While trying to measure the current draw for a battery-operated thermometer, I found that I didn't get any conclusive results with my multimeter and began to look for one that allowed me to measure microamperes, finding that there were some decent ones for less than a hundred USD. Eventually I came upon the µCurrent™ GOLD and found that to be the best option, since I wouldn't end up with another multimeter laying around somewhere while its batteries slowly die.
When I looked at it at the time, I found that it was out of stock, and since the bill of materials is listed, I decided to make my own to remedy the problem. I first decided to replace the CR2032 holder with one that allows the battery to be inserted perpendicularly instead of parallel because I'm not horribly fond of having a large pad for the battery to contact to transmit power (I blame a key-chain sized garage door opener for that).
I also found the switches to be quite large and pricey (based on Mouser's prices), so I looked for something that was smaller and cheaper, and I found something that fit the description.
I tried to use the same reverse-mounted LED, but found that it needed a rectangular hole, which I wanted to avoid because OSH Park might not fabricate it right. I did find something that required a circular hole, and after a quick question to OSH Park, I found that the pads would be fine with the hole there (10 thou would be removed from the pads adjacent to the hole, which I saw on the board after getting it).
I did find a problem with the schematic, which is that pins 1 and 3 of the voltage monitor IC (TPS3809L30DBVR) are flipped, as pin 1 is supposed to be connected to V- and pin 3 is supposed to be connected to SW2-2A.
I also downloaded the Altium PCB files and had a heck of a time getting the right software to be able to open it to poke around with. There's a few things I didn't understand (the traces that go nowhere and the size of the traces for the current input), but it at least helped with ideas for part placement and certain connections (that wasn't clear to me in the schematic).
Once all the parts were created in my library (which I prefer doing, since I don't trust some of the ones from the manufacturers... I blame the pre-made SOIC-8 footprint from back when I was doing the Logitech Optical Trackman power hack...), I started with the placement of the lesser of the parts and ran the traces to them when I had them in a good spot.
After getting all the air wires in, i did some rough placement for the parts and had a tough time, but eventually got the parts arranged and most of the traces routed. I had to add four jumpers to make it possible, since I was trying to use the same trace width for everything after the current input traces.
During all that, I had wondered a couple times if the trace width made a difference, and while i was fairly close to finishing the board, I went and looked it up. Much to my surmise, there was, and I was a bit disheartened that I wasted all the time I did with part placement and trace routing, since I had to throw almost the entirety of the design out. What I found was that it's better to use thin widths for signal traces and thick widths for power traces, and so for the new design, I decided to use the thinnest possible width (6 thou) for all signal traces and anything higher for power traces.
It turned out to give me a lot more leeway with part placement, though routing V+ and V- still wasn't easy, but I did find a way to route it all without having to add jumpers.
Before I forget, I wanted to keep it fairly clean-looking so no exposed vias on the top of the board. I think the µCurrent™ GOLD is actually a multi-layer board with buried vias since there are two traces that via to another layer to hop over other traces. I can't do blind vias with OSH Park, so it's why I wanted to avoid them, or hide them under the switches.
Anyway, I had to utilise SolidWorks a lot to do the math for me to make all the traces as tight and neat-looking as possible (I was also doing this before the redesign). I could've done it all by hand, but it's a bit more time consuming than just drawing it and making all the necessary relations. I also did a lot of manual math with part placement, but it's fairly simple, since I just spaced things 6 thou apart to make it all compact.
Oh, the big white circles are just a reference to keep parts away from the boss of the box I used, and the traces that cross into it aren't a problem, since it's nothing compared to SMT parts.
Once I was done with the part placement and traces (or at least I thought, look at the pad next to 1B of the "upper" switch and you'll see a couple of air wires there), I began cleaning up the non-essential items like the other two mounting holes, the top polygon fill, and the silkscreen. I was pretty much done (again, I only thought I was), and was fairly pleased with the result.
If I remember right, the redesigned board is slightly smaller than beforehand.
Anyway, I noticed the airwire and fixed it (again, using SolidWorks to math the trace nicely), and checked the size of the boss of the adaption plate I was making.
I then did some fine-tuning of the silkscreen before I was happy with the result.
Before and after ratsnest tool.
