When to not shave the yak

Have questions about a SparkFun product or board? This is the place to be.

Moderator: phalanx

Post Reply
kd5crs
Posts: 65
Joined: Thu Jan 14, 2010 1:54 pm

When to not shave the yak

Post by kd5crs » Thu Jul 12, 2012 1:20 pm

It seems like I saw the definition of "shave the yak" in one of the Sparkfun product posts, but it means deciding to use pre-built things vs building your own. Like, if you want to sell t-shirts then sell t-shirts, don't start by shaving a yak for wool (yak wool?).

For those of you who have done home manufacturing of small electronics as a business, here's my question: For $22 I can get a fully assembled, plug and go part for my larger assembly. It is overkill, but it does everything I need. For about $10 I can make my own (price for parts only, with no value assigned for my labor). Each of the $10 boards would require me to solder 60 through-holes.

So if it were you, at what price point for the final assembly would you switch from the build by hand to the pre-made? I mean, if I were selling these for $1000, I'd take the premade, no question. If I were selling them for $20, I'd have to make my own or I wouldn't make any profit at all.

FWIW, I hope to sell maybe 10 units a month, and I'll be doing my assembly nights and weekends as I do work full-time.

Thanks for your advice.

Joeisi
Posts: 418
Joined: Thu Apr 26, 2012 6:59 pm
Location: Lima, OH

Re: When to not shave the yak

Post by Joeisi » Thu Jul 12, 2012 1:48 pm

Through hole is quite easy to assemble. It is not as annoying as SMD to solder, thus reducing manufacturing time and effort. If you are only making a few per month, just do it yourself regardless.

AndyC_772
Posts: 195
Joined: Wed Feb 02, 2011 1:17 am

Re: When to not shave the yak

Post by AndyC_772 » Thu Jul 12, 2012 1:51 pm

Regardless of your product's sale price, it depends on how you value your time. Work out how long it takes you on average to make the part from individual components, including the time taken to order, organise and keep track of your stock. That tells you how long you're working to achieve the $12 saving.

If each part takes you 2 hours to build, that's $6/hr and I'd personally rather spend the time watching TV with my cat. On the other hand, if each part only takes you 10 mins to build then that's $72/hr in savings, and I'd be more inclined to accept a part-time soldering job at that rate.

For what it's worth, I normally prefer to make my own circuits where possible as a matter of principle. If I buy something in, more often than not it's single-sourced, and I've no idea how long it'll remain available - plus there's not much I can do about it if I have a problem with the part or its supplier. With my own circuits I have a much better understanding of how they work, what is critical and what isn't, and I can switch component suppliers if necessary.

macegr
Posts: 380
Joined: Wed Feb 27, 2008 4:44 pm

Re: When to not shave the yak

Post by macegr » Thu Jul 12, 2012 2:16 pm

Surface mount is actually faster to solder, even at home with manual placing. Get a stencil, smear down paste, place all the components on the PCB, pop in the toaster oven. Skips the following steps: bending leads, holding part in place, turning board over, solder for a couple seconds on each and every joint, clip leads, turn board back over, repeat. It is worth knowing how to do, and SMD is much easier for an outside assembler to do later on if you need more volume.

Look at it this way: Sparkfun assembles the majority of their original products. They used to do it by hand, before buying pick and place machinery. How many of those products are primarily through-hole? Get your lessons from the best.

stevech
Support Volunteer
Posts: 2907
Joined: Sat Jun 05, 2004 8:51 pm
Location: USA, California

Re: When to not shave the yak

Post by stevech » Thu Jul 12, 2012 7:59 pm

hire a starving student to assemble?

lyndon
Support Volunteer
Posts: 1641
Joined: Thu Jul 17, 2008 6:37 pm
Location: Minnesota, USA

Re: When to not shave the yak

Post by lyndon » Thu Jul 12, 2012 8:18 pm

First you need to know your total cost: part cost + labor charges at a reasonable rate for your area for building the complete board. If at this point the off the shelf unit is cheaper, then you should probably go with that.

But even if it's more expensive, there is more to consider. By purchasing an off the shelf unit, you may not need to have any more than minimal inventory of one part. If you build them yourself, then you need to inventory multiple parts and in order to get good pricing on those parts, you need to buy relatively large quantities of them. In other words, you may need to purchase all parts at a minimum of 1,000 each in order to be able to build them for the price that the off the shelf unit is sold for. So the more expensive OTS part may actually be saving you money in not tying up your capital in inventory that you may never move.

After doing that, now consider quality and support. Is the OTS unit up to your build quality? Is it likely to cost you more in support than if you built something yourself. Admittedly, when you're selling to hobbyists, this may not matter much, but if your customers are using your product to make money, they will expect good support and high quality.

