Stephanie's MP2x00 Overclock

The Newton MP2000 and MP2100 are both pretty fast to begin with, running the StrongARM processor at 162 MHz. Compare that to the first few generations of Palms at 16 MHz. However, there are some times when we might like our Newtons to really fly!

PixSolution makes a commercial product which will speed up your MP2x00, but it's expensive, requires shipping from overseas, and is basically out of my price range. The product, called Implant, does have advantages though, and is worth considering if you can afford it.

As an avid electronics / hardware hacker and hobbyist, I checked out what could be done on the cheap. I soon found Sine's web site where he outlined a planned MP2x00 overclocking (follow the 'Projects' link). What Sine was discussing is very straightforward, and I found the parts readily available at a nearby electronics shop. This overclock cost me about $10.00 Canadian.

For a look at another angle, and another set of step by step instructions and images, check out Ed's overclock page.

So, on with the project details.

Disclaimer and Warning: The information provided below is provided for entertainment purposes. By attempting any modifications to your Newton you will definately void any remaining warranty you might have, and could destroy your Newton completely. I am in no way liable or responsible for your actions or how you use the information below, nor am I liable or responsible for any unexpected or undesired results arising from the modifications described on this page. Soldering experience with Surface Mount components is required. This is definately not for beginners. You have been warned.

Theory, and Stuff

The StrongARM processor runs at about 162 MHz. This speed is achieved by means of a tuned quartz crystal and a frequency multiplier. The crystal used in the MP2x00 is a 3.686 MHz crystal. So the crystal frequency is multiplied by about 43.97. The top speed that has been reported for accellerated MP2x00's is around 220 MHz. Divide that by the multiplier and we need a crystal of about 5.0 MHz.

Newtons accellerated this way, have a few 'drawbacks'. The sound is sped-up (think Chipmunks). The serial port is unusable, as is the IR port. Some PCMCIA cards stop working (primarily, some modems and some ethernet cards. Memory cards seem to work fine). The time & date are unaffected, though.

So, we don't want a permanently accellerated Newton, but one which we can switch speeds at will.

To achieve this, the quick -n- dirty method is to just use a switch to choose one crystal or the other. The crystal has two conductors, one needs to be switched, the other can be common. I have wired it this way. In the following image, you will see my modified Newton. You are looking at the back of it, with the IR port at the upper left hand corner.

Click here for a larger picture (330 K)

Click here for a huge picture (700 K)

In this image I have removed the original crystal, it used to be in the area indicated by the light green rectangle; above the large Cirrus Logic chip and below the small Cirrus Logic chip. With it gone, I have soldered two pink wires to it's former circuit traces. The one on the left is the 'common' one, and the one on the right is the switched one. The two crystals are positioned, as indicated by the blue rectangle, overtop of the StrongARM processor, using the age-old electronics component, sticky-tape. The slide switch is at the top of the image in the purple rectangle, hidden under a blob of hot-melt glue (another electronics standard).

The smaller crystal is one with the same specifications as the MessagePad's original one. The larger one, as you can see, is 5.000 MHz. You will see that the two crystals are soldered together, the lower pin of the small crystal to the upper pin of the top crystal. The pink wire soldered here is the common one, which goes back to the left-hand trace above the large Cirrus chip, as mentioned previously. The remaining pin on each crystal also has a pink hookup wire soldred to it, and those two wires go up to the top part of the image, where they join to either pole of a miniature slide switch. The common contact of the slide switch, is connected by jumper wire, to the right-hand solder trace above the large Cirrus Logic chip.

In the image above you can see the tiny swich as indicated by the arrow. I just put in a new switch recently, that is easier to activate. The small black dot is the part that slides. To make the hole, I started 'nibbling' away at the case plastic just to the side of the large square hole where the mini-din goes. My advice is to be very careful in that area because *Snap* and away went more plastic than I intended. As you can also see, I've lost the 'door' that's supposed to cover this area, its hinges just seemed to disintegrate one day.

As you can see by this schematic, the circuit is incredibly simple. I must caution though, that while the design is simple, the implementation is anything but. Please re-read the warning and disclaimer above. This is difficult, and there are real risks to your Newton.

Discussion, Pros & Cons

Someone experienced with soldering and electronics, might expect to spend about 30 to 60 minutes on this hack. It requires a steady hand, a good eye, and strong nerves. Actually, two steady hands are preferable.

Someone who has little or no experience with soldering and electronics, should perhaps be discouraged from attempting this as their first project. I would suggest doing some kit projects first, to learn soldering skills.

The benefits of this home-brew overclock, versus the commercial product, are twofold: The price is incredibly attractive, and you have a brute-force no-holds-barred way of turning the overclock off (the switch) when you want to use serial / IR / etc. I have heard of an instance where the commercial product got stuck 'on' and the Newton could no longer be linked to a desktop for installing other packages.

One of the benefits is also the main drawback: The switch. There is no way of controlling what the Newton's processor is doing at the moment you throw the switch. It is possible, very possible, to crash your Newton by switching it while it's busy. In practice, I do not switch speeds unless the Newton is either sleeping, or at very least, all packages are closed (only the Background running). Sticking to this, I've had about 1 'crash' in 50 switches. When crashed, the Newton requires the reset button to be pressed. I have never lost data doing this, and I do not think that the electronics themselves can be damaged by this. (See the disclaimer again! Your mileage may vary!)