I then uploaded the board to OSH Park and checked it before ordering it with a couple other boards.
Ordering the parts was an interesting ordeal, as I had found the 0805 case, 10 ohm, 0.05% tolerance SMT resistor on Digi-Key (Mouser didn't have it) when I was getting footprints, but never paid attention that they didn't carry them in stock. I found that Element14/Newark had them, but didn't feel like messing with them (I blame bad memories with the original Raspberry Pi model B), and decided to try out Arrow Electronics. I was actually surprised because they offered free shipping, and so I ordered three of the resistors from them so that I could fill that spot on the other two boards.
For the output amplifier (LMV321AS5X), Mouser had happened to run out of stock of them while I was ordering, and luckily Digi-Key had them in stock, and I placed an order for three and paid a bit less than three USD for shipping, which was nice.
I had to change the CR2032 holder to a different package (from tray to tape and reel) so that it would ship with the other parts from Mouser, but other than that, there were no other surprises.
I also ordered the adaptor plate from Xometry (as well as a couple other items) and had it made out of ABS-ESD7, which wasn't more expensive than the "first available" or cheaper ABS options.
I obviously received the boards back first, and then set them at my solder station after taking some pictures. The "main" cluster of parts was going to be hell to solder when I looked at the board, but I figured if I solder the ICs and then solder the other parts around it, it should be fairly easy.
The usual set of pictures.
I could've started soldering all the parts, but I wanted to get it all done in one go, so I waited until I got my Xometry order. It was estimated that it would ship 22 May, but I ended up getting it this past Thursday (18th), and with my friend needing time to himself to catch up on a project of his, I decided to get this thing soldered and assembled today (20th).
Oh right, I offloaded the banana jacks from the board to make the board smaller.
Anyway I first soldered the 18 gauge wire to the banana jacks and covered the joint with heatshrink before setting them aside to work on the board itself.
I first started with the speciality 10 ohm resistors and the output amplifiers on the extra boards so that I can shove them into ESD bags and get them out of the way.
R9 and U2 populated on the extra boards.
I then started with my plan to start with U1, and found that I couldn't find the silkscreen dot for pin 1, and after looking at the board in EAGLE, I found it was the one "underneath" the "1" of "U1". I dropped some solder down for pin 1 to make it easy, and also dropped some solder down on pin 1 for U4, though the dot was fairly apparent.
I then soldered R5, then U4, then C3 and C4 (I think). Pretty much I kinda worked my way outward from the inside, though partway, I decided to get the "stray" and "leftover" parts out of the way (R1, R9, R2, and then U3, D1). I held off with R4 because I got confused with the blinker/metronome project.
The LED was somewhat interesting to solder, since I didn't want to use my standard method of putting some solder on the pad and tacking the part to the board before fine-tuning the position and such, regardless I got it to work. I probably could've just used that method just fine though.
The battery holder was interesting since the pads are the same size as the tabs, so I put a bit of solder down, put the tab on top, and then heated the tab up until the solder flowed and I got it into position. I forgot about the other tab and removed the holder before putting a slight bit of solder down on the other pad and re-soldered the tab. I then heated the other tab and applied another slight bit of solder. I gave the holder a slight nudge with my finger, since I wasn't sure if it was soldered, and since it didn't budge, I set the board down to figure out my problem with R4.
After digging through the list of parts of the invoice, I realised that the LED I used isn't any difference in forward voltage or forward current, so I went back and soldered the final 270 ohm resistor to the board.
Components soldered.
Since I didn't get a good opportunity to talk about the adaptor, I'll do it before I move on. While I had designed the board to use the lid holes of the box I used, I wanted it to look nice (and didn't feel like trying to nicely drill holes for the banana jacks), I decided to make an adaptor plate to make it easy and nice. I first used the data sheet dimensions, but eventually loaded the 3D model into SolidWorks and pulled some measurements out.
I added some 0.15mm bosses to one side where the PCB sits so that the top of the PCB would theoretically be flush with the edge of the box. I also added some countersinks so that I could use the flathead screws that come with the box instead of having to find something to shove in there. For the board screw holes, I made them 2.5mm in diameter so that I could use a M3 - 0.5mm tap to thread the holes so I didn't have to use some nuts.
