Simon, ICQ #173250711 

Today i tried to get the switch chips to work with the gumstix and driver i wrote earlier (see part4) without success. All switches are to stay open! The ones i got are TC9163N instead of TC9163AN may that be the problem (negated inputs or so (hm, i should try that)). Unfortunately i can only find datasheets for AN and not for N :(

I still have bypassed the switching and hooked up my speakers just to listen if the sound is ok anyway (only used test signals and headphones until now and square waves don't make you happy). Once turned on i just couldn't stop listening and going through my music collection... sooo beautiful! It is so great to hear fingers moving on snairs and little vibrations in the voices and all other tiny details that make the music live. And finally the musics suggested by Özlem start to reveal their secrets to me :-)

Addition i forgot to add yesterday: I also can confirm now that an excessive load has been the cause of all the troubles. It blew the input and record switches. As a result that the CK and DATA lines of the input select have -14V on it. Which blew the CK line of the big fat luxman driver chip, hence the missing clock signal. Fortunately it didn't seem to cause much more damage. There is one more thing though which i found out yesterday. If i select direct input, the overload protector relay clicks on for a 10 to 20 seconds after which it releases and everything seems fine again. But it indicates some bad somewhere which i need to look into.

Second addition: Just found out i stupidly enough ignored the chip select bits for the switch array. Adding a high for bit 11 did the trick. I uploaded a new and slightly improved driver here: TC9163.

Today written a little gpio driver for the gumstix to drive GPIO 58/59/60 (LDD[012] on the breakout board) as DATA, CLK and ST lines for the TC9163AN. The signals look good (see top image), but can only test the real thing as soon as i have some more time (hopefully the day after tomorrow).

Anyway, the driver can be found here.

Problem 3, cut off of left signal, is fixed.

The soldering of the tone board connecter was very bad (see top picture). I wonder how it ever could have worked before??? Resoldering fixed the problem and left and right are both present again.

Now back to the input selection...

Today i made quite some progress with debugging my luxman A-383 amp.

At this moment there seem to be 3 main problems:

1. Two fuses blow as soon as record selection circuitry is attached
   
2. The input / record selection is f*cked up
3. Somewhere the left channel is chopped off

The input and record selection is build up around two analog switch arrays (TC9163AN) which must have had quite a blast (did i already mention the amp broke because of a _very_ faulty input device?). Replacing both chips solved problem 1. Still the input / record select remains faulty. The right channel is omni-present (comes thru regardles of which input connected to or which input selected) while the left channel does not come thru at all.

The TC9163AN uses a 14 bit pattern for switch selection. This pattern is clocked into its shift register. There are 6 data lines going to the input board: DATA, CLK, STIN, STRC, RCT1 and RCT2 (these lines come from a obscurely large Luxman chip on the front panel). At this moment only the DATA, CLK and STIN lines are important (STIN and STRC work as a kind of chip select). Using the logic analyzer shows that the DATA line is doing what it is supposed to, but... there are _no_ clock pulses! Those pulses are needed to shift the data bits into the register. This could explain why there is no reaction when pressing the buttons. Unfortunately the big luxman chip is as a big black box and is not described in the service manual. Tomorrow i'll write a little gpio driver and use a gumstix to generate appropriate clock and data signals. If we can get the right behavior we nailed down the problem. Only thing left is a solution ;)

Until then i bypass the input by directly wiring the CD input to find out where problem 3 occurs.

After some effort i succeeded to access the tone board (a lot of screws and cable ties), the second step after the input board). Using a little pd program (right laptop in top image) to generate two test signals (10KHz and 1KHz semi square waves for right and left respectively) i can now trace down the signals using the 2 channel scope (can't do without this bitscope thingy already :) ). As it is already quite late i only checked to see if it survives the tone board, which it does (left laptop in top image).

So far for now :)

Long time no post.
Last time i've been really busy with passepartout, a long term research project at V2_. The project is eating all my spare time and blogging is low on priority list ;)

There is one interesting development however; my amplifier broke :)

Well, it is actually very sad, it is a 1993 Luxman A-383. Quite a beast but perfect sound. Thanks to Sven Eiman i now have the service manual with all electric diagrams and part lists.

Also today my pc based logic analyzer and oscilloscoop, a BitScope BS50U pocket version, came in. I'm quite happy now, have wanted such a thing since i was 12 or so :). It is a nice piece of equipment for hobby purposes and for persons like me who don't have that much space to place a 60cm deep analog scope (and not the least important; don't have the big bucks) ;) The best thing is that it is a complete open hardware design, and the software also runs on linux which is rare with these kind of tools.

This weekend i start repairing and hopefully with success....