VG Pocket handheld game problem (dead)

[bluto]

One of my younger customers has a Disney Princess hand held video game (VG Pocket VG-9001) that is apparently dead. The LCD back-light and top LED come on when the power button is pressed, but that is all. Nothing on the screen and no sounds. I tried pressing buttons but still no indication of life. The device will not switch off with the power button, the batteries must be removed.

I was told that the game 'just stopped working one day'. There were no indications of impending failure. The device looks like it hasn't been roughed up too much.

I took it apart and stated replacing capacitors ('cause that's what we do here!). I checked each one that came off (and the two originals that I put back on) for capacitance and ESR and they all checked out OK, so I don't think it is a cap problem. There were no signs of leakage either.

I am looking squarely at the crystal as the possible culprit. It says 6.000 on the top which I assume means 6MHz. I am not sure how to test it definitively. I pulled out my scope-meter and touched it to the top of the can (was soldered down and seemed to lead to the dab 'o glue cpu). The results are pictured. I got a 'triangle' looking waveform. There are two pins soldered to the board underneath the can and probing them gave a weirder looking waveform.

Any thoughts on how to test the crystal better? Any tips for replacing it and hints for finding a matching part on DigiKey?

(Mind the fact that the ribbon cable to the LCD is unplugged in the picture)

[bluto]

I found about a dozen 6MHz crystals on DigiKey that look like this:

http://search.digikey.com/scripts/Dk...name=XC1712-ND

Do I need to worry about load capacitance of the crystal?

[Radio Fox]

I can't see what the timebase setting is set to on your scope. Is it reading the right frequency?

[bluto]

Sorry for the noob-ish question: What time base should I be using? I have not used the scope much since my uncle gave it to me.

Another question, If I were to replace the crystal, does the orientation matter? Or can it be installed in either direction?

[shovenose]

lol, im envious. i wish my uncle gave me oscilloscopes. well he did buy me a ipod video, an ipod touch, a macbook, and an iphone (the only one of which i still have is the iphone...i sold the ipod touch after a year, and i later sold the macbook on craiglist because it ran too hot...

and the point of my long rant? to say that my iphone sucks at everthing it was made to do: the buetooth never worked, its slow, drops about 3/4 of the calls. and this is the iphone 3g not the 4...

tbh i still appreciate it like crazy, it is my first "phone"

[stj]

crystals go in either way.

[bluto]

Well I messed around a little more. I still am learning how to crawl, but I think I am getting somewhere. I started by hooking the black lead from the scope to the negative battery terminal (apparently important). Then I probed both legs of the crystal and took pictures of both waveforms. The scope was set on a 2ms timebase. The first leg produced a waveform that looked very sinewave-ish. The second was a little less sinewave-ish. Each were stable over time, but did fluctuate a few times (or it could have been the operator ).

Any ideas on this one? I still don't know what I am looking at exactly. Is this telling me definitively that I have a good crystal, or is it somewhere along the lines of 'definitely maybe'?

[smason]

I've forgotten how to do the math to determine frequency from a waveform on a scope, but it looks like what I would expect. I'd go out on a limb and say it's working fine.

[shovenose]

Dont these thingw cost like $5? But then again its more fun to fix it. Are u sure a main ic isnt just fried?

[kc8adu]

looks like the clock is running.i would focus on the flash.corrupt or broke loose.its to the right of the cob and has a p on it.the fact that the power button turns the unit on but it gets stuck in a bad state and cannot be turned off tells me its software.

[kc8adu]

and for another test check out u1.
appears to be a +3.3 regulator.i suspect its ok as the thing starts up.see if it stays in spec when the thing locks up.

[Radio Fox]

Looking at those waveforms, you are not reading the crystal frequency, but some other lower frequency.

To convert frequency to time period or time period to frequency, you just invert it (T=1/f or f=1/T). So, 1Hz = 1S, 1kHz = 1mS, 1MHz = 1uS 1GHz = 1nS. A frequency of 6MHz will have a time peroid of 0.167uS (= 167nS).

