Hijack v457 is now available.

This version implements a proper workaround for a dead cs4231a chip, so that full functionality of the player is preserved -- except for visuals in Tuner/Aux modes.

So, in many (or most?) cases, a player with a "cs4231a not found" condition can still be fully used, with the only caveat being that the front screen visuals no longer respond to the beat of the sound when in Tuner/Aux modes (visuals work fine for hard disk playback of mp3s etc.).

This should help save a number of players from the soldering-iron / scrap-heap when the cs4231a dies.

------------

The previous release of Hijack was v456, which added support for configuring/using kernels that mount their "disk" over NFS -- complete diskless empeg operation. This requires a kernel reconfig/rebuild to use. No changes for normal use though. This patch courtesy of Thomas J.

Cheers

Hijack v458 is now available.

This version improves upon the dead cs4231a workarounds from Hijack v457, fixing the "slow mute/unmute" issue.

It also slightly reduces the kernel size, by removing the "recent samples" reporting from /proc that was added a few versions ago.

Cheers

You, sir, are THE MAN.

Hm. I just tried to edit the FAQ and add this information to the "no hard disk found" section of the FAQ, and I saw no results from that edit. It appeared to allow me to MAKE the edit, but after the edit was completed, the text I changed did not exist in the FAQ. Tom, any idea why?

Use the wiki, Luke

You may well argue you are looking out for Number 1, but you are making your fixes / mods / upgrades public, so we all get to benefit as well. Many thanks.

Close-up (blurry):



This unit had a botched cs4231a chip replacement, with several pads missing and misaligned, as well as the usual IDE header issues. I had previously written it off as a goner for spare parts.

But with the recent Hijack kernels, a cs4231a chip is no longer necessary. So I grabbed a Dremel tool and cut the chip free from it's pins, and then took copper braid and a soldering iron and cleaned up the pads (much better than the picture makes it appear).

It's working very well now!

Close-up of the damaged area.



The barely visible whispy-thin debris around the pads is cotton from the Q-Tips used to swab the area with cleaning alcohol.

Impressive!!!!

I wonder if anything interesting could be done with that board real estate...

I believe that you could successfully get a new chip onto that.

Thanks so much, Tom! Working now. Updated the FAQ.

Dredging up an old thread of mine.

This is one of the "dead" units that I acquired from another BBS denizen once upon a time ago.

It worked for me here for a while after the "fix" described above here, but then stopped working, giving tons of I2C errors at boot.

Lately I have been using it to trace circuitry while repairing other units, and now that all of my other players are working well, I figured I'd tackle this one again.

Using the LogicPort analyzer on the I2C revealed that the CPU was sending correct data to the DSP, but the DSP was not ACKing the packets. I then hooked the analyzer up to the DSP oscillator circuitry -- pin5 of the nearby 74LVX74 chip, and discovered NO CLOCK.

Which is odd, because the DSP clock circuit is quite simple on this board. There's just a clock IN, and a clock OUT on the DSP, and then a crystal and the dual flipflop (the 74LVX74).

Oh, wait, there's also a clock POWER pin on the DSP. Mmm.. 0V.. that could be a problem. I tracked this back to where it originates, and that's the working 3.3V supply. So not that.

In desperation/inspiration, I took my trusty soldering iron to the clock POWER pad on the DSP (pin 65), and then also touched up the feeder coil pads (silver can thing, near the far edge of the board beside the SA1100), and then checked the clock again.

Woo-hoo! Nice little square-waves on the LogicPort display now. So I plugged in a hard disk and listened to The Guess Who for a spell.

Now to reassemble it all and perhaps find a home for it.

Cheers

One of the pads in the picture though does have a trace going to it, so it might be important... or not. I don't have the datasheet for the 4231 to hand. The pad in question goes to a via though, so you could always run a patch wire from there if you can't follow what chip it goes to next.

Hugo

Yeah, there are two missing pads that would normally connect to a via there. If I ever get hold of a new cs4231a chip to install, I'll jumper those back into place.

Cheers

In case you missed it, the mystery of cs4231a "failures" has been unraveled here.

A simple pin-lift and jumper wire are all that's needed to work around the design flaw.

Cheers

The missing cs4231a has now been successfully replaced.