PDA

View Full Version : On Regret...



tician
04-19-2012, 06:55 PM
...holding ~7" long strips of "tape" containing 100 pieces each of 0402-SMD resistors and capacitors, and remembering I have to place them by hand... and I still have not made/bought a solder paste stencil for the prototype boards that arrive in the next week or two.


...having two boxes of 170 pieces each of 2.4" square ~1/4W CIS solar panels sitting under my bed for >2 years, and still not having any good project ideas.


...having all the stepper motors, driver boards, extruder barrel and nozzle for a reprap for upwards of three years, and still nothing assembled and printing.


...not joining Dr. Steel's Army of Toy Soldiers before his "retirement". :(


...having been a member of the forums for over a year and a half, and still not yet built a bot for MechWarfare.


...opening the package for two 800+ lumen LED's (here (http://search.digikey.com/us/en/products/XMLAWT-00-0000-000LT50E3/XMLAWT-00-0000-000LT50E3CT-ND/2615397) and here (http://search.digikey.com/us/en/products/XMLAWT-00-0000-0000T6053/XMLAWT-00-0000-0000T6053CT-ND/2451238)) and finding that the aluminum substrate PCB's bought to mount them on do not match (PCB product page gave external dimensions of board, but not the footprint).

sarendt
04-24-2012, 11:35 AM
Sounds like a bunch of goals, not regrets...

Cheers,
Scott

tician
04-24-2012, 06:11 PM
Sounds like a bunch of goals, not regrets...
Glass half-full type?

I'm pretty sure I'm going to start a little design/idea contest thread soon to get some of those solar panels to better homes. I'm going to make a few RF-enabled, solar-powered soil moisture sensors to stick around the flower beds because I'm lazy and occasionally forgetful. I might also make a few small solar bots, but I can't think of much else and it seems like a waste for them to just sit there.

Another future regret? I bought a myKeepon on clearance over the weekend and I'm not sure why. I spent about a half-hour today to break out the power connector and I2C bus connector through the battery compartment (barrel connector on the back is some really small variant I couldn't match to a power supply at 12VDC). I shall call him "woobie" (destroyer of worlds?).

Th232
04-24-2012, 09:33 PM
...opening the package for two 800+ lumen LED's (here (http://search.digikey.com/us/en/products/XMLAWT-00-0000-000LT50E3/XMLAWT-00-0000-000LT50E3CT-ND/2615397) and here (http://search.digikey.com/us/en/products/XMLAWT-00-0000-0000T6053/XMLAWT-00-0000-0000T6053CT-ND/2451238)) and finding that the aluminum substrate PCB's bought to mount them on do not match (PCB product page gave external dimensions of board, but not the footprint).

I suppose the good news is that MCPCBs are relatively cheap. Or if you're attaching them to a copper heatsink and don't have any real space/alignment concerns, you can attach it to the heatsink with a fat piece of wire running along the length of the thermal pad. May require use of both regular and low melting point solder though.

tician
04-25-2012, 07:13 PM
Since I only ordered one of each of those LED's simply to see how bright they really were (and maybe stick one in a maglite), I think I'm just going to go with a small-ish FR4 board with a bunch of vias surrounding the thermal pad to connect the top and bottom thermal planes. Then bolt that to a chunk of aluminum plate with some leftover arctic silver between the plate and the bare thermal plane on the bottom of the board. Should work well enough since I don't currently have any power supplies that can hit the 3A max of the LED's.

Th232
04-25-2012, 07:41 PM
Sounds possible. If you want slightly better thermal conduction (and can get vias that small), place the vias directly underneath the thermal pad then fill them with solder. If you want a stack of ideas on maglite mods and haven't already seen it, CandlePowerForums is a great place to go for that kind of stuff.

tician
05-02-2012, 04:18 PM
Well, my pathetic attempt to make a cheap stencil by hand failed quite miserably, and the later attempt to add solder paste and 0402-SMD parts to the boards sitting on the stove ruined one set (too much solder paste melted off the dental picks from the heat rising off the board and it would not come off easily with solder wick). At least now I know that I can wrangle 0402 parts by hand. I am also still in a state of mild shock from the fact that I did not give myself any new scars (have burned myself with a soldering iron and grabbed hot objects on several prior occasions).

