Author Topic: OSHPark projects  (Read 12849 times)

0 Members and 1 Guest are viewing this topic.

Offline warhawk-avg

  • Newbie
  • Posts: 2
OSHPark projects
« on: March 23, 2014, 01:16:56 am »
I like tinkering, I like open source, more or less an amateur hobbyist (did this to get the kids into electronics and stuff)

Figured I would post up a few projects I have found and completed and got working
Full repository of found "open source" projects I have shared
https://oshpark.com/profiles/WarHawk-AVG

First is a 12vdc desulfator (alot of debate if these work or not (my experience they work), I just wanted to build one)
https://oshpark.com/shared_projects/ATrDSRDV


And a smallish joule thief led (this particular one ran for almost 2 weeks straight on a fresh AA battery)
https://oshpark.com/shared_projects/Pv0rRfN6


I have also recently got into these high power Cree flashlights..and have built some drivers for them
This is a 12A 32*7135 linear regulator board that is controlled by an Amtel ATtiny13A MCU, not fully tested (I got it working but needs some dimensional adjustments [aka stock driver 2mm thick, OSHPark board 1.6mm so a few solder blobs on the outer ground ring should make it fit)
Coincidentally I reflowed this in a stainless steel skillet on my kitchen stove


Another is a rework of a Chinese Nanjg style driver, normally they are 17mm but a few lights have 20mm drivers...so a guy redesigned it on a 20mm board
I stripped the Chinese one with a new hot air rework station and then put all the components on the OSHPark board...works surprisingly well


Only complaint I have is that alot of the shared projects there...people don't provide linkbacks to their project webpages or BOM's or schematics or anything, OSHPark did allow for editing the shares so it can be added but unfortunately many haven't gone back and updated it...thus really difficult to build stuff if you can't find out what the stuff is

Either way...post up pics and whatnot of what you have done or shared...
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: OSHPark projects
« Reply #1 on: March 23, 2014, 04:46:13 am »



Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline Fantasma25

  • Regular Contributor
  • *
  • Posts: 91
  • Country: mx
Re: OSHPark projects
« Reply #2 on: March 27, 2014, 07:28:45 pm »


Wow, how much did it cost you to make these? Are they 4 layer boards?
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: OSHPark projects
« Reply #3 on: March 28, 2014, 04:53:26 am »
About $300
Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline SirNick

  • Frequent Contributor
  • **
  • Posts: 589
Re: OSHPark projects
« Reply #4 on: June 12, 2014, 11:01:04 pm »
What are we looking at here?
 

Offline Richard Crowley

  • Super Contributor
  • ***
  • Posts: 4309
  • Country: us
  • KE7GKP
Re: OSHPark projects
« Reply #5 on: June 12, 2014, 11:38:31 pm »
Only complaint I have is that alot of the shared projects there...people don't provide linkbacks to their project webpages or BOM's or schematics or anything, OSHPark did allow for editing the shares so it can be added but unfortunately many haven't gone back and updated it...thus really difficult to build stuff if you can't find out what the stuff is.

Indeed. And the same problem with the service that SparkFun operated (before they sold it off to somebody else).  There aren't one in 100 boards that had the slightest explanation about what they were.  I marveled that they even went to the bother of putting those pages online when they are so utterly and completely useless.
 

Offline mazurov

  • Frequent Contributor
  • **
  • Posts: 448
  • Country: us
Re: OSHPark projects
« Reply #6 on: June 12, 2014, 11:46:51 pm »
You do it other way around - post a project elsewhere and provide a link to the PCB (example here -> https://www.eevblog.com/forum/projects/inductor-saturation-tester-alternative-route-to-dump-the-excess-energy/msg364816). OSHPark is a PCB fab, not a project repository.
 

Offline Richard Crowley

  • Super Contributor
  • ***
  • Posts: 4309
  • Country: us
  • KE7GKP
Re: OSHPark projects
« Reply #7 on: June 12, 2014, 11:50:04 pm »
OSHPark is a PCB fab, not a project repository. 
Fine. They can do business any way they wish. And more power to them, as far as I am concerned.
But then why do they make a pretense of offering a "project repository"?  That's the part I don't get.
 

