Setup costs are the big expense. In a low-volume run, P&P programming, parts procurement and dealing with any test fixtures will easily comprise 80% of the total cost. It's necessary to have volume to spread these setup costs across a larger number of units.
If PropKey BOM is $25 and you want to build 100 units, a contract manufacturer will charge $15 labor per unit and $1200 setup cost. That works out to be $42/unit.
Even with the same setup costs, labor per unit and BOM costs, 1,000 units works out to be $31/unit.
But as Perry pointed out, you need volume to make this happen (plus a lot of cash that nobody has lying around). So what's the solution? There are at least five of them:
(a) Human P&P plus home reflow oven. Apply solder paste with dispenser. Yuck.
(b) Find a friend who agrees to run them for you. Some friend!
(c) Send it to a Chinese assembly house, telling them you're just testing their setup for larger volume with a trial run. Not necessarily ethical, but works.
(d) Design the PCB for hobby SMT assembly
(e) Place your parts manually, push them into our reflow oven between other boards.
Did I read somewhere that the PCB has parts on both sides? I hope not. . . please tell me it's not true. Dreaded 2nd op.
I already priced boards through DorkbotPDX, they have a medium run service that costs $1 per sqin, which is very reasonable.
I found a desktop reflow oven on Ebay today, $264 as shipped for an 18x23.4cm reflow area. It's IR with profile control, which looks like a good solution for small volume home assembly.
I *can* kit these up relatively easily and already had an idea of how to do it efficiently and with proper QA before I even thought of selling them.
I think the right way to go is to do a partial kit, where the ICs and uSD socket are reflowed by me and I supply a passives kit for the recipient to solder.
Setup costs are the big expense. In a low-volume run, P&P programming, parts procurement and dealing with any test fixtures will easily comprise 80% of the total cost. It's necessary to have volume to spread these setup costs across a larger number of units.
If PropKey BOM is $25 and you want to build 100 units, a contract manufacturer will charge $15 labor per unit and $1200 setup cost. That works out to be $42/unit.
Even with the same setup costs, labor per unit and BOM costs, 1,000 units works out to be $31/unit.
But as Perry pointed out, you need volume to make this happen (plus a lot of cash that nobody has lying around). So what's the solution? There are at least five of them:
(a) Human P&P plus home reflow oven. Apply solder paste with dispenser. Yuck.
(b) Find a friend who agrees to run them for you. Some friend!
(c) Send it to a Chinese assembly house, telling them you're just testing their setup for larger volume with a trial run. Not necessarily ethical, but works.
(d) Design the PCB for hobby SMT assembly
(e) Place your parts manually, push them into our reflow oven between other boards.
Did I read somewhere that the PCB has parts on both sides? I hope not. . . please tell me it's not true. Dreaded 2nd op.
Ken Gracey
LOL, the dreaded 2nd op!. Well, there wasn't room on the 1st side for everything, at least with the first round. I reflowed the bottom side with a hot plate (lab style, not walmart) and a thermocouple to get the proper temps. The first try worked out really well, especially because I knew I couldn't solder the uSD connector with it's pad layout.
I'm pretty ingenious, so I think I can make this work out for the main components. I have to design the process so my wife can do it at home, so there is some curve!
I had a machine shop for several years and figured out how to exploit my software programming background to make setups and "1 off" parts economical for me. The trick is in fixtures and standardization.
I'm thinking of making a board holder and plastic templates for placing the components. I can also have some paste templates made to do it "properly".
The way I see it:
*Apply paste on bottom with template
*mount board in array holder
*lay plastic template over array
*carefully place SMT parts in respective cutouts
*remove template
*reflow
*flip and apply paste to other side
*put in holder
*apply template
*place SMT parts
*remove template
*reflow
*INSPECT with microscope
*rework (if needed)
I figure 5 or 6 boards in array, laser cut polymide stencil, laser cut plastic template.
I have found that with care you can freehand place the components, as long as it doesn't take forever, this would probably work, then a simple board holder made from FR4 could be used.
