Update V, 30 April 2009: No lockups so far. Still haven’t replaced the caps in the second unit. My sister just ordered one and if she starts running into problems, I’ll replace the caps in the second one, swap hers out, then fix hers. With two to be “fixed” I should remember to get snapshots.
Update IV, 01 April 2009: Updated Mouser parts link. Haven’t yet “fixed” the second router as I’ve decided to wait until I determine that this is in fact a fix.
Update III, 28 March 2009: Hmmm… Woke up to a non-working internet. Unfortunately, I had gotten into the habit of cycling the power on everything and had already removed power from the cable modem and router before realizing I needed to troubleshoot everything *except* the newly “fixed” wireless router. Dang it! The short story is that I had the opportunity to find out if the WRT54GL was actually fixed. <Mental note to self: don’t do that!>
Update II, 23 March 2009: The parts came in late this afternoon. It was a bit of a pain to get the old caps out. I kept upping the heat on my Aoyue 908 trying to get the danged solder to melt. I ended up heating the leads to the point that I pulled the cap heads right off. It wasn’t that big a deal, though. The leads came out pretty easily after that. I couldn’t solder wick the remaining solder away, so I used a pin vice with a drill bit very slightly larger than the leads on the new caps and bored the solder out. Worked great. The new caps went in in just a few minutes. All the caps were 20mm tall, so I did end up having to dremel away a small portion of the case to get it back together. All in all it was a piece of cake and I’m running off of the “fixed” router right now. In my haste to get it done, I forgot to take snapshots. I have another one to do, so I’ll try to remember to get snaps then, probably sometime tomorrow morning.
Update, 19 March 2009: First, I don’t want to give you the wrong impression about Tomato firmware. It’s great. In fact, I now use it instead of dd-wrt. Nothing wrong with dd-wrt, either. I installed Tomato and now run Tomato. It’s a tasty Tomato. I changed the post title and link so as not mislead anyone.
Okay, I left a message on Savel’s blog asking which caps need to go where. He responded that the filter for the mains will be around 16v and the others will range from 5 to 1.8 volts – all I have to do is measure them. And so I did. The PCB is easily removed from the housing by first popping off the antenna base covers to reveal a knurled nut and simply unscrewing them. The blue front is attached on the sides by hook-and-groove friction locks. Remove the Voids Warranty sticker (I just cut along seam with the blade of my trusty Leatherman Squirt E4). Now hold the housing such that the top side is toward you and the blue front is facing either the left or right. Using one hand per leg, put your fingers on the insides of the legs and force them apart by pushing against your thumbs against each other. That’s a poor explaination of what to do, but I trust you’ll figure it out. The front will pop right off. Now place it upside down and slide the feet toward the front until you feel it stop, about 5.5mm (~ 0.2 inchs), then lift straight up. The PCB is detached by removing the two screws located near the center of the board just behind the RJ45 ports. Slide it towards the front and lift.
There are four 16v, 22uF capacitors along the edge of the PCB just behind the power port. The one closest to the power port measured 12v, and the others were 3.3v, 5v, and 1.8v moving from back to front. Savel replaced the 12v cap with one rated at 25v, 470uF and the other three with 6.3v, 1500uF caps. He said that while the higher the uF the better, the main concern is finding capacitors that aren’t too tall – you won’t be able to get the housing back on. I see that another concern is the diameter. If they’re too fat, there might be a problem. It looks like I can use caps up to 16mm in length and a maximum of 20mm if I modify the top housing: there’s a support ridge running the width of the top half that’ll need to be Dremel’d away just enough such that the caps can clear during reassembly.
I rummaged through bins of old computer video cards, motherboards, modems, etc., looking for comparable parts, but didn’t find anything I wanted to kill — I really will get around to putting that old Abit dual CPU mobo in a case and turning it into a server! I ended up ordering from Mouser. Using the filtering, I first went for size, ranging up to 9mm in diameter and 20mm in length, then Farad rating, then voltage. I have no clue as to whether these are ideal, but I ended up selecting part number 647-UHZ0J152MPM6 for the 6.3v caps and 647-UHE1E471MPD for the 25v caps. I have two GLs that need to be modified and so ordered 10 of each, more than required, but I kept it to less than $10.00.