Guess I'll just bite the bullet already and buy a stencil from pololu.



Sounds possible. If you want slightly better thermal conduction (and can get vias that small), place the vias directly underneath the thermal pad then fill them with solder. If you want a stack of ideas on maglite mods and haven't already seen it, CandlePowerForums is a great place to go for that kind of stuff.
In my initial board layout I had added as many vias as I could underneath the thermal pad, but then thought about the potential for breakage and possibly too much solder wicking into the vias (lowering the part more than expected and/or balling on the opposite side). Given how much difficulty I had separating the two boards I connected with mouse bites (0.6mm drill hits spaced 1.25mm on center) and a small tab, I no longer worry about breakage. And I can always uses solder wick or some other means to remove excess solder, so I probably will follow your advice.

jwatte
05-02-2012, 05:36 PM
Internet rumor has it that if you smear flux across the board, and add a thin strip of solder paste (from a syringe) across the footprints of your parts, surface tension of the solder will do the job and actually separate into the area of each pad. I have not tried it myself, though. When I finally go SMD for myself, I'll probably just laser cut my stencils at TechShop...

tician
05-02-2012, 06:13 PM
Didn't even need the extra flux for the solder paste to behave like that. The problem was that a large blob of solder paste melted (liquified the binder but not the solder itself) off the pick I was using to apply the paste and spread the solder around. That blob just would not behave correctly, so after ~5 minutes of trying to get some 0402 resistors in place and remove the excess solder, I just gave up on it (also had trouble removing some stray solder from the holes for a connector). At that point, I was more than a bit frustrated as I had just ripped three pads off of another board trying to remove the excess solder with some solder wick (solder would not remelt and I tried to force it). Also there was the smell of something other than flux/solder and I didn't really want to find out what.


"Perhaps your only purpose in life is to serve as a warning to others"

Th232
05-02-2012, 06:17 PM
Well, my pathetic attempt to make a cheap stencil by hand failed quite miserably, and the later attempt to add solder paste and 0402-SMD parts to the boards sitting on the stove ruined one set (too much solder paste melted off the dental picks from the heat rising off the board and it would not come off easily with solder wick).

I'm sorry if I'm reading this completely wrong, but it sounds like you were trying to put the paste on while the board was on the stove? If so, I generally apply the paste and parts before subjecting the whole thing to any heat.



In my initial board layout I had added as many vias as I could underneath the thermal pad, but then thought about the potential for breakage and possibly too much solder wicking into the vias (lowering the part more than expected and/or balling on the opposite side). Given how much difficulty I had separating the two boards I connected with mouse bites (0.6mm drill hits spaced 1.25mm on center) and a small tab, I no longer worry about breakage. And I can always uses solder wick or some other means to remove excess solder, so I probably will follow your advice.

If you do, you could put some solder in the vias first to fill them in and eliminate voids or balling, then reflow the LEDs on top of that. Just make sure that whatever the board's sitting on won't stick to the solder (don't ask...).


Internet rumor has it that if you smear flux across the board, and add a thin strip of solder paste (from a syringe) across the footprints of your parts, surface tension of the solder will do the job and actually separate into the area of each pad. I have not tried it myself, though. When I finally go SMD for myself, I'll probably just laser cut my stencils at TechShop...