A year or so ago at my work some board mfg sent in a sample of a PCB that already had the solder paste on it. It was a special paste and they shipped them with plastic over the paste. You basicly peeled off the plastic and started placing parts.
I can't find it but you should be able to google them.
When I did reflow I used a hot air wand that had a large (about 1") end. It worked really well. You could see what areas hadn't reflowed yet and concentrate the heat on them. This would work even better if you had the PCB on a hot place set to about 100°C.
I would suggest selling as kit, with a markup ($20 ???) to be populated. Then you can just populate them as they are ordered. There is nothing worse than populating a bunch of board and then not selling them.
I communicated with the assembly house more today. The basic gist is that I am stuck at $31-33 each in q100 for the PropKey. What I'm going to do is make an order of ~75 boards with DoPDX and hand assemble those at home. I expect the pilot run to give me a good idea of the demand, then if it is enough, I'll redesign the board. On the new board I'll try to go with QFN parts and smaller passives to get everything on the top side, for easier assembly. I stayed away from the QFN parts because I wanted to be able to hand solder/assemble the boards, but if I know I'm going to have a fab house make them, I'll make it easier to build.
I've got a RAM/FLASH board that I need to send off and test, but it should work real nice with the XMM modes in PropGCC, and in SPIN/PASM if anyone wants. The tentative specs of the RAM/FLASH board are: 512KB 8xSPI SRAM, and 16MB of 2xqSPI FLASH. That's 8-bit wide flash and RAM that can be SPI controlled, including the /HOLD line on the RAM so you can setup a transfer, pause, then clock the transfer on a common bus. This would allow you to do DMA to an LCD display, or stream video data to VGA. You would setup a transfer from RAM or Flash, then run a counter at whatever dot clock you want, to stream the data out of the memory. You can also stream to an LCD this way: setup the RAM transfer, assert /HOLD, setup the LCD to receive, then de-assert /HOLD, and clock the data.
A thought.. Could you make your project as SMT friendly as possible in your design and then push it out as a kit? This might be the perfect project for someone to learn SMT soldering at home isn't as bad as they think.
Yes, I want to offer it as a kit too. There are a couple of difficult parts to solder, I'll put those on first, then ship a kit of parts with instructions.
Once I get this rolling, I want to offer another kit that fits in an Ice Breakers mint container. The new containers are plastic, 76mm dia, and ~14mm deep. I was thinking of designing a PTH kit that has the USB chip pre-soldered, but has a USB type B vertical PTH connector, 40pin DIP, LEDs, and a small prototype area. The USB connector is about the same size as the "to share" flap, and the proto area/LEDs will be under the "not to share" flap. You could store some short wires and stuff instide the container and close the flap, easy to carry around or deploy innocuously.
I'm curious how hard 2-sided reflow actually is...
I recently tried baking a big 128-pin QFP upside down because I wanted to remove it (and didn't have the hot air gun handy).
But, the thing wouldn't come off!
Wonder now if it's really easy to do 2-sided because things on the bottom don't fall off...
On the new board I'll try to go with QFN parts and smaller passives to get everything on the top side, for easier assembly. I stayed away from the QFN parts because I wanted to be able to hand solder/assemble the boards, but if I know I'm going to have a fab house make them, I'll make it easier to build.
Remember with a QFN part you can't place vias (other than to ground) under the chip. I counted (not carefully) 11 vias under the Prop on your board.
In my very limited experience, I've found QFN parts with a center pad, don't offer much of a space savings because of the lost area to place vias.
I'm sure you already know this, but I've been amazed at how difficult QFN parts are to solder compared to QFP parts.
It sounds like reflowing a second side isn't the issue I thought it would be. I had considered (with my planned double sided board) using lead free solder for the first side and then lead solder for the second. I figured the lead free shouldn't melt while the leaded side is reflowing. But it sounds like it's fine to use the same solder on both sides of the board.
