Costs for low qty fab services?

Go To Last Post
34 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Anyone have experience with suppliers for low volume fab of say, 30 components, mix of SMD and others? Say, 50 boards?
Non-recurring and piece prices?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

stevech wrote:
Anyone have experience with suppliers for low volume fab of say, 30 components, mix of SMD and others? Say, 50 boards?
Non-recurring and piece prices?

Depends on how fine pitched the components are, and number of BOM line items (if 20 of those 30 components are the same part, cost goes way down).

I used to use Screaming Circuits, but now use PCA. They're both very good. I think Screaming Circuits might do slightly better work, but they also seem to charge you a bit more for it. You can quote your board on their website, which is nice. PCA is just a bit cheaper and (since my move) is now local.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

for 50 boards the fixed costs will dominate. I think MyroPCB charged me ~$500 for programming, stencil fee, and pcb setup. Non fixed costs were ~1.5 cents per solder.

For 30 components lets say you have 100 solder points, so $1.50 soldering fee per piece.

for 50 pieces pcb fab costs will be around $1.50 per piece.

$500/50 = $10 fixed costs per piece.

So the total would be about $13 + component costs per piece.

Personally I think once you get around 250 pieces of volume then it becomes economical to outsource assembly.

For 50 pieces 30 parts per piece, I would change as many components as I can to SMD, get a stencil + solder paste + toaster oven and do it myself. If you do not have very fine pitched parts (<0.5mm) you could probably get a 95% success rate with practice.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks. I didn't know the costs or breakpoints.
More study needed.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

toalan wrote:
for 50 boards the fixed costs will dominate. I think MyroPCB charged me ~$500 for programming, stencil fee, and pcb setup. Non fixed costs were ~1.5 cents per solder.

For 30 components lets say you have 100 solder points, so $1.50 soldering fee per piece.

for 50 pieces pcb fab costs will be around $1.50 per piece.

$500/50 = $10 fixed costs per piece.

So the total would be about $13 + component costs per piece.

Personally I think once you get around 250 pieces of volume then it becomes economical to outsource assembly.

For 50 pieces 30 parts per piece, I would change as many components as I can to SMD, get a stencil + solder paste + toaster oven and do it myself. If you do not have very fine pitched parts (<0.5mm) you could probably get a 95% success rate with practice.

Interesting topic. I'm thinking about doing some low volume runs with a small board (4" x 6") with a mix of SMD and TH. Where would I get the stencil from? I have a few components on the bottom of the board. How do I handle this?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

PCB-Pool supplies a free SS stencil with each order for SMT boards, some other PCB suppliers probably do the same. Board assemblers can supply them, but they tend to be expensive. There are low-cost stencil suppliers who use Kapton or Mylar, which are OK for building prototypes. They don't last as long as SS, of course.

Leon Heller G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As leon says PCBpool gives a free stencil with each pcb order and the quality of their pcb is superb. PCBpool gives you a metal stencil.

I find the pcbpool stencil to be a bit thick for applying solder paste by hand and a razor. My pads tend to get too much solder on them, not a big deal for most parts but I do 0.4mm qfn stuff and too much solder becomes a royal bitch. Might be that I need a special jig or better quality paste.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

devils4ever wrote:
toalan wrote:
for 50 boards the fixed costs will dominate. I think MyroPCB charged me ~$500 for programming, stencil fee, and pcb setup. Non fixed costs were ~1.5 cents per solder.

For 30 components lets say you have 100 solder points, so $1.50 soldering fee per piece.

for 50 pieces pcb fab costs will be around $1.50 per piece.

$500/50 = $10 fixed costs per piece.

So the total would be about $13 + component costs per piece.

Personally I think once you get around 250 pieces of volume then it becomes economical to outsource assembly.

For 50 pieces 30 parts per piece, I would change as many components as I can to SMD, get a stencil + solder paste + toaster oven and do it myself. If you do not have very fine pitched parts (<0.5mm) you could probably get a 95% success rate with practice.