I added a cutout for the wires on one side (voltage out), since they were at the edge of the board, and would be inaccessible otherwise. I also did the same to the opposing side so that it wouldn't matter if the plate was rotated 180 degrees, and to shave off some costs.
I spaced the holes for the banana jacks equally, so that it would be centred between the outer edge and the inner edge, and divided equally between the width of the plate.
I also added some arches below the long strip, but removed them when Xometry added about five USD to the price.
After getting the part, I had to drill the board holes out to 2.5 since they seemed closer to 2mm (or smaller) and then tapped the holes. I then checked the fit of the banana jack holes and had to file the hole out a bit so that it would be easy to insert the banana jacks, but still give a snug enough fit.
Top view, bottom view, after tapping, and after filing. I used the torch for the last two pictures to try to make the features a bit more clear.
While it would've looked nicer without the hatching, there's no real easy way to avoid that with FDM-style 3D printing. Luckily, it's uniform, so it doesn't look horrible.
Anyway, back to the assembly. I attached the banana jacks to the plate before I found that the 18 gauge wire didn't fit into the hole, which was annoying since I had already shrunk the heatshrink. While I could've sworn I engineered it for 18 gauge, the best option was to use smaller gauge wire (multimeter lead wires are usually 18 gauge, which is why I chose 18 gauge). Luckily I had some 22 gauge wire laying around, so it wasn't as bad as it could've been.
After getting the banana jacks situated with 22 gauge wire and heatshrink, I then reattached the banana jacks to the plate and then awkwardly solder the first two wires to the board before finding a better way with the other two (the wires close to the switches would've still been awkward anyway). And once that was done and trimmed, I screwed the board to the plate and bent the wires until the assembly fit in the box without any force.
Top and bottom.
I was also concerned with the clearance between the battery and the plate, but I had no worries when I was inserting the batteries, seeing that I had taken it into consideration or something.
I actually used an "old" battery out of a garage door opener that seems to not like any batteries under 3.1 volts, and since the voltage monitor IC turns the LED off at 2.64 volts, I might as well utilise the battery.
Picture wasn't really necessary, but too late.
Anyway, it was time to give it a power test, and when i flipped the switch, my heart skipped a beat or two in the couple seconds before the LED came on.
Either the picture with the switch at "On" never came out, or I actually did move it to the "Short" position by accident.
Now that I confirmed power, it was time to do the final assembly, and it was a bit awkward since the screws and the lid are both black because it was hard to tell how far I drove the screw in as I was trying to keep it relatively loose to allow the holes to align.
Different angles to view the silkscreen better.
I had also ordered some resistors later (with the parts to hack my headphones to make them balanced) to test it according to the article, but didn't realise one of the values was actually an SMT resistor (and the style I'm not fond of because of the SMT zener diode I use at work).
I had made a board and sent it out to get made after checking the unconnected layer (where the air wires reside). I did this so that it would be a lot easier than snagging some clipped legs from work and awkwardly soldering the SMT resistors together as well as the legs.
Doesn't entirely look great, but I prefer functionality over aesthetics.
The boards shipped Thursday (18th, and I'm thinking they'll be here Monday (22nd) or Tuesday (23rd), and once they're here, I'll get the board together so I can check the Mini µCurent.
Soldering definitely wasn't easy, as I had a few encounters of possible solder bridges, but only had a couple of them which weren't hard to get rid of. Would I hand-solder 0603-sized parts again? Probably only if I had no other options. Would I put pads 6 thou away from each other again? Probably as long as the parts aren't 0603.
This is definitely the tightest and probably the nicest looking board I've designed, and it was definitely worth all the time I spent in SolidWorks to make it so.
I'll stop ranting here for now until I'm able to test it, which I will write more (though you all will be reading this after the entirety of this post is written).
It turns out the boards were sitting in the post box since about 11 yesterday, but I didn't know until I went to check where the package was before heading to bed last night. Unfortunately it's four right now, so I'll wait a bit before fetching the post. At the least I won't have to keep the "I hope it works just fine" thought in my mind for that much longer (though I should say that's it's more of a "I hope I didn't f**k it up").