Remember with a QFN part you can't place vias (other than to ground) under the chip. I counted (not carefully) 11 vias under the Prop on your board.
In my very limited experience, I've found QFN parts with a center pad, don't offer much of a space savings because of the lost area to place vias.
I'm sure you already know this, but I've been amazed at how difficult QFN parts are to solder compared to QFP parts.
It sounds like reflowing a second side isn't the issue I thought it would be. I had considered (with my planned double sided board) using lead free solder for the first side and then lead solder for the second. I figured the lead free shouldn't melt while the leaded side is reflowing. But it sounds like it's fine to use the same solder on both sides of the board.
I ordered one of those T-962 desktop reflow ovens this week. I'm going to assemble a Propkey using hand assembly when it arrives. Then I can assess any problems or redesign issues and go from there.
I'm on the fence about a redesign of the board, what I've got now, I'm 99.9% sure works fine, it's just more difficult to assemble because it's got components on 2 sides. A redesign would be aimed at putting all components on the top side, so it is a single step. To be honest, the assembly cost wasn't the largest expense of making batches of these, the components and boards were the bulk of the cost of the assembly quote.
I'm leaning towards just sticking to the known design and moving forward. I've got a memory card that I want to test, and I need a fixed target board for that.
I'll post an update once I've used the new oven. I'll probably order prototypes of the memory board when I make a medium run order for the PropKey.
pedward: Nice. I will be interested to hear how the T-962 performs. Presume its not the larger A version. I am interested to see how smoothly the drawer works.
For double sided, they use a red IR type glue dot on the first side and reflow. This effectively glues the part to the pcb as part of the reflow procedure. Then they turn the pcb over and do the second side.
I have a few suggestions as a compromise. There are only a few passives anyway, convert to 0604 for easier tweezer placement. For the first versions, don't try to do a double sided with glue... rather sandwich two pcbs together and use headers to tie each side together. You couldn't use both top and bottom parts of the USB insert, so chop off the part of the PCB that doesn't have the traces on it for contacts. Panelize the boards in sets of around 25. Build a hand rest out of 1/2" aluminum tubes spaced out on 2" rows so that you can rest your hand over any area of the panel when placing boards. Use http://www.hitechstencils.com or other company to make a stainless stencil for the panel. Leaded solder is easier to work with at first, manncorp.com has a no clean leaded that works nice. You could crank out 25 sets in a few hours easily. If sales pick up, then maybe it will justify some outsourcing for fabrication.
Would you be able to add the avaliablity of having an external power source? Or did I miss something in the "schematics"? Is vsub the substitute power source? Other wise, I think this is great.
How much would one cost (pre-assembled... I don't know how to SMD solder)?
Or just make the PCB a little wider, who cares if it is slightly wider than a USB drive. 0603 is just as easy to hand solder as 0805, a lighted magnifier on a swivel arm helps.
Actually I find the resnets easier to hand solder than the 0805 because they sit flat and dont want to tombstone. The 0805 resistors are actually fine, but the larger (as in higher) 0805 capacitors are more of a problem. I wear magnifying glasses - I got them from my jeweler brother.
I received my T-962 desktop reflow oven today and assembled a double sided propkey entirely with reflow. The process was basically what I expected. I did the bottom side first, then went to the bench and populated the top side. I had an old CPU I put in the oven, pin side up, to rest the board on so it would lay flat, and I didn't have any problems with bottom side components during the second operation. Initially I thought I had a problem because BST wouldn't detect the prop, but it turned out I had minicom running in another terminal and it was taking the port.
Now that I have 2 Propkeys assembled here, I officially have a Propkey cluster!
The T-962 seems to be a sturdy and well constructed device. The complaints I read about excessive fumes and heat on the counter were not my experience. The firmware is definitely rough. It appears to do a crude PI control on heat up, but during cooldown it really could use some help. It oscillates between running the fan to keep the actual within 10degC of target, but then it overshoots and turns on the heaters!