This works fairly well for IC packages or other areas where you've got lots of pads sitting in a line, with a few observations:
* The thickness of the copper layer means solder can gather in the valleys between the pads. Recommend a good wash in ethanol afterward soldering to dissolve flux and remove all the excess solder balls. Running a needle or similar between each pad prior to placing the components can't hurt either.
* You need good paste. The stuff we use has a shelf life of 6 months, after around 8 months it's still useable for hand soldering, but other procedures can go funny.
* Different pastes have different ball sizes. This will limit the pad size you can successfully use this technique on.

tician
05-02-2012, 07:35 PM
I'm sorry if I'm reading this completely wrong, but it sounds like you were trying to put the paste on while the board was on the stove? If so, I generally apply the paste and parts before subjecting the whole thing to any heat.

Most of the pads were covered with at least some solder paste, but there was a lot where it should not have been (where components should have been) and cleaning between the pads would pull up most if not all of the paste (it would rather stick to the metal picks than the pads, which is where a decent stencil was really needed). So, I applied it to as many pads as I could, stuck it on a piece of 1/16" thick aluminum plate with one side coated in kapton tape and put that on one of the heating elements (electric stove with built in fan). I waited for surface tension to clean most of it up and started placing some parts before adding a bit more solder paste for the few remaining bare pads (which melted a few inches off the board and would not behave thereafter).


If you do, you could put some solder in the vias first to fill them in and eliminate voids or balling, then reflow the LEDs on top of that. Just make sure that whatever the board's sitting on won't stick to the solder (don't ask...).
Sounds like someone soldered a board to their heating/work surface. ;)
I had a feeling I would have to fill the vias before adding the LED, thank you for the confirmation. The layer of kapton tape on aluminum plate has worked pretty well in preventing accidental solder joints during reflow. Some of the excess solder that I managed to pull off the board with a pick simply balled on it. No damage after sitting there for a while, and just a little char/residue from what I would guess was the flux or binder of the paste (came off with a bit of soap and water).


This works fairly well for IC packages or other areas where you've got lots of pads sitting in a line, with a few observations:
* The thickness of the copper layer means solder can gather in the valleys between the pads.
Which is where soldermask/solder-resist can be quite useful if it is available as an option for the board design and during board manufacture. For some reason the board house messed a little bit with the soldermask of my boards this time around (did not tent the screw holes on the top side or any of the thermal portions of the transistor and LED pads, but they did tent the thermal portion of the vreg, the bottom of the screw holes and avr-isp header).

* You need good paste. The stuff we use has a shelf life of 6 months, after around 8 months it's still useable for hand soldering, but other procedures can go funny.
* Different pastes have different ball sizes. This will limit the pad size you can successfully use this technique on.
I fail pretty badly on that count. Lead-free (http://search.digikey.com/scripts/DkSearch/dksus.dll?x=0&y=0&lang=en&site=us&KeyWords=SMD291SNL10-ND) manufactured in 12/2010 and stored in its syringe in a ziplock bag in a cardboard box on the floor of my bedroom since 5/2011. Applied pretty well with some portions of the hand cut stencil, but the stencil sucked something awful.

Th232
05-02-2012, 07:52 PM
Which is where soldermask/solder-resist can be quite useful if it is available as an option for the board design and during board manufacture. For some reason the board house messed a little bit with the soldermask of my boards this time around (did not tent the screw holes on the top side or any of the thermal portions of the transistor and LED pads, but they did tent the thermal portion of the vreg, the bottom of the screw holes and avr-isp header).

Yeah, I've found soldermasks to be an iffy business. A small warning though, I've seen places where the pitch of some IC leads means there simply wasn't enough space to deposit a mask (certain small and/or leadless IC packages come to mind). Here's part of a board I've had to hand solder before, an FTDI232RL chip (or half of one) sits on those pads.

3979

But I digress, that's not a problem for you though.


I fail pretty badly on that count. Lead-free (http://search.digikey.com/scripts/DkSearch/dksus.dll?x=0&y=0&lang=en&site=us&KeyWords=SMD291SNL10-ND) manufactured in 12/2010 and stored in its syringe in a ziplock bag in a cardboard box on the floor of my bedroom since 5/2011. Applied pretty well with some portions of the hand cut stencil, but the stencil sucked something awful.