Interesting topic. I'm thinking about doing some low volume runs with a small board (4" x 6") with a mix of SMD and TH. Where would I get the stencil from? I have a few components on the bottom of the board. How do I handle this?

I guess toaster oven the side with the most components and then hand solder the stuff on the other side.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

toalan wrote:
devils4ever wrote:

Interesting topic. I'm thinking about doing some low volume runs with a small board (4" x 6") with a mix of SMD and TH. Where would I get the stencil from? I have a few components on the bottom of the board. How do I handle this?

I guess toaster oven the side with the most components and then hand solder the stuff on the other side.


In my experience, lightweight components will stay on when inverted. So just solder the side with the heavy components second. Both sides can be done in the toaster oven.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think assembly houses use some type of glue for securing parts to the underside before it goes into the oven

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

toalan wrote:
I guess toaster oven the side with the most components and then hand solder the stuff on the other side.

I did a run of about 100 boards a while back--50 parts, 35 BOM lines, two sides. 33 parts on the top, 17 on the bottom. I chose to do the top side using a stencil, and the bottom side by hand.

All that soldering got really tedious after a while. Stenciling helps, but you'll still need to populate the board by hand, and that can take a while for 50 parts.

If I were doing a similarly small production run again, I would seriously consider whether or not I could build the cost of a contract manufacturer into the price of the product. If I can't, the question becomes, just how passionate am I about the project? What is my own time worth?

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

stevech wrote:
Anyone have experience with suppliers for low volume fab of say, 30 components, mix of SMD and others? Say, 50 boards?
Non-recurring and piece prices?

I'm interested in this topic too.

For small runs like this, are reels needed? Tubes? Trays? If and how does this affect the cost? Will they accept cut tape?

For example, I'm not going to go out and buy a reel of 1000 parts when all I need is 50 of them, just to satisfy one production house :)

-- Damien

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

damien_d wrote:
stevech wrote:
Anyone have experience with suppliers for low volume fab of say, 30 components, mix of SMD and others? Say, 50 boards?
Non-recurring and piece prices?

I'm interested in this topic too.

For small runs like this, are reels needed? Tubes? Trays? If and how does this affect the cost? Will they accept cut tape?

For example, I'm not going to go out and buy a reel of 1000 parts when all I need is 50 of them, just to satisfy one production house :)

-- Damien


Some companies will offer discounts if you have everything or most everything on reels. Other companies will refuse the job if you don't have reels. Others don't adjust their price either way.

Generally, small local houses that specialize in low volume won't complain about cut tape, while larger places that do higher volume will be unhappy.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
the question becomes, just how passionate am I about the project? What is my own time worth?
Indeed, for an avocation-to-semi-vocation endeavor, that is the question. I see so much low volume stuff in hobby/student-lab markets that I asked here, in hopes there's a magic answer as to how to do lot 1 and 2 and hope for escalating sales. But the catch-22 is the sell price of lot 1.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

stevech wrote:
But the catch-22 is the sell price of lot 1.

One more thought...

When I initially talked to a contract manufacturer, I was surprised to see that the per-unit cost going through them was about the same as doing it myself. Basically, they saved a bit of money buying parts cheaper, and then charge about that much back to me for labour.

So, suppose that your per-unit costs are roughly the same either way--say, $20. You'll want to sell them for about 2.25 times that. This allows a 50% margin for you, and another 50% for resellers. So your price is then about $45.

Initially, you may not use resellers, but I still think it's a good idea to include them in your price. If you decide to use them later on, and have to increase the price because of it, your customers may wonder why the price suddenly jumped. It just doesn't look good.

Suppose you don't use resellers initially, so you can pocket both margins. Then your profit per-unit will be about $25. If you do a run of 50, that's $1250.