The oven works for the intended purpose, and the tray and interior are all made of stainless steel, quite sturdy too. The slides on the drawer are a little stiff, but not jerky. I suspect loosening the bolts and allowing it to realign may help, that or just working the slides a bit. It appears to have a pair of ball bearing slides, but I didn't look real hard.
I read that it uses an AVR to run the oven, but others said ARM. Perhaps when I get bored I might put a prop in it!
After assembling 1 unit by hand, I think it could be doable to hand assemble at home, but I'll need to get a solder stencil for sure, the syringe method didn't work that great.
I'm using some Amtech 63/37 no-clean solder I've had in the fridge for 10 years It seems to still work fine, but I may consider getting some more paste in a jar for stencil use.
I really want to build a PnP machine, and I've got some ideas I shared with the guy that started OpenPNP. I just need to get a couple of tools moved into my shed so I can do some prototyping.
For hand soldering, a fine solder and a temperature controlled iron will be fine. But the real secret here is to use a flux pen to wipe on the pcb and the underside of larger IC pins. They then solder quite well and after a little practice you will not require solder wick to take off the excess. Any bridges can be easily wiped by your iron after removing sloder on the sponge first. I dont use a stencil for hand soldering.
Interesting comments regarding the T-962 software. If its an AVR then you can reprogram with the prop. But I agree, probably easier to control with a prop
I've got a temp controlled iron and some .020 solder, but that's a bit heavy for the fine pitch stuff. Sure, it works, but it isn't pretty. I was using solder paste applied with a syringe and cut down 20ga needle. It worked, but was a little messy and difficult to control because the force required to get the paste to come out was high, and then the paste was a little finicky. I can get a Polymide stencil for a reasonable price, the chinese folks sell metal stencils for $35, but it's chemical etch and that company has some quality issues. The guy who does the Polymide stencils uses a laser cutter to make them, so they are fairly accurate.
I am using .020" solder. As long as the older joint is good, then a little too muchsolder is not a problem for this market. If I was selling the boards commercially for $x00's that would be a completely different matter.
Yes I understand the polymide stencils are fine for low volumes.
Another tool we need is a prop controlled syringe paste delivery system
Well, I successfully produced 2 boards tonight, the solder paste application proves to be the most hassle. I used a large bore tip tonight and that was much easier, but messy. I really need a stencil for these.
The reflow oven works quite well. I still haven't had any problems with the bottom side on the second operation. The problem with too much solder seems to be that components move around too much, so I had a bit of hand rework, and solder bridges, to fix.
Placing the components is pretty simple, especially since I picked similar part values for many of the components.
Next up on the list is to order a medium run batch of boards, a stencil, and components. I should also make a test fixture to ensure all the I/O's are operating correctly and not bridged or open. I'll look for some solder pogo pins.
Also, thanks to Tubular's information, I'm going to pick up the 2.65mm tall uSD sockets, making it much less likely to break a uSD card where it plugs in.
For those few parts and just for a handful of boards, I would not use a stencil or oven. I would use a combo of rework station and pencil. Radio Shack sells a very fine silver bearing solder that is great for SMT work. I would put some flux on the pads, use the fine solder and coat the pads, then place parts with tweezers and use hot air from a rework station. If you were doing multiples on a panel and had a stainless stencil then it would be much easier.
T Chap, I'm trying to work out a usable small scale production solution. I want to offer the PropKey, but commercial manufacture is too expensive and requires a good chunk of capital that I don't have. Ironically, the most expensive single component in the commercial quote was the PCB. At the end of the day, I'll probably make around 75-100 PropKey's at home and re-evaluate commercial production. I'm also working towards an OpenPnP solution for home assembly.