I can't speak for that particular variety of solder paste, but for ours we've found that storing it in the fridge really helps with prolonging its life.

tician
05-02-2012, 09:45 PM
Yeah, I've found soldermasks to be an iffy business. A small warning though, I've seen places where the pitch of some IC leads means there simply wasn't enough space to deposit a mask (certain small and/or leadless IC packages come to mind). Here's part of a board I've had to hand solder before, an FTDI232RL chip (or half of one) sits on those pads.

Indeed. There are a number of MEMS rotational rate sensors, accelerometers, and compass ICs available only in leadless packages that permit no soldermask anywhere within a few mils of the exterior of the IC or a proper solder joint will not be created during reflow. If I had been even dumber than usual and attempted to use the MLF-32 package instead of the TQFP-32 for the ATmega88/168/328 on the board, I would have had to be very wary since the default soldermask exclusion area for each pad exceeded the distance between the pads.



I can't speak for that particular variety of solder paste, but for ours we've found that storing it in the fridge really helps with prolonging its life.

Honestly, I have been using the solder paste so infrequently (and only for personal prototypes) that it does not really matter much to me how long it is 'good,' so long as it is 'good enough.'

Th232
05-02-2012, 10:12 PM
I just realised I forgot to ask, what exactly are you making? Since you're using 0402s and a TQFP package I presume it's pretty small, so now I'm curious.

tician
05-03-2012, 01:23 PM
The horseshoe shaped board is a possible replacement for the HUV robotics Foot Pressure Sensor board for the Robotis Bioloid foot plate. I used 0402 just because I kinda like the challenge, since it could just as well have been 0603 or slightly larger. The holes on the arms and sides are for reverse mount LED's (eye candy) that I decided against buying because they were a bit expensive and I was not sure the board would even work. If I get it working at all, the next version will have only four with each connected to an 8-bit PWM pin so that each LED will reflect the current loading of the sensor in each corner of the foot. I also need to move the arms about 1~2 mm towards the center to better align with the polyester leads and solder tabs/pins of the 0.1" round FSR's. I should also move the top pads for the FSR's about 2~4 mm towards the front of the board, or go with a through hole connector on the bottom for all the FSR's. If I stick with soldering them flush to surface pads, it might be a good idea to make it three boards that mate together to improve ease of assembling the foot. As is, the length of the board makes it a bit difficult to attach the front FSR's to the bottom of the board (but not impossible).

The top board was added just because I was already paying for the largest rectangle of the FPS board, so it seemed like a waste to not put something in that big area in the middle of the arms. It is a 6 LED board that will hopefully work better than the predecessor without an avr-isp header (yes, I did place it under the ATmega, but it is meant to be accessed from the other side). If it does work, it may also be useful in providing consistent illumination for the HaViMo2's in the lab so that they do not need to be re-calibrated any time the lighting changes (the 4 M2 screw holes are on the 16mm square common to Bioloid frames).

The face (Toy Soldier smile) and quote (Gaius Valerius Catullus - Catullus 101) are because I only learned about Doctor Steel's 'retirement' a little while ago (several months after the fact).

Th232
05-03-2012, 07:10 PM
Those look very nice, I'll be interested to see how the pressure sensor board works in particular. Neat idea to provide visual feedback with the LEDs as well, if you don't mind I might use the concept in my own robot.

I thought "Ave atque Vale" was just a general goodbye, guess I've learnt something new today. Catullus was a seedy guy...

Gertlex
05-03-2012, 07:29 PM
I too find the LED brightness indicator to be an awesome idea :D

tician
05-03-2012, 08:08 PM
"ave atque vale" - "hail and fairwell" to his dead brother. Was thinking of "diruit aedificat mutat quadrata rotundus" or "nostalgia is necrophilia" (two of my all-time favorite quotes - I've got an ever growing text file filled with them), but neither really seemed to fit the board's purpose.