This is a very good question, BTW. I wish there were more discussions like this on the web.

kd5crs
Posts: 65
Joined: Thu Jan 14, 2010 1:54 pm

Re: When to not shave the yak

Post by kd5crs » Fri Jul 13, 2012 1:06 pm

Thanks, very good points. Some answers:

It is all through-hole because that's what I know and that's the design I got working. Surface mount seemed harder to get started with, but I'll get there one day.

I can build each one in 10 minutes. The soldering is actually not the slow part, it is populating the little board. I have to take all the shortest components, put them in place, flip the board, and solder. Then I get the next tallest, repeat, etc. One barrel-type capacitor is the tallest part, so it makes up the fourth flip. Each flip is a chance to drop it all on the floor. I guess if I wanted to go full speed I would get a piece of wood and drill it so that all the components fit in there upside down at the correct height, then I put the board on top and go to solder town.

The whole assembly is going to be potted, so if someone sends one back because it broke I think I'm out the entire value of the assembly regardless, right? There's really no good way to un-pot something, as far as I know.

All my prototypes have been assembled by me, so I've ordered some of the off the shelf boards to try that way.

lyndon
Support Volunteer
Posts: 1641
Joined: Thu Jul 17, 2008 6:37 pm
Location: Minnesota, USA

Re: When to not shave the yak

Post by lyndon » Fri Jul 13, 2012 1:13 pm

How are you holding the parts in place when you flip? I found that if you use thick foam rubber, you can hold a pretty wide variation of component heights in place. The foam has to be heat resistant of course, and it may take some force to hold everything in place but it does help.

That said, when you transitionto SMT you'll be amazed at how much faster it is with no wasted motions.

AndyC_772
Posts: 195
Joined: Wed Feb 02, 2011 1:17 am

Re: When to not shave the yak

Post by AndyC_772 » Fri Jul 13, 2012 1:42 pm

lyndon wrote:That said, when you transitionto SMT you'll be amazed at how much faster it is with no wasted motions.
Definitely +1 on this, through-hole is really only used by hobbyists and for prototypes these days. SMT is not only quicker to assemble by hand with practice, it's quicker and easier to replace parts too. Throughout my career I've done a lot more damage to PCBs from trying to desolder through-hole components than from replacing SMT, and I've changed a lot more SMT parts than through-hole.

Joeisi
Posts: 418
Joined: Thu Apr 26, 2012 6:59 pm
Location: Lima, OH

Re: When to not shave the yak

Post by Joeisi » Fri Jul 13, 2012 5:21 pm

AndyC_772 wrote:
lyndon wrote:That said, when you transitionto SMT you'll be amazed at how much faster it is with no wasted motions.
Definitely +1 on this, through-hole is really only used by hobbyists and for prototypes these days. SMT is not only quicker to assemble by hand with practice, it's quicker and easier to replace parts too. Throughout my career I've done a lot more damage to PCBs from trying to desolder through-hole components than from replacing SMT, and I've changed a lot more SMT parts than through-hole.
I will go back on my comment earlier and say yes, SMT is the way to go. Now that I think of it, SMT should be faster. I just design weirdly (I put components on both sides. My applications are very space demanding). For large production runs SMT uses less solder and such, reducing cost further. I sometimes cheat and just hot air everything, applying solder with the other hand. Gives you more room to look. I myself have odd practices so you probably shouldn't follow much of what I say. When prototyping, as said above, it takes maybe 10 seconds to remove something small(like a 0603 chip, or a SMT v-reg) and not much longer to remove a larger component (QFN, TSSOPs, etc.). So prototyping that way is painless. DIP packages have taken me a few minutes sometimes. Production in large volumes with a Pick-and-place machine usually tend to be SMT designs due to the above mentioned bending of leads, and the large object to heat up.(Since it is a larger chunk of plastic, it needs more energy to heat up, and thus more time wasted).

Consider all of the above, and the fact that SMT sometimes cost less (both money and PCB space wise), SMT is the way to go.

And SMT makes you look like a pro.

kd5crs
Posts: 65
Joined: Thu Jan 14, 2010 1:54 pm

Re: When to not shave the yak

Post by kd5crs » Mon Jul 16, 2012 12:14 pm

lyndon wrote:How are you holding the parts in place when you flip? I found that if you use thick foam rubber, you can hold a pretty wide variation of component heights in place.
I've been using a flat piece of cardboard, but I have such a piece of foam. I'll try that.

So I guess the consensus is that by switching to a surface mount design, the resulting improvements in manufacturing (and board cost, can't forget board costs) would drop the make-it-myself price and difficulty down to no-brainer levels, vs the pre-assembled overkill part. I'm inclined to agree with that.

Quite honestly, even with the design complete, several working prototypes including some that have been in daily use for 2 years, and a fair amount of money in the bank, it still seems like actually selling something is lightyears away. I gotta say though, I'm going to be so dang proud the first time I sell one.

Post Reply