I´m Doing A Voltagemod.


Fishbong

man of poor reputation
Joined
Apr 23, 2004
Messages
2,730
Age
38
Location
Hildesheim, Germany.
Website
Visit site
Allright, at the PSX4gp2x beta tester thread we have been discussing the possibilities of a Voltagemod to increase the GP2X´s overclocking capability.

What is a Voltagemod?

A Voltagemod or V-mod is a little hardware modification to increase the core voltage of a processor. This makes the processor run more stable at higher speeds, however it may be very risky: The processor consumes much more power and can get thermally instable. This can lead to its quick destruction. Therefor, it is important to increase the voltage in small steps. Another disadvantage is that the higher power consumption will lead to a reduced battery lifetime.

How is it done?

This is not (yet) a tutorial. I haven´t done it yet and have still to gather information and make some measurements to find out how to do it safely.

However, i have already taken the first steps and have a clear picture of what to do.

OK, the gp2x´s processor is a MP2520F "magic eyes". It has a core voltage spec of 1,8V. That means there´s got to be a little stepdown converter on the GP2X´s board. Those mostly consist of one or two capacitors, two resistors, a coil and, most important, a small, simple chip. The task is to find this chip and find out how to adjust it in order to increase the output voltage.
OK, i found it. I´m quite sure its this chip (in the red circle)




Now to the adjustment.
Usually those chips are being adjusted by two resistors using this formula:
formula8me.png

R1 and R2 are the two resistors, on the picture marked with a blue circle. I have to unsolder them to measure their resistance and then i can use the formula to find out Vref ( given that Vout is 1,8V). then i can change the resistors according to the formula to increase the output voltage (Vout) in small steps.

I won´t be able to do this today, expect updates inthe next couple of days.
 
Last edited by a moderator:
It looks a bit like a regulator. If it is then you could replace it with a 7805, It requires 7v+ to run but outputs 5v. Also since you want to do it in small steps i'm pretty sure there are some adjustable regulators out there.
 
dj-spuddy posted on May 25 2006 at 08:17 AM said:
It looks a bit like a regulator. If it is then you could replace it with a 7805, It requires 7v+ to run but outputs 5v. Also since you want to do it in small steps i'm pretty sure there are some adjustable regulators out there.

God what are you smoking?

He is going to change the resistors and raise the voltage.

If this works then all we need to do is use pencil lead to write on those two resistors until it changes them to another value, hello pencil mod!!

I would much prefer a micro Potentiometer and use a jeweler's screwdriver and a DMM (DVOM) to adjust, I would just need to hide it under the battery lid so I can adjust it from outside for kicks.

But I know me, I would just grab a pencil and color on those resistors while checking it with a DMM.

*Digital Volt Ohm Meter, Digital Multi-Meter
 
Last edited by a moderator:
Well, while I don't know too much about the overclocking itself, I do know for sure that the chip you found there is the step converter that goes poof when you use the gp2x with an AC adaptor at wrong polarity or wrong current.

It's also broken if your gp2x is only running with a PSU but not with batteries anymore.
Did change it quite a few times now, so I'm pretty sure about that.

If it still has something to do with the voltage for the CPU, I don't know.
 
Well, I hope all goes well with this mod. I was planning to do it on my new gp2x, but I decided, I wont void my warrenty on the new one.
 
I just had a talk with ED, and found out that the chip in question does most likely not control the core voltage. That means there´s got to be another stepdown converter. Hope i find it soon.
 
The ARM920T core voltage is taken in on its own pin. It seems to utilize, by default, 1.8vdc. It can run at a minimum of 1.6vdc or a maximum rated 1.95vdc. However, you may be poking around a lot before you find the exact lead that carries the 920T core voltage since the MMSP2 is a FPBGA package and none of the pins penetrate the board.

Isolating the ARM920T core voltage and not overvolting the whole MMSP2 is a much wiser move, as you will extend battery life (components like the scaler will also have increased voltage, and they use much more power than the 920T, apparently).. increase stability to the SoC by not taking more components out of spec than needed, and decrease heat output (although this is really a non-issue, odds are.) You might also want to tie in a filter capacitor to ground to remove stray AC interference that might be making the core voltage uneven-- this will improve stability, if only slightly.

Best of luck.
 
