupt. It
>> would also need a serial port of some kind to receive commands form the
>> master control, and and ADC that can be used to monitor the voltage
>> across
>> the coil. I like the Atmel stuf because the chips are cheap, many being
>> under $5, and they have free development software.
>>
>
> If you're going to do that, why not just go with one chip with more pins?
> Or do the individual controllers have to be in different locations?
>
> Zack
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>

> As long as you aren't going to need more than 127 controlled by the same
> master I2C sounds just about perfect for the application.  You specified:

The most I could see needing in a stator/rotor group is 55. That would be a
5 ring setup and the diameter starts getting up there. 4 rings will likly be
a better upper limit keeping the diameter down while allowing more room for
bearing mounts. At 4 rings there would be 44 coil pairs. Any more and it
would be another group on the shaft slaved to the first.

> moment.  If you could wait for the new ATTiny24/44/84 series of 14 pin
chips
> you can save yourself some space (14 vs. 20 pin, not bad).

Now I know why Digikey dosn't list the 24 :). 14vs20 pins is not much of a
factor at this point. Bigger problem is trying to build on a perfboard using
the newer surface mount styles. Once/if I get a working design for more
expermenting, I could go to surface mount parts and have the PC boards made
buy someone like pcb123.com.

Timothy D. Lenz wrote:
> I have been looking at using a 2313 for an expermental pulse motor, but by
> the time I add enough address decoder parts to do what I want, the circuit
> starts getting a bit complex. I started think about maybe using a controller
> per coil plus 1 master controller. The problem I see with the little 8 pin
> atmel chips is, while they have a lot of stuff in them, you still end up
> needing a number of external parts to demultiplex and get at them. The coil
> controller needs at least 2 output bits, one input and one interupt. It
> would also need a serial port of some kind to receive commands form the
> master control, and and ADC that can be used to monitor the voltage across
> the coil.
 You just need to move up to an AVR  with more I/O pins, is all.

>  I like the Atmel stuf because the chips are cheap, many being
> under $5, and they have free development software.
>   

  Prezactky and excisely.


tom

Again, I say, look at the Tiny26, which does come in a DIP.
I am using it for very similar type control, using solenoid
coils and linear-stepper-motor coils.
Ken

> I've looked at the higher pin count parts some but there are a couple of
> problems with them. This will be just a hand wired on perf board circuit
> to
> test and those higher pin count chips tend to be in the newer high density
> packaging. I have found a few sockets that seem to be for doing that which
> will likly end up using at some point anyway. The other is, Im not sure
> the
> ADCs will be fast enough to do what I want even with one chip per coil.
> It's
> not that the cpu won't be fast enough, but I'm not sure I'd get the
> sampling
> rate I need. The goal is to try and monitor when the coil becomes
> saterated
> by monitoring the voltage drop across it and then cutting the power and
> allowing a FWBR capture the back EMF. The turn the coil on again repeatin
> until the rotor has passed a certain point at which time the coil would be
> left off until time to fire again. each coil cpu would get 2 rotor
> signals,
> one for polarity passing magenets and when they are in line and the other
> a
> pulse train to provide postion between pole shifts. the cpu would provide
> 2
> signals used for controling the coil. The master control cpu would then
> decide how many coils need to be used at the time.
>
> The first test motor will only have 6 magnets and 5 coil pairs. But latter
> versions will have more rings of coils and magnets with each ring
> increasing
> number of magents and coil pairs by 4. It would get to be a lot of cpus,
> so
> moer centralized would be prefered. I may be able to do compairs against
> set
> voltages insted, since I'm mainly looking for the voltage to reach a
> certain
> high or low. I haven't seen any min compair speed yet lookin over the
> pdfs,
> but havn't had time to look much for that.
>
>
> ----- Original Message -----
> From: "Zack Widup" <w9sz@prairienet.org>
> To: <AVR-Chat@yahoogroups.com>
> Sent: Monday, March 13, 2006 8:37 AM
> Subject: Re: [AVR-Chat] selection for coil control
>
>
>> If you're going to do that, why not just go with one chip with more
>> pins?
>> Or do the individual controllers have to be in different locations?
>>
>> Zack
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>

> Now I know why Digikey dosn't list the 24 :). 14vs20 pins is not 
> much of a factor at this point. Bigger problem is trying to build on 
> a perfboard using the newer surface mount styles. Once/if I get a 
> working design for more expermenting, I could go to surface mount 
> parts and have the PC boards made buy someone like pcb123.com.

Also - check out SurfBoards (DigiKey, Fry's) -- I find that I quite 
like SMT prototyping when I have boards to mount the chips and bring 
out to pads.  Too bad the larger circuits look like 3D Rubik's Cubes 
(apply power to opposite corners ;).  But it lends itself to 
modularity....

Also (I haven't used) www.pcbfabexpress.com for cheap 5 day turns.  I 
use Target3001 (http://www.ibfriedrich.de - last time I used Dey's 
Electronics [www.deyselectronics.com] as the US reseller) to do the 
layout (same thing as PCB123 iirc, except you can move the design back 
and forth to other programs and you get simulation, etc).

All the PCB layout programs seem to have their own quirks, but at 
least Target is cheap.

-Thor Johnson

Thanks, I'll checkout those sites.

----- Original Message ----- 

From: "thormj_tiburon" <yahoo.20.tjtiburon@spamgourmet.com>
To: <AVR-Chat@yahoogroups.com>
Sent: Wednesday, March 15, 2006 8:47 PM
Subject: [AVR-Chat] Re: selection for coil control


> Also - check out SurfBoards (DigiKey, Fry's) -- I find that I quite 
> like SMT prototyping when I have boards to mount the chips and bring 
> out to pads.  Too bad the larger circuits look like 3D Rubik's Cubes 
> (apply power to opposite corners ;).  But it lends itself to 
> modularity....
> 
> Also (I haven't used) www.pcbfabexpress.com for cheap 5 day turns.  I 
> use Target3001 (http://www.ibfriedrich.de - last time I used Dey's 
> Electronics [www.deyselectronics.com] as the US reseller) to do the 
> layout (same thing as PCB123 iirc, except you can move the design back 
> and forth to other programs and you get simulation, etc).
> 
> All the PCB layout programs seem to have their own quirks, but at 
> least Target is cheap.
> 
> -Thor Johnson