Anyway, one thing I forgot to mention is that the "B1" silkscreen ended up getting covered by the battery holder, since I forgot to flip the orientation in my head since it's on the bottom and not the top (like in the datasheet). I moved the silkscreen in EAGLE, but it's a shame I'll not be seeing it that way (well, not like I'll be seeing the bottom of the board much anyway). It was actually confusing me for a moment when I was soldering the holder on, since I thought I flipped the orientation, but again, was just that my spacial module crashed with the item while doing the silkscreen.
While arranging the parts, I was a bit hesitant with having the pads for the switch so close to the battery even though the battery is raised up some by the holder. When I was soldering the switch, I found that the legs didn't protrude all that much, so now that I think about it, I doubt it's a problem... Actually I'm gonna go take a quick look before it drives me mad...
Yep, a tonne of space there.
Anyway, I could've also designed a panel that allowed the board to be mounted underneath to lower the outside profile, but that would require labels or something. Perhaps I could've made something with the Front Panel Designer software from Front Panel Express, but I would've had to do a bit more work than I would like to (I would have to figure out all the text positions and how to put it into the programme). I'll digress slightly and say I could've spent a lot less money on the solder station power panel if I had known about Front Panel Express, and it would've came out much nicer as they do print and silkscreen. C'est la vie.
I was worried about the strength of the adapter with where the board is and the thickness (why I put the arches in temporarily), but it's actually much stronger than I thought once assembled. The banana jacks feels solid when pressing them down, but considering the location of them, it's fairly expected. It does give slightly when I press down on the board, but it's not all that much. I actually forgot about those worries when I played with the switches after soldering them to the board, finding that they're quite easy to move (I was expecting something like the voltage switch on a PSU for whatever weird reason).
I had also thought of how to make the board even smaller, and the only way would be to offload the battery (and maybe use something like a CR213A, which would probably last friggin' forever if a CR2032 lasts approximately 50 hours of use). Offloading the battery would also ease the traces, as the placement for V+ and V- would be much more flexible, and probably reduce the top-side trace jumps.
The remaining way to make the board smaller is to force the use of an adaption panel and move the mounting holes after redesigning the board after offloading the battery.
I kept the CR2032 because I didn't want to have a battery and battery holder floating around in the box, otherwise, I probably would have offloaded the battery.
On the name, I went with the USB name basis where you have the large and somewhat square type B plug as the norm, which is followed by mini and micro. So since this is the reduced size, it's "Mini µCurrent". I would say offloading the battery would probably be called "Micro µCurrent" ("µ µCurrent"? XD), and then making the board as small as possible would probably be called "Nano µCurrent". Since I won't be designing either board, it won't be my call (cheers if you do design the smaller boards and utilise the names).
On the "Econo Purple" part of the name, since the main distinguishing feature between the original µCurrent™ and the µCurrent™ GOLD is the gold-plated banana jacks, I decided to go along with the schema there and because OSH Park's boards are purple, that's where the "Purple" comes from. "Econo" is (obviously) short for "economic" which just comes from the fact that this is cheaper than the µCurrent™ GOLD itself (at the least for those in the States).
While checking the name of the original µCurrent™ (I was double checking to see if the "™" was there), I over-scrolled (thanks to the hyper-scroll function...) and ended up seeing an explanation of the strange traces. Rather, a question about those strange traces and a reply to that with a video. If you're just reading through here, I won't torture you by forcing you to watch a 30-minute video of David L. Jones jabbering away to get an explanation of those funky traces I saw (or watch it if you'd like an in-depth explanation). Basically they're traces for testing and connect to other traces in the Gerber file of the panellised boards. Luckily, they were nothing I really needed to worry about.
While looking at the Wikipedia article on him, I found out why he uses Altium - it's because he used to work for them (can't complain with free or reduced-price software).
Anyway, it's getting close to six now, so I'll maybe fetch the post in a couple hours (it's 3°C outside atm, supposedly), but I'll quit typing for now.
I fetched the post approximately one hour ago and took the beauty shot of the Mini µCurrent Econo Purple before opening the package. While dumping the boards (and sticker) into my hand, one fell on the floor, which I caught a quick glimpse of, and it didn't seem quite right. After picking it up, I saw that the board was riddled with vias, which I knew wasn't right, and once I looked at the boards a bit closer, I realised OSH Park had sent me someone else's boards.
