Page 1 of 1

Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 3:15 am
by sharpcoder
Hello!

Thanks for taking a look at my post. Long story short, I'm trying to make an IoT telescope powered by the pi. Which sounds cool, and the software was super interesting to work on. But I am drowning in questions regarding the physical design of my project.

I have a 12v Stepper motor (and successfully wrote a java program to interface with it). I want to learn more about what kind of designs exist for a rotating platform (just trying to peg the X axis movement with this one). Or even just find a good resource (book, website, blog, etc) that maybe explains a bit more about hardware "design patterns" (if such a thing exists). I'm not looking to solve this one exact problem, per se. I'm hoping to find something that can help me start to think like a robotic/mechanical engineer/hobbyist who can do more than just program :)

Any guidance would be greatly appreciated. Thanks!

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 3:24 am
by W. H. Heydt
Look up the various telescope mounts that have been devised? Possibly the most interesting are the ones designed for really bi telescopes.

Edit to add.... I'm not sure that x-, y-, z-axis movement is relevant. Don't telescopes usually need rotational movements like polar and equitorial?

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 3:29 am
by Graymalk
There are quite a few, ranging from the very basics of electrical engineering to highly specific circuits like reference designs. You could always start with the same material engineering students work through. We used this book: http://global.oup.com/us/companion.webs ... edrasmith/

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 3:34 am
by sharpcoder
@W. H. Heydt
I've researched a few mounts. One promising one is called a dobsonian mount. But to make it controllable by the raspberry pi requires integrating a motor, and as far as I can tell, some parts of a dobsonian are traditionally friction based. Which is fine, that just means I'll have to improvise. And that's more what I'm after here: any resources that might help me understand what kind of patterns exist for solving hardware problems (such as using motors to make a platform rotate).

Also, the movement of celestial bodies can be described as polar coordinates. However, there are mathematic formulas that exist to translate a celestial polar coordinate into x/y (altitude and azimuth) for any point in time.

Edit: That's how you can do it in normal amateur astronomy too. Any point in the sky can ultimately be flattened out into those two dimensions. How far do you look left/right. How far do you look up/down.


@Graymalk
Thanks! I'll take a look.

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 4:11 am
by W. H. Heydt
There are a couple of ways to make a motor turn a platform. It depends on size, speed (and accuracy) that is needed. One method is a small spur gear at the motor (either directly on the shaft or through a series of reduction gears) engaging gear teeth at the edge of the platform. The other is to have the driving shaft (again, either directly or though a reduction chain) tanget to the edge of the platform and use a screw to engage the gear teeth on the platform.

The need for reduction gears is probably less than it used to be, given how common stepper motors with reasonable torque are. The reduction gears permit continuous motion by slowing the speed of the output shaft in exchange for higher torque to drive the motion. In any case, you want to reduce "lash"--the slack in the gear train that limits the precision of motion, especially when reversing direction.

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 4:57 am
by Gavinmc42
OpenSCAD means you can use code to make shapes.
http://www.openscad.org/

Think it might work on a Pi.
Source does not look big, an easy compile?

For a telescope maybe a belt drive so there is no backlash.
Great big belt pulley laser/cnc cut for the Azimuth axis

Dodsonian is probably the easiest telescope mount to make.
http://www.instructables.com/id/Homemad ... Telescope/

Could use a full circle drive pulley for altitude axis.
Plenty of books/plans for DIY Dobsonian scopes.

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 2:53 pm
by spock
openscad doesn't have to be compiled. it is in the raspbian repository. (at least it was last time i tried.)

for a fast viewport you have to enable the opengl driver.

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 4:43 pm
by mikronauts
How heavy is the scope?

For X axis, you could make a turntable, or just buy a "Lazy Susan" and modify it to be turned by a motor.

For the Y axis, assuming you can mount it at the center of gravity of the scope, another stepper should do it.

For positioning, you could use a tilt compensated electronic compass, or something like an HMC5883L and any accelerometer with some code to save money.

If you mount the compass above the center of the turntable & center of gravity of your scope rough positioning will be easy.
sharpcoder wrote:Hello!

Thanks for taking a look at my post. Long story short, I'm trying to make an IoT telescope powered by the pi. Which sounds cool, and the software was super interesting to work on. But I am drowning in questions regarding the physical design of my project.

I have a 12v Stepper motor (and successfully wrote a java program to interface with it). I want to learn more about what kind of designs exist for a rotating platform (just trying to peg the X axis movement with this one). Or even just find a good resource (book, website, blog, etc) that maybe explains a bit more about hardware "design patterns" (if such a thing exists). I'm not looking to solve this one exact problem, per se. I'm hoping to find something that can help me start to think like a robotic/mechanical engineer/hobbyist who can do more than just program :)

Any guidance would be greatly appreciated. Thanks!

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 6:06 pm
by sharpcoder
@W.H.Heydt
Thanks for the in-depth information! I really appreciate it. You make some good points and now that I think about it, I'll probably need some way to make my motor achieve more accuracy than just 200 ticks per rotation (which is what it's rated for). I will think on this further.

@Gavinmc42
I never thought about a beltdrive. That sounds like an interesting proposal.

@Mikronauts
Yes! A lazy suzan. I think that might work really well. Thanks for the suggestion. Funny you mention the HMC5883L. I actually already have one and have written a driver for it in my java program. I imagine the altitude axis will be more challenging than moving a lazy suzan, but I guess I'll take this one thing at a time :)

These are all great suggestions. I appreciate all the input!

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 7:12 pm
by stderr
sharpcoder wrote: I'll probably need some way to make my motor achieve more accuracy than just 200 ticks per rotation (which is what it's rated for). I will think on this further.
Microstepping and/or gearing down to allow more revolutions of the stepper for each revolution of the telescope. If you are going to allow full 360 degree unlimited movement, you'll need some sort of slipring to get power and perhaps even data to and from the scope itself, assuming you are planning on having a pi camera on it. You could use wifi for the data but somehow power has to get to the parts that are turning.

Re: Is there such a thing as hardware "design patterns"?

Posted: Wed Oct 12, 2016 8:44 pm
by jamesh
With an equatorial mount, you only need to move one motor for tracking purposes. Dobsonian, you need two motors. because you are moving in two axis.