Here's what I'm thinking: Presumably the number 50 came up because you're fairly sure you can sell that many. If you keep your costs low initially by selling them yourself, then you can use the $1250 profit to pay the contract manufacturer's setup costs. If the product flops, then you should roughly break even. If it goes well, then you'll already be set up for large-scale manufacturing, and the next batch of boards is just a phone call away.

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Another consideration is that since it is new product and you want to use the product to build business, you want to put your best foot forward.

It is not realistic that you can by yourself with a stencil + skillet/toaster oven can rival the quality of a dedicated machine run by professionals that do it for a living. I am talking in terms of the look of the assembled pcb and also reliability.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

crwper wrote:
stevech wrote:
But the catch-22 is the sell price of lot 1.

One more thought...

When I initially talked to a contract manufacturer, I was surprised to see that the per-unit cost going through them was about the same as doing it myself. Basically, they saved a bit of money buying parts cheaper, and then charge about that much back to me for labour.

Michael

This is true especially for asian based assembly places, they can find passives and connectors for dirt cheap, even for brand name ICs they can typically undercut digikey/mouser/etc.. prices considerably. However giving the assembly free reign on sourcing parts can easily bite you in the ass since they will typically source the component based on price and not on quality.

Generally for small quantity, assembly places do not want to go through the hassle of sourcing components for you, they just want a digikey parts list and source from digikey.

I get myro to source stuff that I feel is vastly overpriced on digikey/mouser/etc... and there is not much room for poor quality. So far I have only done this for pinheaders.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

toalan wrote:
It is not realistic that you can by yourself with a stencil + skillet/toaster oven can rival the quality of a dedicated machine run by professionals that do it for a living. I am talking in terms of the look of the assembled pcb and also reliability.

This is a huge consideration. In the first 100 boards I produced by hand, failure rate was about 5%, largely due to the 0.5 mm QFN AVR. In the 200 boards produced since then by the contract manufacturer, I have not yet run into a manufacturing error.

toalan wrote:
However giving the assembly free reign on sourcing parts can easily bite you in the ass since they will typically source the component based on price and not on quality.

One thing that may help here is to specify what can be substituted in the BOM. I allowed my local manufacturer to substitute basic passives freely, a few mid-level components with approval, and only a handful of top-level parts not at all. This gives them a great deal of flexibility in sourcing, while maintaining some control over the critical components.

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
It is not realistic that you can by yourself with a stencil + skillet/toaster oven can rival the quality of a dedicated machine run by professionals that do it for a living. I am talking in terms of the look of the assembled pcb and also reliability.

I do it for a living and I would not dismiss manual smd assembly so quickly. The critical issue is quality printing of the solder paste. If a stencil is used, you can have a high quality attractive pcb whether the parts are placed with a machine or a pair of tweezers. I manufacture boards similar to what the OP describes, and lots of them. My initial production was by hand, and it took me 5 minutes to place a board, so I could easily do 50 units a day. The irritating thing about using tweezers is that when you arrange little piles of resistors onto a table, half will be upside down so you have to flip them over using a micro-tiddlywinks technique. Using my CSM84VZ P&P, I crank out 200 units in three hours, including reflow and all inspections. My two computer controlled batch ovens (aka "toaster" ovens) can easily reflow the boards faster than what the P&P can place, so a conveyor oven is not really needed. Hand placed pcbs should be indiscernible from machine placed ones.

ps: with about 15k various units in the field, failure rate so far is exactly zero

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

tpappano wrote:
Quote:
It is not realistic that you can by yourself with a stencil + skillet/toaster oven can rival the quality of a dedicated machine run by professionals that do it for a living. I am talking in terms of the look of the assembled pcb and also reliability.