I was (and still am a bit) quite disheartened, not because I can't get this project out of the way (if I get impatient enough, I'll solder the SMT resistors together and use 22 gauge wires as makeshift legs), but because that someone else has to wait that much longer before they can finish theirs (I'm just hoping their project isn't time-sensitive).
I went and got all the prices together from what I spent, which turns out to be a total of 59.05 USD without the adaptor plate, and 70.77 USD with the adaptor plate. Buying a µCurrent™ GOLD, would've cost me about 63.80 USD (according to the currency convertor thingy at the EEVblog store, 66.36 USD according to Google) for the item itself, and I'm guessing 30 AUD for shipping, so 109 AUD total which calculates out to be roughly 78.14 USD (81.27 based on converting 109 AUD to USD with Google). Since it's out of stock, I couldn't get an accurate total to convert.
Total shipping costs that I spent was 7.84 USD (there's also a total of 19 cents tax, but let's just ignore that since it's not worth mentioning), so subtracting shipping, I spent 51.21 USD (62.93 USD with the adaptor plate).
While I was working on the board at work (during a short break), the electrical engineer was asking about it before saying that it would cheaper for me to just buy it. I knew I could make it for equal or lesser cost, and I did, even with the adaptor plate. I'm debating whether or not to rub it in his face that I proved him wrong. I'll think about that more.
All for now until I'm able to test it (not that anyone's reading this until it's posted anyway).
I got impatient and tested it out. At first I got -600 something mV reading (which would be -600 mA), but found that when I hooked the 47 ohm 1 watt resistor and power supply up, the reading didn't change, even when I swapped the resistor for 47 kilohms. My heart sank a lot and I went over the board layout, schematic and screenshot of the µCurrent™ GOLD.
I stopped poking around a few minutes after noon to go eat and pondered it a bit more, figuring I should check for shorts and whatnot. After a bit of time, I found a short between R12 and R11, and once I removed the solder bridge, I was getting proper readings. I got 111mA for 47 ohms, 111µA for 47 kilohms, and was happy that things were working right. It was time to solder those SMT resistors together, and I happened to have some legs lying at the solder station.
Very resistor, many makeshift, wow.
When I turned the Mini µCurrent back on in the nanoampere scale, I found that it was bouncing around roughly 30-40nA and after I connected the makeshift resistor, it went up to 60 something nA - for reference I should be getting around 100nA. Needless to say that I was fairly ambivalent, since I was happy that it works otherwise, but upset that it doesn't properly test out in the nanoampere scale. While I highly doubt I will ever need the nanoampere scale, it's still sort of a bummer to have something work partway.
Perhaps the solder bridge between R11 and R12 damaged something enough to where it disrupted the nanoampere resolution, but not enough to throw off the other two scales (which should give 106mA and 106µA readings). After taking another close look at it (physically) and making sure there was nothing wrong with the makeshift resistor (I guess my Fluke metre can't measure 50 megaohms), I decided to just leave it be, as I was more wanting something that reads microamperes.
I was going to release the EAGLE file, bill of materials, and 3D models of the adaptor plate, as well as sharing the board on OSH Park, but with how it turned out and the fact that I have no idea what's wrong with the nanoampere scale, I think it would be safest to not do so.
I also was going to give away the other two boards with R9 and U2 attached, but again, I'm going to keep them for now until I find out what's wrong.
Kind of a shame it turned out the way it did, but at the least it's not completely unusable like I thought it was before lunch.
This project was definitely a challenge in many ways, but it was mostly worth the trouble.
I have learned to turn off the top, bottom, via, and pad layers to make the air wires more clear, so I don't miss any connections, and have applied that to the µCurrent tester board and the metronome/blinker board.
I think the other take-away from this is just devising a soldering order when dealing with close or small parts.
That's it for this post and the next one will be the metronome/blinker board when I get the board tomorrow.
P.S.
Oh right, while watching that video, I saw the two traces on the µCurrent™ GOLD that jump topside to hop over other traces, and that it's hidden under one of the switches, so it actually is a 2-layer board.
No comments:
Post a Comment