A Newtontalk reader has suggested that the small lengths of hookup wire may act as antennas, radiating the crystal frequency and/or harmonics of it. To minimize this risk, hookup wires should be kept as short as possible. The Newton's case has some shielding which will also minimize the risk of stray RF being radiated out. If you use your Newton in an RF-sensitive environment, you may want to think twice before doing this modification, or refrain from using the modified Newton in such locations. Once again, I do not take any responsibility for unexpected / undesired results arising from the modifications outlined on this page.

Results

As expected, the serial port was unusable. The IR port will only work if you are beaming to another Newton accellerated by the same ratio. Otherwise, it's inoperable at high speed. My modem card - a generic 56k unit - works fine. My ethernet card - a 3Com 3C589D - works fine, nice and fast. And (joy oh joy) my 802.11b card - Lucent Orinoco WaveLAN Silver - works fine too, roaring wireless speed. Too cool for words.

PixSolution has some freeware available on their website, and one of their programs does a status test that probably is like a 'diagnostic' for their Implant product. It has the nice benefit of showing the current clock speed, so I have used it to get the following 'before' and 'after' screenshots, so you can see the difference in processor speed.

Before	After

Parts & Tools Needed

Parts available at better electronics shops. Online, try DigiKey.

Y1a - 3.686 MHz Crystal

Y1b - 5.000 MHz Crystal

SW1 - Small, or Surface Mount Slide Switch: SPDT

Fine Hookup Wire

Fine electronics solder - Rosin core (Not Acid core!!!)

The two crystals I used are the 'parallel, 18pf' kind (as opposed to 'serial'). I know there's a difference, but I'm not sure what it is. For the switch, if you have a choice, break-before-make is what you want. This means that as the switch moves, there is a moment where neither crystal is selected. My experience has shown me that this is preferable to having a moment where both crystals are selected.

Tools available at electronics shops, even Radio Shack

Fine tipped soldering pen. 15 to 20 watts maximum!!!

Tweezers

Fine probe or needle

Step by Step Instructions

1. Back up your Newton.
2. Back up your Newton.
3. I know it's non-volatile Flash memory. I don't care. Back up your Newton.
4. Remove stylus, both PCMCIA cards or card blanks, and the battery. Remove the screen door.
5. Place your Newton face down on a level surface, perhaps with a soft cloth under the Newton to protect it's screen.
6. Remove the four Phillips screws from the back of the Newton. Put them somewhere safe. Remember which holes they came from. The screws are interchangable, but there are more holes than there are screws.
7. Remove the back cover. This requires some careful handling. Start at the battery opening, work your way along the bottom and up the hinge side. If you need to, use a slot-screwdriver blade to help pry. Be careful though not to damage the case.
8. Once the case has popped off, set the back case aside for now.
9. Now is a good time to decide where to put your switch. I stuck mine near where the Interconnect Port is. I cut away part of the case underneath that little protective I/O door and stuck my switch there. Another option is to stick it in the available open space that was for a phone jack or built-in mini-din port. This is not an option, however, if you have a SER-001.
10. Locate the existing crystal, using the above image. It will be a short silver-coloured rectangular metal 'can' above the large Cirrus Logic chip and below the small Cirrus Logic chip.
11. This crystal must be removed. This is the single most dangerous and difficult part of the procedure. This crystal is Surface Mount, most of the solder you need to remove is hidden underneath the crystal body. If you rush it and try to pry it free, you will probably tear the copper traces off of the circuit board - And kill your Newton in the process.
12. The best way is to use a magnifying glass, a needle, and a fine tipped soldering pen. Go one side then the other, heat, pry gently, then switch sides. Expect to take up to about 10 to 15 minutes. One person suggested adding a bit of solder to the tip of your iron, to maximize heat transfer into the existing solder under the crystal.
13. Once you have removed the old crystal, set it free. Or, if the leads are in good shape, you can opt to re-use it, since you know it's definately the right speed for the low-speed setting.
14. Prepare the new crystals, soldering two leads together as I have done, and position them with glue or tape or whatever.
15. Use your hookup wire, to solder from the two common leads on the crystals, to the left-hand trace of the original crystal.
16. With your switch positioned, solder hookup wire from it's common terminal, down to the right-hand trace of the original crystal. Nearly there.
17. Now solder from the unused pin of your 3.686 MHz crystal, a hookup wire over to one of the remaining terminals on your switch.
18. This is a good time to actually test things. Set the switch to the terminal you have soldered to the first crystal. Carefully turn your Newton over, plug in AC power, and power it up. If it works, great! If it doesn't work, remove AC power and inspect your work. Try not to let your nerves or churning stomach distract you.
19. If all is well, then proceed.
20. Solder a piece of hookup wire from the unused lead on the 5.000 MHz crystal, and to the remaining empty terminal of the switch.
21. You can now test this setting. Set the switch for the fast crystal, and power-up your newton. It should power up quicker, and the 'Chime' will sound faster and higher-pitched.
22. If it works, then use some tape or non-conductive glue to cover any exposed leads or connections, to prevent things shorting out. Test it again, and if all works, then you can put it back together. Just reverse the steps for taking it apart.
23. If it doesn't work, then once again, inspect it carefully. Look for shorts. Look for solder bridges. Make sure the crystals are wired properly. If all else fails, try soldering your new 3.686 MHz crystal directly to the two traces where the original crystal came from, and test that. If that works, you know you had a wiring problem. If that doesn't work... well you read the warning and disclaimer, right?