I’ll update once the parts arrive. Probably a bad snapshot or two as well.
Tomato, like dd-wrt, is 3rd party firmware that turns your $50.00 Linksys WRT54GL into a $500.00 wireless router. I’d been having odd problems in that every 6 to 8 weeks the damned thing would freeze up and nothing would bring it back to life in just a few moments. Not a reset, not unplugging and waiting, nothing. I gotta have my ‘net fix and I gotta have it 24/7. Fortunately I have a duplicate, so all I’ve been doing is swapping them out. By the time the “good” one goes bad, the “bad” one is good again. Horribly frustrating. I’ve been using wireless technology since the B days, a Netgear setup which cost some $300.00, and this particular wireless box pretty much since the original G model came out – going from B to G was like going from dialup to broadband, I tell ya! Anyway, late last year I happened to be watching a TIVO’d episode of (I think) Systm on such 3rd party firmware, particularly Tomato, and decided to give it a go. Yeah, well, it didn’t help. I ended up installing Tomato on both and I had to swap them out just last Saturday around midnight (“Honey, is there something wrong with the wireless?!”).
I’ve built at least two dozen MintyBoosts – some with 4 x AA alkalines, some with rechargeable 2 x A123 NiCads, and of course the standard 2 x AA – and finally decided to build a rechargeable version. I’ve long had the idea, but wasn’t too keen after doing a bit of research on lithium polymer batteries and what is truly meant by “vents with flame” (see this and this, for just a couple of examples). I finally got over the fear and built two in a single morning. I used LadyAda’s MintyBoost circuit board, SparkFun’s LiPo charger, and a couple of the dozens of mint containers I have sitting about. The larger tin contains a 2200 mAh battery I got from RapidRepair for a measly 6 bucks (purchased 3, actually) while the smaller uses an iRiver Clix 720 mAh battery I got from BatteryShip (and cost more than 3x more than the 2200, though it did come with a nice set of minature tools).
My SparkFun charger is an older version of the one depicted at the link above, the only difference being that mine didn’t have the JST (?) connectors. I had originally purchased and planned to use SparkFun’s 2000 mAh battery and decided on a layout which required a bit of modification to both the battery and the charger such that everything could be attached to the bottom of the tin – no mounting of components on top of the battery. When the 2200 mAh iPod batteries arrived, I was pleased to discover that they were perfectly suited the task, but because it was larger I’d have to place the recharging and boost circuitry on top of the battery. Not a problem, really, as for a number of years I made a living installing Nortel and Alcatel SONET equipment and have a roll or two of fish paper electrical insulation left over. I cut the fish paper to size, used spray-on adhesive to glue it to the battery, hot glued the battery into the tin, then aligned and hot glued the circuit boards to the fish paper.
The mini Penguin was a bit different. I had to desolder the DC barrel jack from the board and cut as much off of the board behind the connector solder pads as was possible. I didn’t modify the boost circuit board, but when I do another one I’ll do a bit of reconfiguration to avoid having to cut a hole in the lid. I think next time I’ll eliminate the 8-pin socket and solder the uC directly to the board to get it out of the way – it *barely* closes as it is. I also need to find a shorter inductor. I’ve pulled parts from tons of old electronics boards, but have yet to find a suitable part, probably an SMT (surface mount) part. And while I’m at it, replace everything I can with SMT. Heck, it might even get to the point where I actually make my own boards designed specifically to use SMTs.
In testing, the larger unit recharged my Android Dev 1 with no problem. Hardly a dent in the 2200 mAh battery. However, the Lil Penguin didn’t fair so well. It took my Dev 1 to only about 65%. That isn’t bad, though, and it’ll do in a pinch. The key here is that the device is only for emergency use and it’s tiny, so I’m driven to improve the design.
UPDATE: I now have a couple of lithium polymer powered rechargeable MintyBoosts.
Lady Ada of Adafruit Industries sold me on yet another kit, the MintyBoost. It’s a USB charging device that fits into an Altoids chewing gum tin.
I ordered one entire kit, plus five extra PCBs and MAX756-CPA+ power regulators from Adafruit. From other vendors such as Maxim IC, Mouser, DigiKey, and Futurlec, I bought parts enough to build at least 25. Looks like I’ll have to either make my own circuit boards or order more from Adafruit.