OpenServo.com

dsandber · Joined: 01 Aug 2006 Posts: 52

I think if we can get enough people together to agree to buy some ASSEMBLED PCBs from Olimex, we can get them to make and assemble them for us for not too much.

I'm going to propose that they get the parts, make them, and assemble them for $40 a piece. UPDATE: Olimex estimates $30/piece, tested AND programmed. Does this sound reasonable? I figure it would take me about 20 hours to do myself, so if I value my time at $2/hour ( I value it more! ) I will be coming out even.

I will start off with 3 orders. I'd like to get 25 total to place a decent size order.

I haven't worked out whether we should each order the completed units from Olimex separately, or whether we should get them shipped to the US in bulk and then distributed individually.

Who's with me? Smile

robotjay · Posted: Sun Sep 17, 2006 6:49 pm Post subject:

I'm SORT OF with you. I think paying $40 per assembled board is still too expensive. Considering the material costs of 100 boards would be less that $10, $40 per board just seems ridiculous to me. I personally am building 2 bipeds with these boards, so I will need a total of 40. At $40 each, It would cost me $1600 just for the boards, not including the mechanical hardware. Too steep for my budget. A more reasonable price to me would be $15 per board. If you are willing to contact Olimex (or any other PCB house that assembles too) and can figure out quantities that will bring the price down to $15 per assembled board, then I'm in.

I've been doing a lot of writing, a lot of reading, and a TON of negotiating with distributors, and my cost per board is going to be less than $5 per board for 40 boards. Of course, I will have to assemble them myself, so I haven't included how much MY time is worth. But to me, the time loss is worth not having to pay exorbitant prices for labor.

I am going to post my own methods for obtainig parts for next to nothing, so read that if you are looking to get these boards for cheap.
_________________
"Nothing is fool-proof; For we fools are ingenious and will find a way."

dsandber · Joined: 01 Aug 2006 Posts: 52

I heard back from Olimex and they are interested!

They estimate it would be $30/board --- INCLUDING assembling, programming the AVR, routing the board, and testing!!!

They said they can sell bare boards for $5/each.

Yeah, if you need 40 boards I could see why it would be different. But still, take $30, subtract out the PCB cost, parts cost, testing time, programming time, ... I would guess that it would be faster to work extra hours at McDonalds at ( $6/hr ) than to do it yourself. And you prob. don't work at McDonalds.

ginge · Posted: Sun Sep 17, 2006 9:23 pm Post subject:

$30 sounds fairly reasonable. Just out of interest did they quote the price in $ or euros or other.

Now... $30 only sounds reasonable as I am in the UK. Because of the weak dollar and strong pound is very cheap (?16).

Assuming that an employee of McDonalds earns $6/hr (I still can't believe that!) It looks a lot like it isn't all that cheap for you US guys.

Did they set a minimum quantity for the assembly?

Notice above I said fairly reasonable...
To date, I have managed to get my PCB's for $3 and components for $6 (large quantity). I can build one of these boards in about 30 mins (say $6/hr so $3). I am not counting flash time, as it takes less time than my post has taken to write. This makes one of these boards relatively cheap.
If the price per board can be brought down to the $20 mark by Olimex, we are talking business.

Still, nice to have the option.

My 2p
_________________
http://www.headfuzz.co.uk/
http://www.robotfuzz.co.uk/

dsandber · Joined: 01 Aug 2006 Posts: 52

Minimum wage in the US is $5.15/hour, though some individual states and counties have higher minimums ( up to $9.50 apparently ).

The minimum I asked about was a 25 unit initial order.

dsandber · Joined: 01 Aug 2006 Posts: 52

Olimex said they can make the 'small' version of the PCB for less than $1/board. Apparently the $5/board price was for the 'large' PCBs.

Stefan, who made the Eagle PCB layout for Olimex suggested that if we were only going to create one version, it should be the small one since the small one can be used for big motors as well.

Stefan also said that the PCB layout has some minor issues that need to be worked out before it is ready for production.

ginge · Posted: Mon Sep 18, 2006 10:08 pm Post subject:

I was just looking into those issues. I should have them worked out fairly soon. If anyone knows of any other issues, then let me know.