Good luck Fishbong! :)
Hope you don't kill your GP2X! If you do though, be sure to put up a donation link so I can donate. :p

Btw, check this out. It's on the magiceyes development forum. Anyone know what this is about?

http://forum.mesdigital.com/development/do...fnc=read&idx=47

Title: Core Voltage value change
Quote: "Voltage value change of MP2520F related to internal FPLL.doc"
Filename: [Notice]Voltage_value_change_internal_FPLL.pdf



If you need to know the exact pin, the GP2X Wiki has that info on it in a PDF. I can look it up if needed.
 
zodttd posted on May 25 2006 at 11:30 PM said:
If you need to know the exact pin, the GP2X Wiki has that info on it in a PDF. I can look it up if needed.
Thank you, but as epicenter already said, we can´t access the pins anyway, so it´s not really important to know the exact pin.

In other news, there´s no second stepdown chip on the gp2x´s pcb, so this one has to be related to the processor, i just don´t know how closely.

I made some measurements and found out that there´s a voltage of 1,82V between pin 4 and 8 of the stepdown chip. At first i thought this might be the voltage source of the processor, but now i really don´t know.

Evil Dragon gave me the exact ID of the chip, so i´ve been looking into its datasheet, and what i found is a bit disappointing.

tyicalapp6ai.png


This is the layout for the simplest application of the chip.
The pin 8 is FB, the feedback pin. It reads the resisance of R1 and R2. This is needed to determine the output voltage.
It is close to impossible that someone would connect an actual device to the feedback pin as it would mess up the feedback from R1 and R2.

So, at the moment i´m not sure what to do.
 
Last edited by a moderator:
Fishbong posted on May 26 2006 at 01:28 AM said:
So, at the moment i´m not sure what to do.
Can you trace the 3.3volts out as far as it will go? Does the 3.3volts lead to the 920T power circuit?

Is there any possibility of one of those tiny SMC transistors being a power supply transistor?

Check for Resistor pairs next to them, I am sure I saw a data sheet for a 7805 that was a micro surface mount with 3 legs, could be something.


Did you see that 3.3V line @1 AMP!!!!

Wow, that is Craaazeee. A lot of current for sure.

I am so using 3 NiMH when I get a third GP2X, I will get a single AA holder and tie it to the back of the battery cover, then I will feed 3.6 volts to this chip all day long :) :). (To connect it I will use zip ties, or screws, heck I could even buy a spare cover and epoxy it on there, or above the bulge for that matter.)

It might make a lot of difference to the battery life.

Does anyone remember the console that used 3 AA, was it the Lynx 2?
 
Last edited by a moderator:
Well, that answers what the max safe voltage to use as DC input is: 5V. (Although, the application note says the acceptable range is 1.8 - 3V, not sure if they're just writing typical battery voltages, or if it really won't work properly above 3V in that setup)
 
BradN posted on May 26 2006 at 10:48 PM said:
Well, that answers what the max safe voltage to use as DC input is: 5V. (Although, the application note says the acceptable range is 1.8 - 3V, not sure if they're just writing typical battery voltages, or if it really won't work properly above 3V in that setup)

I am the info man (Im a googler :)):

http://www.linear.com/pc/productDetail.do?...031,C1061,P1901

Check out this link, here are pertinent qoutes (bold mine):
# Synchronous Rectification: Up to 97% Efficiency
# 2A Switch Current Rating
# Fixed Frequency Operation Up to 3MHz
# Wide Input Range: 0.5V to 5V
# Very Low Quiescent Current: 38µA (Burst Mode™ Operation)
# 2.6V to 5V Adjustable Output Voltage

Um, here is the upconversion chip that we need for making the USB fully functional, just saying guys.

Use 3 AA and another one of these and real USB Host looks possible. (I know it feels weird, but I think it will catch on).

On a side note I have a sweet solid-state monster cable brand PSU that makes 4volts @900mAh, I think I will just feed that to the battery terminals, this chip will take it and I don't need to use that dumbass side plug in for power (this thing is tiny, I can carry it anywhere with my GP2X).
 
Last edited by a moderator:
Fishbong posted on May 25 2006 at 08:12 AM said:
OK at the gp2x´s processor is a MP2520F "magic eyes". It has a core voltage spec of 1,8V. That means there´s got to be a little stepdown converter on the GP2X´s board. Those mostly consist of one or two capacitors, two resistors, a coil and, most important, a small, simple chip. The task is to find this chip and find out how to adjust it in order to increase the output voltage.
OK, i found it. I´m quite sure its this chip (in the red circle)


Just a quick thought, what is that group of 9 pads doing there?

Check out what is happening there, is it possible all the voltages are checkable there? Or is it possible that is helpful at all for checking the different voltages needed?

If we can find the pad for checking the voltage to the 920T at the factory we can feed it some juice just for experiment's sake and see how it likes it, or just start tracing from that pad back if we can.

If they were just there to join layers of power or Ground, wouldn't they be screened over like all the others in this picture?
 
Last edited by a moderator:
Back
Top