OpenServo.com

jharvey · Posted: Mon Jun 01, 2009 10:21 am Post subject:

amazingmagnets.com is a good source for custom magnets. They have someone machine the raw stock, then they magnetize it and send it to you. My plan is one magnetic rod, the reddish color is one pole, the blueish color is the other.

I plan on placing an order fairly shortly, so where do you get these you ask, in a short time you can get them from me. The noted quote above indicated that the majority of the cost is machine setup, so the difference in price from 10 to 100 is next to nothing. I'm planning on needing 10, but I'll be getting 100. I'll have extras, and I'll sell them for a couple bucks apiece.

A couple pictures of the removed shaft

Just for the heck of it, here's a couple pictures showing the basic positioning.

BasicFox · Joined: 15 Mar 2009 Posts: 59 Location: Belgium

I milled my defect pot open to see its makeup and have found an easy solution.

So u have the rod of the standard pot:

just like yours (also from a MG955 servo)

Now if i put it together again the rod is kept inplace by the gear and the outside of the pot.

So if we drill a hole into the rod and glue a magnetic rod from your website into it think this one would go well:
R250-063 ~ 1/16" Dia x 1/4" Long NdFeB Rod Magnet, Ni-Cu-Ni plated.
(1 Item = 48 magnets) Price US$5.28

then there is no need to mill your complicated magnetic piece

BasicFox · Joined: 15 Mar 2009 Posts: 59 Location: Belgium

the magnet will be too small, in the AS5046 datasheet they talk about a recommended magnet: Ø 6mm x 2.5mm for cylindrical
magnets

the minimum diameter is 4mm, your drawing has a 3mm diameter, u will have to work with 2 pieces

Also they always put the magnet on top of the IC, never on the bottom... don't know if the hall sensor array is effective that way

jharvey · Posted: Mon Jun 01, 2009 4:27 pm Post subject:

The key things that matters to the chip, is that the flux density is high enough, the flux is on center, and that it's wide enough to pass through the hall sensors. To give a gut feel, the vendor tossed out a typical value that could obtain those goals.

I looked at a bunch of washers, but didn't find one that I thought could work. Key issues include the space between the chip and the magnet, and axial alingment. You have to be spot on with your drilling operation, such that you are nearly exactly in line with the orginal shaft. You have to stay .015in on center. Not sure what happens when you go outside of that dim, but the data sheets notes to stay with in that area.

I've looked briefly at a couple magnetic FEA packages that should allow predictions that may indicated if this design will work. However, I also looked at the price of abut $150, and eh. Could go either way.

I know I have a smaller dia, but also I have longer shaft. I suspect the magnet flux coming off the end of this shaft is very similar to the flux from the vendors reference.

I'd be happy to see you modify your broken pot to bering it back from the dead, but I'm a bit of skeptic at the moment. I like the sounds of a lower cost approach.

BasicFox · Joined: 15 Mar 2009 Posts: 59 Location: Belgium

have a lathe here... so drilling spot on the center is not very difficult

like you said, lets each try something out and then compare the solution.
But now i'm studying for my exams (explaining the active internet checking :p)

jharvey · Posted: Fri Jun 05, 2009 2:41 am Post subject:

I finally got some time to run a FEA on this version of the modified custom magnet. I used FEMM. Below are the results.

This is a half axisymmetric cross section of the modified POT. The left is the custom magnet as currently drawn. On the right, is the vendor recommended 6mm dia by 2.5mm thick magnet. The lower square on both is the area where the AS5046 is located, you can see the magnetic part of the case influencing the field on the one on the left.

Hmmm, looks like those that cautioned me about the change from a 6 mm dia to a 3 mm dia were correct in their cautions. I'd guess this is not going to work as well as wished. I'll stew over the design some more and see if I can find a good approach. Back to the drawing board.

jharvey · Posted: Sat Jun 06, 2009 10:43 am Post subject:

I've done a bit more design work. Here's the latest FEMM prediction.

I couldn't find a close enough magnetic washer, and I couldn't find a good way to use a sold disk, so it's looking like it's a custom magnet. Here's a link the magnetic washer PDF. I've currently got in on quote, and expect it to cost about $1.5/ea in qty 100.

http://jaredharvey.com/openservo/servo/POT_shaft_2D-rB.PDF

Hmmm, if it doesn't download, browse to it. There may be a frames issue.

I don't see why you can't take out the existing shaft, and glue this magnet to the shaft that was removed. However glue isn't the best or most reliable mechanical fastener. So I'm also planning on having a bronze rod made that will fit this washer, and includes a rivet feature. Here's a link to that PDF, I'll put that out for quotes shortly.

http://jaredharvey.com/openservo/servo/POT_shaft_B.PDF

I'm planning on using some drill rod as a my rivet punch. I found McMastercarrr number 2900A215 at $2.10 with an OD tolerance of .0002in.