Cheers,

Barry
_________________
http://www.headfuzz.co.uk/
http://www.robotfuzz.co.uk/

ginge · Posted: Mon Sep 18, 2006 10:13 pm Post subject:

After reviewing the board, I see no real issues. The only issue that Stefan identified was probably due to my parts list being wrong. The capacitor that got modded on his boards is not an issue if you purchase a different component (I forget which, but I can easily find out). If anyone wants a larger landing pad for the 33uF cap, then I can just about squeeze it on (it's tight though)

Barry
_________________
http://www.headfuzz.co.uk/
http://www.robotfuzz.co.uk/

dsandber · Joined: 01 Aug 2006 Posts: 52

Here's what Stefan told me the issues were:

Unfortunately the boards are not 100% perfect at the moment. I've encountered the following difficulties while assembling them:
* The landing pads of the 33?F tantalum capacitor C8 are too small. You have to solder it upstanding with an additional small wire.
* I couldn't find a 0.1Ohm Resistor wich can handle over 2Watt and still fit on the 0603 landing pad of R9. We probably have to choose a bigger package for it. At the moment I have soldered 4 resistors in a stack together to gain at least 2Watt.

In any case, we need people to commit to purchase some of these things!

ginge · Posted: Mon Sep 18, 2006 10:40 pm Post subject:

I see with the 33uF cap. I use an aluminium vertical type, but I will make the pad larger after seeing that people are using horizontal.

I will get onto the resistor pad issue. I will see what is the most common resistor type, and work from that.

Thanks for the refresher.

Barry
_________________
http://www.headfuzz.co.uk/
http://www.robotfuzz.co.uk/

ginge · Posted: Mon Sep 18, 2006 10:55 pm Post subject:

the 0.1R resistor is proving to be troublesome. the 2W version comes in pack 2512, which is massive. it's physically larger than the FET. What to do?

Barry
_________________
http://www.headfuzz.co.uk/
http://www.robotfuzz.co.uk/

dsandber · Joined: 01 Aug 2006 Posts: 52

I guess there's no shortcuts, if you need to dissipate 2W, you need a certain amount of surface area. Is 2W really necessary? Is there room to stick two 1W resistors in separate places?

[ If you can't tell, I haven't followed the hardware development ]

ginge · Posted: Tue Sep 19, 2006 12:25 am Post subject:

Personally I use 1W. I can't even remember where the 2W figure came from.

My experience with the 1W is good. Those come in a more workable package. At least if we get 1W in there, you could stack to get the required wattage. I will look into putting another resistor in, but space is getting tight. Without going to 4 layers, I think we might be hitting the limit without a new PCB design and layout.

Barry
_________________
http://www.headfuzz.co.uk/
http://www.robotfuzz.co.uk/

stefanengelke · Joined: 03 Jan 2006 Posts: 59 Location: Stuttgart, Germany

Barry, you're right, I've just ordered the wrong 33?F capacitor. I've used the same tantalum cap Mike did, but haven't changed the landing pads in the design. The cap in your part list is not a tantalum. Is there any difference? The easiest way is, just to use the one from your part list (Farnell#3199680). The only thing we have to check is, if there is enough space to solder it, because it's pretty tight besides the mosfets. I will order some and test it with the current PCB-design.

The issue with the 0.1Ohm resistor is, that the servo was behaving pretty strange with only one 1W resistor. It seemed like the motor power was reduced to a minimum. Just enough to move very slow. So I have stacked 4 of them and the problem was solved.
If we assume a maximum current of 2 A at peaks there will be a voltage of 0.2V at a 0.1Ohm resistor. This means the resistor has to dissipate 4W, isn't it?

There are 0.25W resistors whichs comes in 1206 (Farnell#1107359) pack or 0.5W in 2010 (Farnell#1107376). I will try to fit at least the 1206 landing pad into the current design.

-Stefan

dsandber · Joined: 01 Aug 2006 Posts: 52

> If we assume a maximum current of 2 A at peaks there will be a
> voltage of 0.2V at a 0.1Ohm resistor. This means the resistor has
> to dissipate 4W, isn't it?

P=VI = .2V*2A = .4W, not 4W. Is this just a current sense resistor, or what?