I can relate. I created an app for a PNP that drives a Prop board via USB, the board is dedicated to 3 steppers on a CNC. The board is switchable from USB to parallel port so CNC software can be used or the PNP app. I export the XY coords out of eagle for the boards, load them in the app, save the board with a name. I have a wall with cut tape in strips of 15" pieces, much faster and easier to manage that loading full reels. The tape is manually incremented with a push button, and manually loaded for each tape. Press a but and the part gets placed, increment to the next part and place it or load the next tape. It is slow compared to a real PNP machine, but is a lifesaver since all you do it insert the tape, hit Place Part. An hourly employee runs it, so for example a part that has 125 passives takes an hour, and my cost is under 15 in labor for the board, which I cannot get close to for small volumes with anybody else. Sunstone makes the boards in 12" x 14" panels, with various boards on a panel. They are a 2 day turn, and all I have to do to order is upload the Eagle .brd file only and specify how many. Hi Tech Stencils is pricey but work great.
I use a Rockwell Bladerunner to cut the boards out, and a belt sander to clean the edges.
If you have a CNC, you can rig up a PNP with a vacuum and syringe, I have some of the sites for the needles if you need. With a cheap benchtop CNC you can crank out your small boards in a day, easily 75 to a 100 in a day or 2 if there were panelized.
TChap, I had a MaxNC 15 once, terrible quality on the spindle assembly. The collet runout on the 1/8" shank was > .005 and caused a load of broken tools. Of course I wasn't aware of this until years later when I fired it up before selling it, then I said to myself: "No wonder I had so many problems!".
I'd recommend Sherline, although the fancy anodizing and purpose-built design hooked me when I bought it! ;(
I'm going to build a gantry design with linear rails on an 18x18 piece of Aluminum. I've already outlined the design, I just need to prototype it. The intention is to have a bunch of space at the front for either reel feeders or cut tape, both will feed with the same method. Then room for trays, with a fixed X/Y zero set of rails to form the left and top sides of the fixturing system. I don't want to go into too much detail at the moment, but once it's ready to show, I'll post.
Comments
If PropKey BOM is $25 and you want to build 100 units, a contract manufacturer will charge $15 labor per unit and $1200 setup cost. That works out to be $42/unit.
Even with the same setup costs, labor per unit and BOM costs, 1,000 units works out to be $31/unit.
But as Perry pointed out, you need volume to make this happen (plus a lot of cash that nobody has lying around). So what's the solution? There are at least five of them:
(a) Human P&P plus home reflow oven. Apply solder paste with dispenser. Yuck.
(b) Find a friend who agrees to run them for you. Some friend!
(c) Send it to a Chinese assembly house, telling them you're just testing their setup for larger volume with a trial run. Not necessarily ethical, but works.
(d) Design the PCB for hobby SMT assembly
(e) Place your parts manually, push them into our reflow oven between other boards.
Did I read somewhere that the PCB has parts on both sides? I hope not. . . please tell me it's not true. Dreaded 2nd op.
Ken Gracey
I found a desktop reflow oven on Ebay today, $264 as shipped for an 18x23.4cm reflow area. It's IR with profile control, which looks like a good solution for small volume home assembly.
I *can* kit these up relatively easily and already had an idea of how to do it efficiently and with proper QA before I even thought of selling them.
I think the right way to go is to do a partial kit, where the ICs and uSD socket are reflowed by me and I supply a passives kit for the recipient to solder.
I think I could offer them for $40 as shipped.
How does this sound?
LOL, the dreaded 2nd op!. Well, there wasn't room on the 1st side for everything, at least with the first round. I reflowed the bottom side with a hot plate (lab style, not walmart) and a thermocouple to get the proper temps. The first try worked out really well, especially because I knew I couldn't solder the uSD connector with it's pad layout.
I'm pretty ingenious, so I think I can make this work out for the main components. I have to design the process so my wife can do it at home, so there is some curve!
I had a machine shop for several years and figured out how to exploit my software programming background to make setups and "1 off" parts economical for me. The trick is in fixtures and standardization.
I'm thinking of making a board holder and plastic templates for placing the components. I can also have some paste templates made to do it "properly".