Pretty sure I destroyed another set of boards today with another attempt at a hand-cut stencil. While more successful than the last attempt (except I may have killed a couple LEDs and vregs instead of just cheap passives), if I cannot get every single pad individually (read: flawlessly) solder pasted and every component placed beforehand then I should not get anywhere near a heat source. Also confirmed that while the kapton does not decompose at the temperature I have been using, the adhesive definitely does and it is not produce a pleasant aroma.

Th232
05-03-2012, 08:38 PM
This just pegged, how are you hand cutting a stencil for 0402s? Sounds painful to me. Maybe a good microscope and a finer needle would work better for them?

tician
05-03-2012, 09:18 PM
Tried 9mil photopaper yesterday thinking it would be easily reused (way too thick and tears a lot), and then the thinnest paper I had on hand today (better but still tore a lot). Printed the cream layer and tried to cut carefully with a utility knife (an Xacto knife would be better). I am pretty badly near-sighted so I can go a while staring at small objects close-up before I go cross-eyed. The problem is that I am using a blade and not a fricken' laser to cut the stencil, so if the pads are too close and I am not careful enough then it tears. If I still had some printable transparency sheets, I would have used them from the start (more likely to be reusable and less likely to tear or contaminate the paste with fibers).

I did better today with a thinner layer of paste and better cleaning of the excess between pads where the paper tore, but still failed pretty badly. When trying to fix the single bad resistor and clean some of the excess solder off the TQFP pads, something started stinking and I noticed the LED's were pretty badly discolored. Pulled the poorly placed ATmega and the metaphorical plug (then spent a few hours priming the dining room before the fumes finally made me want to spew).

Th232
05-04-2012, 12:12 AM
Ouch... If it wasn't for the fact that I'm in Australia I'd offer to make you a stencil (laser cut 50 um thick steel foil) and send a few fine gauge syringe needles as well. The latter might raise some interesting questions at customs though.

Given the size of 0402 pads, maybe a better way would be to just punch a round hole in the stencil (~0.5-0.7 mm dia)? Since the solder will spread out anyway when you heat it all up the shape doesn't have to be perfect.

tician
05-04-2012, 01:15 AM
Ouch... If it wasn't for the fact that I'm in Australia I'd offer to make you a stencil (laser cut 50 um thick steel foil) and send a few fine gauge syringe needles as well. The latter might raise some interesting questions at customs though.
Considering I've only got one set of the soon to be revised boards left, it would be a major waste anyway. My syringe of solder paste did come with a large bore needle, but the viscosity of the paste is a bit high to force out at room temp with much control. I haven't attempted to use it with the needle since 04/2011, and even then I had a lot of trouble getting it to stick well to the pads (0603, SOT-23-3, SOT-89, and TQFP-32) without using the picks and a lot of patience.


Given the size of 0402 pads, maybe a better way would be to just punch a round hole in the stencil (~0.5-0.7 mm dia)? Since the solder will spread out anyway when you heat it all up the shape doesn't have to be perfect.
Wish I had thought of that. I guess it is worth a try tomorrow. Definitely ordering a laser cut stencil with the next version of the board to avoid all this mess. Then all I would have to worry about is my manual dexterity (lack thereof).

Th232
05-04-2012, 01:35 AM
Considering I've only got one set of the soon to be revised boards left, it would be a major waste anyway. My syringe of solder paste did come with a large bore needle, but the viscosity of the paste is a bit high to force out at room temp with much control. I haven't attempted to use it with the needle since 04/2011, and even then I had a lot of trouble getting it to stick well to the pads (0603, SOT-23-3, SOT-89, and TQFP-32) without using the picks and a lot of patience.