I do it for a living and I would not dismiss manual smd assembly so quickly. The critical issue is quality printing of the solder paste. If a stencil is used, you can have a high quality attractive pcb whether the parts are placed with a machine or a pair of tweezers. I manufacture boards similar to what the OP describes, and lots of them. My initial production was by hand, and it took me 5 minutes to place a board, so I could easily do 50 units a day. The irritating thing about using tweezers is that when you arrange little piles of resistors onto a table, half will be upside down so you have to flip them over using a micro-tiddlywinks technique. Using my CSM84VZ P&P, I crank out 200 units in three hours, including reflow and all inspections. My two computer controlled batch ovens (aka "toaster" ovens) can easily reflow the boards faster than what the P&P can place, so a conveyor oven is not really needed. Hand placed pcbs should be indiscernible from machine placed ones.

ps: with about 15k various units in the field, failure rate so far is exactly zero

I think you are the exception rather than the rule. I can always tell if a product is hand soldered vs machine soldered.

I find it tough to be consistent with applying the solder paste, never mind placing the components so they are straight and the text are aligned in the same direction. I think you need a $300 jig for applying paste with a stencil, my razor blade on a flat table ain't cutting it.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

crwper wrote:

This is a huge consideration. In the first 100 boards I produced by hand, failure rate was about 5%, largely due to the 0.5 mm QFN AVR. In the 200 boards produced since then by the contract manufacturer, I have not yet run into a manufacturing error.

Michael

5% is fantastic with QFN, I run into 10% with 0.4mm QFN. I go through each qfn chip after wards with a fine iron and flux, and I still get ~10% failure.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I think you are the exception rather than the rule. I can always tell if a product is hand soldered vs machine soldered.

Just to clarify, I'm saying with a stencil and reflow process of some sort you should be able to attain "machine built" appearance and performance. I agree that hand soldering with an iron is tedious and yields an obvious hand made appearance.

Quote:
I think you need a $300 jig for applying paste with a stencil, my razor blade on a flat table ain't cutting it.

I'm not using any sort of commercial stencil printing fixture as yet because bench and floor space is at a premium right now. I use scraps of pcb taped to a table as guides for holding the pcb in position. The stainless steel stencil is also attached to the table with a strip of tape forming a hinge. Results are very good when reasonable care is taken by the operator. The paste needs to be the proper consistency, though. The paste I use will gradually go from a nice "creamy" texture to a "doughy" consistency which signals the end of its life. When it becomes doughy, it just won't print and must be discarded.

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Sorry, when I said hand soldered I meant not machine soldered.

I can get close to the appearance of machined soldered stuff with my stencil and paste, but I have done it alot of times and have had time to perfect it. My first few boards looked like absolute crap.

When I used AMTEC paste, it would get all runny when it got old. When I used some no name paste the stuff doughy but not runny. I keep my paste in the fridge, and when I use it I place it next to an icepack. Colder temps seem to help for both old and new paste to keep a good consistency when applying.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The paste needs to be "sheared" that is stirred to give it the correct rheology.

Paste tends to separate so that the solderballs and carrier material separate somewhat. Mixing it thoughly resoters the paste consistency to an optimal level.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This is an interesting thread. I'm also in the process of setting up small run of about 25.

My board is 5x7, 65 BOM lines, and 181 total parts with a 100 pin QFP and 32 pin QFN. I seriously considered doing them all myself, however I found the most challenging part is the PTH soldering, not the surface mount.

I was doing all my prototypes with hot air and a tweezers and it wasn't that bad. But I've got a LOT of PTH pins that just looked really amateur when completed. Each pin didn't look that bad itself but when it was completed it just had a bad look to it.

I ended up getting an assembly quote from goldphoenix for around $23 per board which I figured is much cheaper than I could do it myself. I'm hoping it all turns out well.

*fingers crossed*

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Gordon,

Does the $23 include the setup charges (NRE) that someone mentioned earlier?

Cheers,

Ross

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Ross,

Yes I believe so. It was only a preliminary quote, but I should getting a more official quote sometime today. I'll report the exact numbers when I find out.

Dan

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok I got the official quote. I requested prices for x25, x50, and x100.

