[NTLK] Implant 2017
jake at allaboutjake.com
Sun Mar 26 09:28:19 EDT 2017
So here’s my first prototype: http://i.imgur.com/DJhC1Hd.jpg
One of the things I’ve been playing with is overclocking the newton. PixSolutions used to have a product called the “Implant” for overclocking. It was essentially a clock generator that could be controlled with the power switch. By default the Newton would book in 222mhz mode, unless you held the power switch on wake, where it would then fall back to 162mhz.
I decided to take a different approach for prototyping. Since I know how to control the GPIO line of the internal serial slot, this could be used to select the clock speed. Also there are modern clock generators that can make almost any frequency, and can switch between them “glitchless” so that changing clock speeds at the wrong time doesn’t crash the processor. This allows you to overclock the Newton on demand from software.
There’s a minor error on the PCB (I left off a trace) that is easily patched by bridging a pin to a nearby capacitor. I have 330-ohm resistors that are supposed to allow you to program the ATTiny in circuit without the clock generator IC interfering, but it seems unreliable. (Could also be my programmer) You can program and then add the resistors or program the ATTiny before soldering it to the board.
You still have to desolder the crystal and hook up to the same pad as you would for the PixSolutions product, but there are fewer wires to hook up then the original Implant, since it is powered and controlled by the serial slot.
Basically there’s an ATTiny85 processor that watches the signal from the GPIO line and sends commands to the clock generator when a clock switch is requested. I read in the SA1100 docs that you’re supposed to use 1v peak-to-peak signal when clocking the processor from an external clock generator. I tried this and it didn’t seem to work, so I’m doing the same thing that the original implant did, and just use a 3.3v signal. (Not sure if this is a good or bad idea)
Probably won’t be building these for sale or anything. I’ve found that my reflow toaster solution isn’t the best, and tends to overcook boards, and I’m going cross eyed from manually placing the small surface mount components. I can publish the design for anyone who’s interested, along with a utility to toggle the GPIO pin as desired.
It’s a good proof of concept, and can be incorporated into other designs if desired. If we are able to develop a new ROM board, we might be able to push the speed even more, but I’m not sure at what point things will start to overheat or fail.
More information about the NewtonTalk