Offline Wilksey

  • Super Contributor
  • ***
  • Posts: 1171
Re: OSHPark projects
« Reply #8 on: June 13, 2014, 12:30:14 am »
Hi MarshallH,

You say your boards cost $300 to make, is this just PCB fabrication or assembly as well?
I often wonder where the best and cost effective place would be to get boards assembled, there are some quite difficult (BGA, LGA, etc) components that cannot be hand placed (easily or reliably) and the cost of assembly here in the UK seems quite high! Especially for the hobbyist running 1 off proto's and the like.

Anyone recommend a "cheap" assembly house, much like we have to PCB fab?

Cheers
 

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 1602
  • Country: us
Re: OSHPark projects
« Reply #9 on: June 13, 2014, 01:17:32 am »
Cheap professional assembly jobs just don't exist.  And won't exist for a while.

The problem is machine time.  Lets assume you are a real professional assembly house taking orders from many customers.  Your machines are your capital investment and you have to make that back on the number of boards you assemble.  Any time your machine is not running boards, it's not making money.  That includes setup time for each and every project you take on from a customer.

Lets assume your project has x SMD line items and y through hole parts.  That means for your boards to get assembled these things need to happen:
1) Someone has to program your BOM and pick and place locations into the machine and do the programming for the machine movements.  That assigns reels/tube/trays/etc to specific feeder locations on the machine and then calculates the best path for the machine to take to put all the parts on.  Think of is like CNC operation programming.  Repeat orders of exactly the same board won't need this programming charge, so you save like $200USD per order after the first one.
2) Someone then has to physically load all the reels/tubes/trays/etc on the machine.  This takes a surprisingly long time.  If you didn't buy machine friendly quantities of parts, so cut tape instead of reels, then the assembly house has to re-reel your cut tape or hand install those parts.  Same with lose ICs and SMD connectors and stuff.
3) Then your boards go through the actual automated assembly process.  Chips stuffed, reflow oven, cooling.
3.a) If you have a double sided load with parts on both sides, the boards go back through #3 again on the other side.  More machine time.
3.Note) You better hope your assembly doesn't have more line items than the biggest machine has feeders or else it will need to tie up two machines.

That's the bulk of the "automated" part, but your boards aren't done yet.

4) Then depending on the system process, the through hole parts need to be installed.  This is usually done by an actual person physically sticking all the parts in the holes.  They will either hand solder or if you are really high volume they might flow solder with a fixture covering the other SMD parts so they don't all fall off.  Those fixtures will cost around $1200USD but save a lot of hand soldering labor.
5) Your boards then usually go through a wash to get rid of the flux and have to dry.  There are other processes that can happen to clean the boards too.
6) The next step is inspection where they try to make sure assembly mistakes don't go out the door.  For BGAs you might even get xray inspection.  They might do inspection at various times in the process.  It doesn't have to happen just here.
7) Pack, ship, done.  The boards show up on your loading dock and you send the assembly house a fat check.

That's not even including functional testing.

As you can see there is a hell of a lot of manual processes involved and a big chunk of the money to have boards stuffed is just setup and manual labor getting the machines ready.  This gets worse because a small run of boards also has to buy full reels (or pay $12USD extra per line item for digireels) or else the assembly house has to do more manual labor getting your parts ready for the machine.

Small batches have to amortize that setup downtime over fewer boards and they come out extremely expensive. 

Even the availability of smaller at-home size pick and place machines won't turn into profitable cheap small prototype businesses because someone will still have to do the same setup as a big machine and then babysit the machine since it will be small and cheap and likely jam and break often.  If you wanted to put in the initial investment and buy your own cheap machine for a few thousand USD then go ahead and do it yourself and not counting your time your boards will be super cheap.  That's about it.

Wow, that turned into more of a rant than I intended.
 

Offline SirNick

  • Frequent Contributor
  • **
  • Posts: 589
Re: OSHPark projects
« Reply #10 on: June 13, 2014, 01:41:48 am »
Useful rant though.  I had absolutely no concept of how much it costs to have automated assembly done.  None of Dave's videos on the topic included pricing, that I recall.  I'm curious how this worked out with the uC.  Not exactly a high-volume product...
 

Offline Wilksey

  • Super Contributor
  • ***
  • Posts: 1171
Re: OSHPark projects
« Reply #11 on: June 13, 2014, 01:52:21 am »
Smokey,
What a wonderful explanation!

