any cost reduction ideas?

Status
Not open for further replies.
I

instrumentek

Well-known member
Hi,
Just looking on any pointers for reducing costs on a project I want to market. I am using a Teensy LC and assembled PCB with approx 50 parts. It also incorporates an extruded aluminum case, display, and some other misc parts. I have done some stuff already:

- Source all electronic parts from one distributor to lower shipping costs
- I have a goal to have assembly on one side only (possibly solder any remaining bottom side parts myself)
- Design with 2 layer board only
- Some non essential parts I am buying through Ebay because they are so cheap and free shipping

Only making 10 - 30 at a time so i am just going to use the Teensy LC and not migrate to a board that only uses the IC.

Any input appreciated.
 
Don't under estimate your tech support costs.

And try to add in some diagnostic indicators. Obviously this hits your part count where it hurts but have a think about possible modes of failure and how you would diagnose those as quickly as possible via email. Especially you want to be able to identify any external factors that are impacting your product and preventing it working to avoid replacing serviceable units because somebody else's produce wasn't providing power, data or whatever. This also goes into the software design to always assume any input can be corrupt or invalid and ensure your code handles it without collapsing in a heap, and if possible usefully informing the user.

And if you are thinking about 3D printing parts:
https://www.kickstarter.com/project...ffordable-laser-cutter-engraver/posts/1078565
Thought that said having assembled the finished product that team made they had a lot of parts that clicked together in a fast and unambiguous way that beat the pants of some of the laser cut and bolt together kits I've met, so 3D printed jigs might be useful at the 20 unit production run scale to get things together nice and quick.

I think it's mentioned in one of the Bunnie blog posts but designing anything that will be assembled by a human so that if possible there is only one way to assemble it, or failing that, so that incorrect assembly is obvious is a good idea. And if it's not obvious design a test to make it obvious

So like typed components together, taking parts in just the right number of parks to make your boards so any extra's mean you missed something. Get accurate scales and weigh finished parts. And of course work out how you will test it quickly cheaply and completly
https://www.sparkfun.com/news/1575 (see final note for how they modified their test jig)

and of course:
https://www.kickstarter.com/project...rm-cortex-m4-usable-in-arduino-a/posts/305527
 
budget for... (excluding software and product fabrication)
  • warranty returns & shipping costs
  • time spent in tech support - especially in education - unless you do all that gratis
  • billing
  • shipping
  • receiving and incoming inspection/test
  • Paying sales taxes on not-for-resale items
  • City tax & business license(s)
  • Publishing DBA (fictitious business name)
  • gross margin as cost of sales
  • Marketing, advertising


cost drivers are often
  • enclosure
  • power supply (unless this is to be user-supplied)
 
FWIW, I design my SMD parts to be on one side, the PTH on the other side.

Maximize the number of SMD parts, where practicable. They are faster to place and secure.

Test points on the board do not have to cost anything other than just a little board space.

Similarly, consider using connector-less systems like the tag connect for programming and diagnostic purposes. Who says it has to be just for JTAG- why not USB / Serial/ etc. also? Pauls bed of nails is a good illustration of how to program without connectors at a higher manufacturing scale.
 
Thanks for all the good ideas,

One thing i will change is adding test points. I currently have a company and make similar products so some items mentioned I have considered before, thinking about it now some carefully placed test points would have made some trouble shooting a lot easier on my previous projects. my ideas on test points would:

- pos / neg power supplies
- between op amps on my input circuit
- parts that will be burred under the Teensy

I designed the product so i could use commonly available parts all from Newark and eBay , so i don't think 3D printed custom parts will be necessary. I had a target price of 100$, but going through it now looks more like 140$ the board and assy represents the majority of the cost at my production levels. That is why my focus has been moving to all smd parts and a 2 layer board with all parts on one side. I think i will have a couple parts left on bottom side but i will just put those on myself. The enclosure is an extruded aluminum box so it is only 7$ the display is going to be a nokia 5110 eBay special and there like 4$.
 
Yeah, but at $5.6 the touch TFTs offered on ebay are pretty cheap and add option for instrument control. May end up being cheaper overal.
 
ebay sellers as source... long term availability OK?
watch for BOM cost vs. sell price... ratio at least 2x, often 4x in consumer items.
 
Constantin said:
Yeah, but at $5.6 the touch TFTs offered on ebay are pretty cheap and add option for instrument control. May end up being cheaper overal.
Click to expand...

That would be a neat and save costs overall because i could eliminate my currently planned interface. but i think for my application not the best, i'm making a tool for technicians that will be using gloves and in a dirty greasy environment. I was going to use a small rotary encoder with the push option to do all the interface.
 
stevech said:
ebay sellers as source... long term availability OK?
watch for BOM cost vs. sell price... ratio at least 2x, often 4x in consumer items.
Click to expand...

Its a risk for sure but its for parts like banana jack connectors, and the molex style wire leads ect, so i could find them elsewhere if needed.
 
Agree re: TFT in the intended role. Won't work. However, the displays themselves (i.e. minus the mounting board) are relatively inexpensive to source from multiple places. You simply have to be OK with soldering a flexible header wire. A XPT206 chip and a few resistors are needed for the touch readings... overall though, very little gear! And likely cheaper than other approaches, assuming the duty cycle allows the use of a touch screen.
 
Status
Not open for further replies.
Back
Top