The way I see it:
*Apply paste on bottom with template
*mount board in array holder
*lay plastic template over array
*carefully place SMT parts in respective cutouts
*remove template
*reflow
*flip and apply paste to other side
*put in holder
*apply template
*place SMT parts
*remove template
*reflow
*INSPECT with microscope
*rework (if needed)
I figure 5 or 6 boards in array, laser cut polymide stencil, laser cut plastic template.
I have found that with care you can freehand place the components, as long as it doesn't take forever, this would probably work, then a simple board holder made from FR4 could be used.
I can't find it but you should be able to google them.
When I did reflow I used a hot air wand that had a large (about 1") end. It worked really well. You could see what areas hadn't reflowed yet and concentrate the heat on them. This would work even better if you had the PCB on a hot place set to about 100°C.
I would suggest selling as kit, with a markup ($20 ???) to be populated. Then you can just populate them as they are ordered. There is nothing worse than populating a bunch of board and then not selling them.
Bean
I've got a RAM/FLASH board that I need to send off and test, but it should work real nice with the XMM modes in PropGCC, and in SPIN/PASM if anyone wants. The tentative specs of the RAM/FLASH board are: 512KB 8xSPI SRAM, and 16MB of 2xqSPI FLASH. That's 8-bit wide flash and RAM that can be SPI controlled, including the /HOLD line on the RAM so you can setup a transfer, pause, then clock the transfer on a common bus. This would allow you to do DMA to an LCD display, or stream video data to VGA. You would setup a transfer from RAM or Flash, then run a counter at whatever dot clock you want, to stream the data out of the memory. You can also stream to an LCD this way: setup the RAM transfer, assert /HOLD, setup the LCD to receive, then de-assert /HOLD, and clock the data.
A thought.. Could you make your project as SMT friendly as possible in your design and then push it out as a kit? This might be the perfect project for someone to learn SMT soldering at home isn't as bad as they think.
OBC
Yes, I want to offer it as a kit too. There are a couple of difficult parts to solder, I'll put those on first, then ship a kit of parts with instructions.
Once I get this rolling, I want to offer another kit that fits in an Ice Breakers mint container. The new containers are plastic, 76mm dia, and ~14mm deep. I was thinking of designing a PTH kit that has the USB chip pre-soldered, but has a USB type B vertical PTH connector, 40pin DIP, LEDs, and a small prototype area. The USB connector is about the same size as the "to share" flap, and the proto area/LEDs will be under the "not to share" flap. You could store some short wires and stuff instide the container and close the flap, easy to carry around or deploy innocuously.
I recently tried baking a big 128-pin QFP upside down because I wanted to remove it (and didn't have the hot air gun handy).
But, the thing wouldn't come off!
Wonder now if it's really easy to do 2-sided because things on the bottom don't fall off...
Remember with a QFN part you can't place vias (other than to ground) under the chip. I counted (not carefully) 11 vias under the Prop on your board.
In my very limited experience, I've found QFN parts with a center pad, don't offer much of a space savings because of the lost area to place vias.
I'm sure you already know this, but I've been amazed at how difficult QFN parts are to solder compared to QFP parts.
It sounds like reflowing a second side isn't the issue I thought it would be. I had considered (with my planned double sided board) using lead free solder for the first side and then lead solder for the second. I figured the lead free shouldn't melt while the leaded side is reflowing. But it sounds like it's fine to use the same solder on both sides of the board.
That is correct -- But with some correction. On 2 sided PCB's that is correct but on 4 layer PCB's it give much more place to spell on.
I'm on the fence about a redesign of the board, what I've got now, I'm 99.9% sure works fine, it's just more difficult to assemble because it's got components on 2 sides. A redesign would be aimed at putting all components on the top side, so it is a single step. To be honest, the assembly cost wasn't the largest expense of making batches of these, the components and boards were the bulk of the cost of the assembly quote.
I'm leaning towards just sticking to the known design and moving forward. I've got a memory card that I want to test, and I need a fixed target board for that.
I'll post an update once I've used the new oven. I'll probably order prototypes of the memory board when I make a medium run order for the PropKey.