All in all, I think that should make a fairly reliable and accurate device.

jharvey · Posted: Thu Jun 11, 2009 10:08 am Post subject:

Just got the quote from amazingmagnets.com for the open-encoder magnet. Here it is.

Pricing is as follows.

Material: Licensed NdFeB
Grade: N40
Coating: Ni+Cu+Ni
Magnetization: Diametrical
Shape: Ring
Dimension: D6.1 x d1.6 x 3.05
Tolerance: +/0.1
Min Qty: 100
Unit Cost @ 100: $1.50 $1.50/pc
Unit Cost @ 1k: $0.35 /pc
Unit Cost @ 5k: $0.20 /pc

Production time: Approx 3 weeks.

I'm ordering 100 of these. The tolerance-ing on thickness is a pinch looser than I had drawn, so I'm updating the drawing and PDF to rev B. I have also update the link in the above post to point to the rev B copy.

It also occurred to me, I haven't posted a picture of what it currently looks like. So here's a shot of it.

Here are some general theories about this design. Tear apart an existing POT from the servo, and install the magnet either with the original brass shaft and glue, or with the replacement brass shaft and rivet feature. The PCB has a hole slightly larger than the magnet, this functions as an alignment tool, keeping you on center when you install the PCB and bend the tabs back down. With a couple wires down to the OSV3, we should be good to go.

jharvey · Posted: Fri Jun 12, 2009 10:22 am Post subject:

Looks like amazingmagnets has made this into an order that any one can order. Here's a link

https://www.amazingmagnets.com/index.asp?PageAction=VIEWPROD&ProdID=700

The product number is D118-240-063-DM

I just ordered 100 of these, I seem to recall it will take up to 3 weeks.

On a down side, pricing for the bronze shaft replacement is stupid expensive. So far the best price has been emachineshop.com. In qty 100, they charge about $4.50 each, in qty 50 they charge about $7 each, in qty 10 they charge about $30 each.

I got a quote back from a fellow with a screw machine, he wanted more than $11 each min qty 100, and I provide the material.

Here's one option for the material McMastercarr 8947K22 cost under $30, so the large majority of the cost is labor and machine wear.

I think I'll go the emachineshop rout. Because I should be able to glue the first ones in, I'll hold off a bit before I order the replacement shafts.

BasicFox · Joined: 15 Mar 2009 Posts: 59 Location: Belgium

and u can't put a connection pin into the old copper shaft and the magnet?

Also the field lines are much different with the hole in it... are you sure the last drawing is correct... normally the hole would also have horizontal fieldlines if you compare it to the 3D drawing

jharvey · Posted: Fri Jun 12, 2009 4:01 pm Post subject:

I believe it is correct. I tend to think the FEA is a more accurate representation of what will really happen, than my gut feel. On one hand I see the magnet as nearly the same as the vendor recommended, but see the fields are quite a bit smaller than the vendor recommended magnet. Then i see how quickly the field drops off when it leave the magnet, and having that in the middle of the it, well I can see how it would decrease the magnet quite a bit. Here's a direct link to the most recent FEMM file

http://www.nedharvey.com/jaredharvey/openservo/servo/custom_diskD.FEM

The other files are in that same directory and can be compaired against eachother.

About the pin, are you thinking of drilling a hole in the bottom of the existing shaft, then fastening it with a kind of nail? I believe that would still require folk to have a lathe or similar to create the hole in the existing shaft.

BasicFox · Joined: 15 Mar 2009 Posts: 59 Location: Belgium

yeah it would still need a lathe, but if you would do it that way and you use magnet that is custom made, it would be better to use a magnet with the pin already on it in stead of the hole so there is no hole in it:

jharvey · Posted: Sun Jun 14, 2009 7:41 pm Post subject:

One down side, amazingmagnets told me that can't do "steps" those lips I had on my first attempt of the custom shaft, they couldn't do that. They can bore it, put on the flats, OD and length. Beyond that the tooling cost go way up.

subculture · Joined: 13 Aug 2006 Posts: 15

There's a very small 1.7 gram servo made here in the UK by an outfit called Falcon models.

It uses an Allegro hall effect sensor and a couple of off the shelf magnets mounted in a little plastic holder which rotates with the servo arm-

http://www.falconmodels.uk.com/acatalog/Falcon_Mark_III_1_6g_Servo_Information_page.html

jharvey · Posted: Sun Jun 21, 2009 1:32 am Post subject:

I have modified a MG995, and posted a fair bit of how I did the mod. That page can be found here.

http://www.openservo.com/moin.cgi/ConstructionTutorial_MG995