I know that the assemblers cant do what PCB fabs do and pool designs (unless they  use the same parts or there are enough spools left for the extra components), I am quite surprised that nobody has yet to come up with an idea to reduce the costs of putting say, 1 or 2 hard to fit (BGA, LGA, et al) to a board for a reasonable cost for hobbyists.

I can understand not wanting to do an entire assembly, from your description of the workings of such a task, but would 1 or 2 components be more palatable for a assembler to put on?
 

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 1602
  • Country: us
Re: OSHPark projects
« Reply #12 on: June 13, 2014, 02:36:02 am »
Hundreds of boards is probably ok for something like the uC with a small number of line items.  It's not usually the size of the board, it's the number of line items.

For more reference, we just got an 8in x 8in double sided load board quoted from our local (in the USA) assembly house.  They do great work and we can drive over and actually sit down and go over any issues that might come up. 
The stats are lead free with around 100 line items and around 600 components total per board.

The One-Time NRE charges are (All in USD):
(NRE = non reoccurring engineering)
Programming = 250
Solder paste stencil = 400
Flow solder fixture (this is optional, but the per board cost goes up if you don't do it since there are a lot of through hole parts on this design.  This is a big board so the fixture is really expensive.  Plus you get more than 1 fixture for this price) = 3500
... So total NRE (without the fixture) = 650 one time.  Doesn't matter what the run size is, this is the price.  If the PCB changes at all, you have to pay both again.  If the BOM changes, you have to pay the programming. 
Another fun fact about programming.... The computer calculates the optimal order of parts to install.  If you tweak the BOM for more than a couple parts, they can't just manually change the program because the optimal route can change significantly.

Per board cost (just for assembly):
25 piece price =  you don't even want to know....ok.. it's about 225 each
50 pieces = about 75 each
100 pieces = about 60 each
200 pieces = about 50 each

Wilksey:
In order to place something like a BGA reliably they really need to make a stencil so the paste is even and have a machine apply it.  And it has to have the right reflow thermal profile.  I mean sure they could hand wipe the paste on with a coke can stencil and reflow it in a toaster theoretically, but if every 5th or even 10th or probably even 20th (or more) board didn't come out right and had to be reworked they just lost all their profit if they aren't charging much. 
Plus with the time it takes them to setup to do your 3 BGAs they could be setting up a machine for some customer that wants 2000 boards loaded. 

SirNick:
If I remember correctly, when Dave was doing batches of 50 uCurrents back in the day he was having a guy in his basement assemble them for him by hand.  Once he went up to the thousands quantities of the kickstarter it was straight to the full assembly house with multi-board panels and stuff.
 

Offline mazurov

  • Frequent Contributor
  • **
  • Posts: 448
  • Country: us
Re: OSHPark projects
« Reply #13 on: June 13, 2014, 03:19:55 am »
Wow, that turned into more of a rant than I intended.

Small quantities of the board like @marshallh has posted can be done manually much faster than any machine (plus setup, obviously) from loose parts. I can stuff 3 of these in a day (and often do despite having a PNP - it's not feasible to spend any time setting up even one reel for small quantity. I pick and place parts that happen to be already set up on the machine though) , a professional assembler will do it even faster. If said assembler makes US$30/hr then the expenses are $240 + $60 for a mylar stencil = $300.  I think the price for 3 of those would be in $1500 - $2500 range.
 

Offline SirNick

  • Frequent Contributor
  • **
  • Posts: 589
Re: OSHPark projects
« Reply #14 on: June 13, 2014, 03:37:19 am »
Once he went up to the thousands quantities of the kickstarter it was straight to the full assembly house with multi-board panels and stuff.

Ah, that's the piece I missed.  Thanks for the break-down, that was a good read.
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5549
  • Country: us
Re: OSHPark projects
« Reply #15 on: June 13, 2014, 04:14:15 am »
I'm pretty sure it wasn't some guy on his basement, I remember it was a local house (not as in a home) that did the assembly manually, no basement mentioned.
 