For double sided, they use a red IR type glue dot on the first side and reflow. This effectively glues the part to the pcb as part of the reflow procedure. Then they turn the pcb over and do the second side.
It should be fine without. The smt lines we used were not that vibration free - a conveyor style.
Would you be able to add the avaliablity of having an external power source? Or did I miss something in the "schematics"? Is vsub the substitute power source? Other wise, I think this is great.
How much would one cost (pre-assembled... I don't know how to SMD solder)?
Kind Regards,
John
Now that I have 2 Propkeys assembled here, I officially have a Propkey cluster!
The T-962 seems to be a sturdy and well constructed device. The complaints I read about excessive fumes and heat on the counter were not my experience. The firmware is definitely rough. It appears to do a crude PI control on heat up, but during cooldown it really could use some help. It oscillates between running the fan to keep the actual within 10degC of target, but then it overshoots and turns on the heaters!
The oven works for the intended purpose, and the tray and interior are all made of stainless steel, quite sturdy too. The slides on the drawer are a little stiff, but not jerky. I suspect loosening the bolts and allowing it to realign may help, that or just working the slides a bit. It appears to have a pair of ball bearing slides, but I didn't look real hard.
I read that it uses an AVR to run the oven, but others said ARM. Perhaps when I get bored I might put a prop in it!
After assembling 1 unit by hand, I think it could be doable to hand assemble at home, but I'll need to get a solder stencil for sure, the syringe method didn't work that great.
I'm using some Amtech 63/37 no-clean solder I've had in the fridge for 10 years
I really want to build a PnP machine, and I've got some ideas I shared with the guy that started OpenPNP. I just need to get a couple of tools moved into my shed so I can do some prototyping.
For hand soldering, a fine solder and a temperature controlled iron will be fine. But the real secret here is to use a flux pen to wipe on the pcb and the underside of larger IC pins. They then solder quite well and after a little practice you will not require solder wick to take off the excess. Any bridges can be easily wiped by your iron after removing sloder on the sponge first. I dont use a stencil for hand soldering.
Interesting comments regarding the T-962 software. If its an AVR then you can reprogram with the prop. But I agree, probably easier to control with a prop
Yes I understand the polymide stencils are fine for low volumes.
Another tool we need is a prop controlled syringe paste delivery system
The reflow oven works quite well. I still haven't had any problems with the bottom side on the second operation. The problem with too much solder seems to be that components move around too much, so I had a bit of hand rework, and solder bridges, to fix.
Placing the components is pretty simple, especially since I picked similar part values for many of the components.
Next up on the list is to order a medium run batch of boards, a stencil, and components. I should also make a test fixture to ensure all the I/O's are operating correctly and not bridged or open. I'll look for some solder pogo pins.
Also, thanks to Tubular's information, I'm going to pick up the 2.65mm tall uSD sockets, making it much less likely to break a uSD card where it plugs in.
Please post some pics when you can!
I use a Rockwell Bladerunner to cut the boards out, and a belt sander to clean the edges.
https://www.rockwelltools.com/us/BladeRunner_Saw-P1564.aspx
If you have a CNC, you can rig up a PNP with a vacuum and syringe, I have some of the sites for the needles if you need. With a cheap benchtop CNC you can crank out your small boards in a day, easily 75 to a 100 in a day or 2 if there were panelized.
I'm not sure if you know this, but SparkFun sells pogo pins.
If someone has a less expensive source, I'd like to know about it.
I'd recommend Sherline, although the fancy anodizing and purpose-built design hooked me when I bought it! ;(
I'm going to build a gantry design with linear rails on an 18x18 piece of Aluminum. I've already outlined the design, I just need to prototype it. The intention is to have a bunch of space at the front for either reel feeders or cut tape, both will feed with the same method. Then room for trays, with a fixed X/Y zero set of rails to form the left and top sides of the fixturing system. I don't want to go into too much detail at the moment, but once it's ready to show, I'll post.
And it certainly will use a Propeller!