You are seeing roughly 2.6 cycles for each 2mS division, so approx 0.77mS which is just 1.3kHz. This could be a reading of the crystal frequency after it has been divided down by one of the chips (6MHz/4096=1.46kHz).

Also placing probes directly on the crystal can cause the frequency to change or even stop. You would normally see low value capacitors connected to the crystal. The extra capacitance of the probes could upset the working of the circuit.

btw...does your scopemeter have cursors or markers? If so, you can use these to get a direct frequency reading. Line up a vertical cursor, or marker with any part of the waveform, then adjust the second cursor, or marker to line up with the next occurence of the same part of the waveform.

[ratdude747]

you know these toys are throwaway junk... they figure by the time it dies, the kid will have gotten bored with it or a new and better version will steer kids away from he old one.

good luck fixing it... more proof why quality is hard to find these days...

[bluto]

Understand. It is just that the customer was attached to this one. I may have to buy another one so I thought I would give it a shot to try fixing it. Seemed like it should be repairable. And it looks like a cleanly made board. The only thing that I don't understand is the 'dab of glue' cpu. Is that when the manufacturer draws a custom part right on the board and covers it with a glue-like substance? I guess it is cheaper than fabbing a fully packaged chip? Possibly due to its application-specific nature?

I am amazed that throwaway junk technology has come this far. 12 games in one, color LCD screen, etc. beats the crap out of the hand held 'simon' game that I had as a kid.

[bluto]

Checked a few things. I get 3.6v out of the regulator. The other IC (u6) is the audio amp. I get 3.3v going in. I pulled the data sheet on the flash chip. It is a Macronix MX29LV320ABTC-90G - 32M-BIT. On weird thing, the manufacturer logo was mechanically scratched off the top of the package (visible in the picture). Would this tend to indicate that this was a 'factory second' aka 'picked up off the factory floor' batch of flash chips? It would help keep the cost down...

[Badcappy]

>only thing that I don't understand is the 'dab of glue' cpu. Is that when the manufacturer draws a custom part right on the board and covers it with a glue-like substance?
It's called COB (Chip-On-Board), essentially a bare die that's wirebonded directly to the board and covered with epoxy. It's quite common in high volume, low cost applications.

My guess is a corrupted flash. Somewhat unusual, since that is NOR flash and should be more robust than the NAND parts that are more common these days. Using chips that failed QC in some minor way is not uncommon for a cheap non-mission-critical device like this. Even though it's Disney, who knows what OEM they use.

I doubt you'll find firmware for it anywhere, so unless you want to dump the flash and embark on a possibly interesting reverse-engineering project, consider it scrap. Maybe the LCD and caps are still good...

[kc8adu]

i would expect many very cheap items like this to contain low grade or even recycled parts.flash chips get binned and the top grade go into name brand and critical apps.lesser bins get derated and tested.like at a lower speed.the real dregs go to counterfiet flash drive and crappy trinket makers.like this toy.
if i sold salvage i would destroy the logo too.

Originally Posted by bluto

Would this tend to indicate that this was a 'factory second' aka 'picked up off the factory floor' batch of flash chips? It would help keep the cost down...

[bluto]

Makes sense. I guess I will consider this one toast. And thanks for the great insight. Even when I can't fix something, I still learn amazing things from all of you that will help fix the next item!

[shovenose]

Im surprised about the 4megabyte flash chip

[smason]

Originally Posted by Radio Fox

To convert frequency to time period or time period to frequency, you just invert it (T=1/f or f=1/T). So, 1Hz = 1S, 1kHz = 1mS, 1MHz = 1uS 1GHz = 1nS. A frequency of 6MHz will have a time peroid of 0.167uS (= 167nS).
.

Thanks. Having a good frequency counter has made me lazy, and stupid