I've found a similar situation with the viscosity of the paste. I mainly used the needle like a spatula to scoop the paste and put it on the pads. Can't say I've tried dental picks but I suspect it'd be similar to when I tried sewing needles and so on, the paste just slipped off and there just wasn't enough control.

tician
05-15-2012, 07:43 PM
So, I've made a little bit of progress over the last week in overcoming ~25 years of 'meh'. Besides finishing a few renovation projects and buying enough 1/4" MDF to make the frame for the router/printer (still need to cut it), I built a small frame out of 1/16" aluminum angle to mount 15 of those little solar panels. They are all stuck in place with silicone and sealed around their edges to make what I hope to be a water-resistant protective top-cover/weather-shield and power supply for a rather small outdoor rover. However, it was while I was putting the panels into the silicone that a certain something finally sank in (might have been the fumes): putting 15 solar panels at ~3.6VDC each in series gives ~54VDC (going to use a switching reg to bring it down to 5V and maybe a 12V line as well).

So... crazy stupid/awesome idea: 300 panels at ~3.6VDC for what could potentially be a ~1080VDC (~70mA) power supply. Of course, I'm not sure for what I would actually use such a creature if it even worked. Tesla coils are most impressive in the dark, so that idea's out. Maybe I could finally get around to making that coil gun I've been wanting... but then I'd need more capacitors, some really beefy transistors/solenoid-relays, and quite a bit of plate aluminum (got a kinda strange design in mind).

Warnings? Suggestions?

Th232
05-15-2012, 08:15 PM
Hmm... I figure that since the main issue with solar power is how much sunlight you're getting, any project should be able to work over a large voltage range. A solar powered coil gun would be interesting to see, but the question of recharge times might make it more practical to just power it from some batteries or mains with an appropriate boost circuit.

One thing I've been contemplating around my place is some solar cells wired up to some LEDs that are indoors. Efficiency to convert sunlight to electricity then back to light will be terrible, but for areas in the house that just don't get any sun it could be surprisingly useful. Variations could include using the solar cells to recharge a battery instead, a switch to change between solar/battery/mains as so required. Not the most impressive application, but a more practical one. Would also be a use for those XM-Ls you have.

tician
05-15-2012, 10:10 PM
If I were to buy some 0.5[V] 2+[A] cells (or maybe try solar thermal with a couple TEC's), that would result in a panel producing a useful amount of power at a usable voltage and current for household stuff.

Three-hundred 3.6[V] 1/4[W] panels in ~1[m^2] should yield about 75[W] at most (even here in it's-not-the-heat-it's-the-humidity, scorched-earth, redneck, paddle-faster-than-the-banjos, super-duper-bible-land - ಠ_ಠ (http://youtu.be/GYL28a0LM_A)). Not a terribly useful power output or voltage, except for one or two panels in series for low power wireless sensors or low power rechargeable path lights (pretty sure that was the manufacturer's intended use), or lots of panels in series for novelty projects. Even the 3x5 panel that I built is not well suited to its intended purpose of powering a rover due to the low power and high voltage (though the epoxy coating allows me to use the panels as semi-structural components instead of needing to build a polycarbonate case to protect them).

Th232
05-15-2012, 10:35 PM
Maybe I'm a bit out of touch regarding solar cells, but I was under the impression that they could be put in parallel to increase the current? Given that you're pretty much only talking about putting them in serial I'm wondering if I'm missing something.

tician
05-15-2012, 10:56 PM
Indeed, but they are also a power sink when not in sunlight. A rectifier diode is typically added in series to prevent this reverse current flow/drain, but at the voltage-current combination of these little panels I would have to buy quite a few diodes and take a pretty big hit in lost power to keep the voltage reasonable. A 16S array for 57.6[V]/~70[mA] switched down at ~93% efficiency to 12[V]/~310[mA] (~3.7[W]) versus a 4S4P array for (14.4-Vdiode)[V]/~280[mA], where Vdiode can typically range from 0.3~1.5[V] depending on the diode (3.9~3.6[W]). Of course, looking at the math now, it does not seem to be that bad of a loss at even 12[V] or higher if I use the right diodes.