Stencil base: $200
x25: $300 + $100 shipping or ~24$ per board
x50: $350 + $120 shipping or ~13.5$ per board
x100: $650 + $200 shipping or ~10.5$ per board

This also doesn't include what it will cost me to ship the components to China by DHL so I'll have to figure that out too.

Looks like there's a HUGE price break at x50 so I'm really considering going that route even though I'll have to buy twice the components up front. Nearly halving the assembly cost is very tempting.

Edit: Looks like shipping to china will be ~$100 too.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

tpappano wrote:
ps: with about 15k various units in the field, failure rate so far is exactly zero

Is this including fine-pitch QFN parts? For my boards, I was using a technique very similar to yours (spare PCBs to hold the working board in place, steel stencil). It took me about 10 minutes per board to stencil/place one side, but the reverse side (hand soldered) took me closer to 30.

All of my failures occurred on the 0.5 mm QFN part. Although I got into the habit of giving it a quick drag solder to remove bridges/dry pads left after reflowing on the skillet, it was not always possible to correct problems this way.

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

$100 shipping? Wow.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

GordonFreeman wrote:
Ok I got the official quote. I requested prices for x25, x50, and x100.

Stencil base: $200
x25: $300 + $100 shipping or ~24$ per board
x50: $350 + $120 shipping or ~13.5$ per board
x100: $650 + $200 shipping or ~10.5$ per board

This also doesn't include what it will cost me to ship the components to China by DHL so I'll have to figure that out too.

Looks like there's a HUGE price break at x50 so I'm really considering going that route even though I'll have to buy twice the components up front. Nearly halving the assembly cost is very tempting.

Edit: Looks like shipping to china will be ~$100 too.

Give myropcb a try and see what they quote you, GP and myro use a similar excel format for the data so it is not much more work.

I got a quote from GP for 250 pcs of a design, myro gave me just about the same price, I think myro was a few bucks cheaper. I went with myro because I have used them in the past and my opinion is that myro has better customer service.

Why are you supplying the components? GP can source all the stuff from digikey/mouser/etc...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

toalan wrote:
Give myropcb a try and see what they quote you, GP and myro use a similar excel format for the data so it is not much more work.

I got a quote from GP for 250 pcs of a design, myro gave me just about the same price, I think myro was a few bucks cheaper. I went with myro because I have used them in the past and my opinion is that myro has better customer service.

Why are you supplying the components? GP can source all the stuff from digikey/mouser/etc...

I'll do that, thanks for the idea. They say they will only source components for larger orders so I had to provide them all.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Is this including fine-pitch QFN parts? For my boards, I was using a technique very similar to yours (spare PCBs to hold the working board in place, steel stencil). It took me about 10 minutes per board to stencil/place one side, but the reverse side (hand soldered) took me closer to 30.

All of my failures occurred on the 0.5 mm QFN part.

I'm not using any QFNs at this time, but I do use some TSSOPs. I get an occasional bridge on a TSSOP which must be picked off post-reflow. It seems to be a slump issue with the paste I'm using. I print a whole batch, then place the whole batch, and while waiting the solder on the TSSOP pads will sometimes creep. Placing and reflow right after printing may be the simplest fix, or perhaps a thinner stencil or tweak to the aperture shape.

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

GordonFreeman wrote:
toalan wrote:
Give myropcb a try and see what they quote you, GP and myro use a similar excel format for the data so it is not much more work.

I got a quote from GP for 250 pcs of a design, myro gave me just about the same price, I think myro was a few bucks cheaper. I went with myro because I have used them in the past and my opinion is that myro has better customer service.

Why are you supplying the components? GP can source all the stuff from digikey/mouser/etc...

I'll do that, thanks for the idea. They say they will only source components for larger orders so I had to provide them all.

I think what they mean by sourcing components is doing the work to find asian sources for components. I think they will be happy if you gave them a digikey list.