PDA repair

I've wanted to post this for quite some time now, but laziness has so far stopped me from doing so, and since this was finished quite a while ago, some of the details are already a bit hazy, so to keep the info from deteriorating further, I've decided to do it now.

A while back I got a broken PDA for a relatively good price. Broken screen and mobo failure. It was some HP model I don't really remember. After opening it up I found that there was one fried component on the mobo, all the SMD fuses were intact, and since the fried part was a Maxim I decided to free-sample it and try a replacement. The part was a MAX1586, a "Power Management IC" for PDAs in TQFN40 (6mm x 6mm). It seemed to handle pretty much everything power related for the PDA. The original part was a bit melted-through and would get real hot, real quick, when power was applied, so it obviously needed a replacement.

Desoldering TQFN40 without any of the usual tools meant for doing so is not easy :)
What I went with was a hot-air blower, that wasn't at all meant for SMD soldering. (I think it was meant for burning away old layers of paint)
The mobo was wrapped in aluminum foil, in an attempt to shield other components from the hot air. This was mostly ineffective.


Below is the mobo after the the extra blown-off components were hastily reattached, before cleaning up the mess.

I free-sampled MAX1587, which is a bit higher power than the 1586 was, no special reason behind this, if I remember correctly it was the only part available for free-sampling.

As for soldering in the replacement that was mostly done with luck, an ordinary soldering iron, and lots of soldering resin ( I didn't have no-clean flux at the time ).
Thankfully TQFN pads have a "shoulder" that allows them to be soldered in with a regular soldering iron. This was not easy, nor would I recommend it to anyone, but it did turn out "good" in the end :)(

Below, the mobo with replacement PMIC in, after cleaning up.


After about an hour of playing around with setting the chip right, the device did power on. I don't remember if by that point I already had a replacement display, or if I ordered it afterward, but after that was installed the device came to life, functioning perfectly, complete with the original owner's data, which I proceeded to wipe. I kept the device around for about a week, then sold it, with very little financial profit, but lots of experience gained.


Notice the broken watch effect :)

Glossy paper toner transfer test

Part of a circuit onto a junk piece of pcb, with glossy paper result:


While it does seem OK at first look, after further testing I would not recommend this method to anyone, as the results are not so spectacular. I'll stick with using satin-matte photo paper and PnP Blue Transfer film, as they both provide superior accuracy.

Surface mount hot-air soldering Iron

Here's my hot air SMD Soldering Iron.

Based on this one.

I used a different desoldering iron design than the radioshack one, added a real cheap aquarium airpump. And I mean REAL cheap.


And the end product is this:


The original tip's opening was too large and trying to pinch it resulted it in breaking so I went looking for a replacement and found that the tip of a metal refill pencil works great.

One advice for this type of iron.. You don't want to unscrew those two screws when changing the tip. If you do and start turning the tip you WILL ruin your iron. Since I'm used to oldschool soldering iron I did it without thinking. I ended up having to order another one.

The mod is really quite simple, It's basically what's visible in the picture plus adding some desoldering braid into the pipe and tip to achieve better temps. After a 20 minute preheat it melts semi-large lumps of solder in 2-3 seconds.

Salvaging LCDs from all the wrong places

I've always had this urge to salvage components from unused, unneeded, broken gadgets, especially when it comes to LCDs. This one is in an Ascom Eurit 22 ISDN telephone, the handset I've already modified (added a pair of jacks ) for use with a PC, was using it mainly for Skype.

The base unit however resisted a previous attempt at modification where I tried reading the flash and modifying the firmware to make it do something else other than what an ISDN phone is supposed to do. There wasn't much info on the MPU either.. I remember finding some compiler for it on a Renesas site but never actually got down to anything serious because of the issues with reading the flash.

A few days ago out of boredom I took another look at the PCB and the display controller. It's a PCF8578T from Philps. I realized that it works through I2C so I booted up Knoppix , quickly compiled milksop's milk app, connected the I2C pins to the lpt port and started a CheapI2C sniff, after several failed attempts in Windows with BSLA's I2C decoder (had to poke the exe into actually doing anything without a BSLA besides showing demo data) I had some trouble with milk app at first because it doesn't like hyperthreading.

After disabling hyperthreading it worked like a charm though. I got the output:

Milksop GPL Reflasher - 0.20 - (c)2002 andy@warmcat.com CheapI2C on printer port at 0x378 Entering capture mode, your PC will appear frozen - press any key to end Starting realtime... Stopped realtime... Monitoring completed 12.064mS: 3C W 55 1.867 S: 3C W E0 88 70 00 28 00 44 00 28 44 00 00 00 00 00 00 00 00 00 00 00 00 00 00 28 3A B8 1B B0 01 28 00 . p . ( . D . ( D . . . . . . . . . . . . . . ( : . . . . ( .

Now 3C being 0111100 is the I2C Slave ID for the PCF8578, 55 is a SET MODE command that just sets up and activates the display, and on the next line is an actual RAM ACCESS command that is used to modify the display ram aka. write to the display. I did several I2C captures this way.

Then I cut the trace connecting the PCF's I2C pins to the phone's MPU and coded up an app to drive it from the LPT port. For every character on the display 2 bytes are sent after E0 88 70, then after the 14 characters of the display there's a final byte the works the last 3 icons ( the A B C )

The configuration is like this:

So here's the outcome. Driving it off the LPT port.