Offline theatrus

  • Frequent Contributor
  • **
  • Posts: 346
  • Country: us
Re: OSHPark projects
« Reply #16 on: June 13, 2014, 06:14:14 am »
I recently slammed a few of my older projects into OSHPark:

https://oshpark.com/profiles/theatrus

Links head to the Github pages for most relevant information
Software by day, hardware by night; blueAcro.com
 

Offline salfter

  • Contributor
  • Posts: 23
  • Country: us
    • My Blog
Re: OSHPark projects
« Reply #17 on: June 26, 2014, 01:58:37 am »
Here's my most recent work:

https://oshpark.com/shared_projects/e7OaI1G2

It's an I2C and 1-Wire interface board for the Raspberry Pi.  It includes space for a clock chip with battery (connected over I2C) and a relay controller (connected over 1-Wire).  I built it to go atop my fermentation fridge, where it replaces an Apple II that previously did this job.



I2C and 1-Wire go through a set of level converters; they and a 5V supply are brought out to an RJ45 jack.  A 5V wall wart with a barrel plug can be used to provide power to the stack in place of the micro-USB jack on the Raspberry Pi. 

This is an earlier revision of the design.  The current version moves the power jack next to the GPIO header, so it can't short against the RPi's USB jacks.  (I put some packing tape on the USB jacks to avoid this problem.)

Here's the whole stack put together, with a touchscreen from Adafruit to provide the user-interface hardware:



...and in operation:



The project page at OSH Park links to the design files at Upverter:

https://upverter.com/salfter/b0ef63c2cd7a37c7/rpi_i2c_1w/

I had to write a Linux kernel driver for the DS2406 addressable switch...it's already been accepted into the linux-next tree at kernel.org, or you can get a diff here:

https://github.com/adafruit/adafruit-raspberrypi-linux/pull/4.patch

The code that makes a refrigerator controller out of it is also available:

https://github.com/salfter/pifridge

It's a web-based interface...assuming that I've not switched it off, you can check the temperature of the fridge in my garage here:

http://salfter.gotdns.org:81/

The touchscreen displays the exact same interface, implemented with X, Ratpoison, and Chromium in kiosk mode.

Of course, you can use this board for other purposes as well.  I'm not even using I2C offboard for the fridge controller (though the DS1307 RTC is connected on-board through it).
 

Offline LukeW

  • Frequent Contributor
  • **
  • Posts: 673
Re: OSHPark projects
« Reply #18 on: June 27, 2014, 05:31:22 am »
OSHPark is a PCB fab, not a project repository. 
Fine. They can do business any way they wish. And more power to them, as far as I am concerned.
But then why do they make a pretense of offering a "project repository"?  That's the part I don't get.

I think the point of it is that I can easily make PCBs that I've laid out for my project available for you to purchase.

You don't have to do the work of going to my website, finding the Gerber files for download, uploading them to the OSHPark website, seeing what the price is, and ordering them.
It's all provided for you very conveniently with just one click if you like my project and want to order some boards yourself.

However, it is not a substitute for a complete website and repository for your open source project, with documentation, construction details, schematics, BOM information, design details and EDA/CAD source files etc.
 

Offline Richard Crowley

  • Super Contributor
  • ***
  • Posts: 4309
  • Country: us
  • KE7GKP
Re: OSHPark projects
« Reply #19 on: June 27, 2014, 05:47:32 am »
You don't have to do the work of going to my website, finding the Gerber files for download, uploading them to the OSHPark website, seeing what the price is, and ordering them.
It's all provided for you very conveniently with just one click if you like my project and want to order some boards yourself.
I fear you are all missing the point.  Having web pages you can link to for boards is excellent.  But why bother making a public directory of them when not 1 in 100 of them have the slightest bit of information about WHAT they are?

Quote
However, it is not a substitute for a complete website and repository for your open source project, with documentation, construction details, schematics, BOM information, design details and EDA/CAD source files etc.
I would settle for simply requiring a link-back to the original info.
 

Offline mazurov

  • Frequent Contributor
  • **
  • Posts: 448
  • Country: us
Re: OSHPark projects
« Reply #20 on: June 30, 2014, 04:32:25 pm »
First is a 12vdc desulfator (alot of debate if these work or not (my experience they work), I just wanted to build one)

Thanks! I ordered a board and building one atm - have several derivatives already but never built the original circuit.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf