Wires connected to goldstar and yamaha chips.

[carlywhoo]

I took a spare STe apart that I was given, ages ago.
I noticed that there's two wires coming from Goldstar chip.
One goes to a Yamaha chip which I'm guessing is something to do with sound.
The other lead goes another chip, which has pins bent upwards on it.

I can't remember if the M/B works. Just wondered if any one knows why would someone wire up the chips like that.

[Arne]

Someone else on a German forum with a 1040STE reported about the same/a similar patch. The Yamaha is responsible for doing sound, the parallel printer port (in conjunction with MFP IIRC) and for generating SideSelect and DriveSelect signals for both floppy drives.
The "Goldstar" chip is just a 74LS164 shift register - pretty COTS.
So it seems to be an original patch done by Atari I guess.

[carlywhoo]

I don't have this patch in my other STe's though.
And what does the Goldstar chip do exactly (in laymans terms).

[1024MAK]

I don't have this patch in my other STe's though.
And what does the Goldstar chip do exactly (in laymans terms).

This may help a bit... http://www.interfacebus.com/ic-4-bit-sh ... -chip.html (have a read through, but note that the "8-Bit Shift Register IC" is the bit about the Goldstar 74LS164 shift register).
Its exact function very much depends on how the circuit is wired...

[Stefan jL]

I don't know why the mod is done but i could guess it is Atari that missed some traces on the PCB and had to fix them after manufacturing

My Atari STE does also have those wires and also a wire on the LMC1992 connecting it with the quad op amp 5v-pin.
Here is a pic of my Atari STE PCB: http://www.ym2149.com/atariste_pcb.jpg

[Cyprian]

first wire is connected to PSG clock input pin "Clock <- 2MHz"
second one is connected to MFP TAI - Timer A clock input.

...My Atari STE does also have those wires...

mine STE too

[carlywhoo]

first wire is connected to PSG clock input pin "Clock <- 2MHz"
second one is connected to MFP TAI - Timer A clock input.

But why?

[Dal]

likely reason is that those tracks are missing on that particular revision of the board. The decision to patch the boards may have been for a number of reasons for example:
1. Supply of boards were too low for demand, so rather than take the board back through a design and testing phase, they added an additional step in the production phase.
2. They may have spotted this problem during production. There are already board manufactured so a small patch is better than trashing a bunch of boards and introducing more delays by going back to design.
3. It's entirely possible that the changes they made to that rev of board couldn't allow for those traces. So they deliberately 'hacked' it with some wires.
4. They used a slightly different shift register to the one they designed the board for meaning that differences in the pinouts had to be accounted for late in the day.

If you compare that board against your other STE, check the rev numbers and also chip used in U211 for differences. This would give the best indication.

[Cyprian]

below you can find serial numbers for STE with patch and w/o patch, and on the image we can see a pattern - STE with two chips below SIMM are patched, and in case of three chips patch is not applied

Without Path:
3392 1991 CA401177
1690 1989 CA4003290
2690 1989 CA4003290

Pathed:
3290 1989 CA4003290
4489 1989 CA4003290

[dml]

My STE board also has this mod. The board is marked: GX-211 / VO 4089 / REV ____ / 1989 / CA4003290

I guess one way to find out what it's for, is to remove it and test a bunch of audio, floppy and HD access etc. to see what breaks


The 74x164 (8-bit, serial-in, parallel-out shift register) chip is connected between the CLOCK pin on the sound chip and the TR (DMA Transmit Ready) pin on the MFP (multifunction peripheral) chip - and one connection to the chip it's sitting on top of (U211). I didn't check the 74164 pinout, but given that this 'mod' has so few connections, it's probably not being used as a shift register. It's being used as a clock divider, and stealing a clock then dividing to a few kHz it to feed something else.

The original STE schematic shows the TR pin on the MFP to be unconnected, and I don't remember the MFP having any DMA capabilities in the ST line.... So it seems reasonable that the 'fault', whatever it is, is either the clock feed to the the YM2149 or something to do with U211.

[dml]

...actually I should have realized that the shift register can't be a divider, but rather an artificial delay - there's something on this page about it (in French):

http://hxc2001.free.fr/yaquoidedans/ata ... index.html

...purpose is still a bit of a mystery though.

[Stefan jL]

I have heard some STE's had a bad DMA chip... maybe this i s a kind of fix for that?

[ralcool]

No... not nothing to do the bad DMA issues..... @dml has the idea. Its a botch job.

Detailed comparison between the STFM schematics and STE shows an unusual wiring alteration to TA1 by instead being connected back to the IO0 output of the same MFP (& incidently the #Busy signal of the parallel port), to partly now controlled by 'XSINT' from the new GSTMCU.... In addition to the Mono detect signal. The French documents suggest a small timing extension. But the STE has many unanswered questions.....

From my examination, it is to do with Timer A as previously explained... but Atari was controlling it differently now with the GSTMCU- that didn't quite meet spec with the old motorola MFP.
Seemingly it had to do with monochrome (or colour... regardless that signal is used).. so perhaps a minor timing bug somewhere in the GSTMCU & Timer A in different graphics modes.???

They delayed (or extended) it through a incorrectly wired gate- that had to be patched post production to sell it. Rushed engineers I guess. The chip has no other purpose.

ST

Edit I found this in the hardware register....

########User Assigned Interrupt Vectors ######

-------+----------------------------------------------------------------+-----

$000100|ST-MFP-0 - Centronics busy |SD

$000104|ST-MFP-1 - RS-232 DCD |SD

$000108|ST-MFP-2 - RS-232 CTS |SD

$00010C|ST-MFP-3 - Blitter done |SD

$000110|ST-MFP-4 - Timer D (USART timer) |SD

$000114|ST-MFP-5 - Timer C (200hz Clock) |SD

$000118|ST-MFP-6 - Keyboard/MIDI (ACIA) |SD

$00011C|ST-MFP-7 - FDC/HDC |SD

$000120|ST-MFP-8 - Timer B (HBL) |SD

$000124|ST-MFP-9 - Send Error |SD

$000128|ST-MFP-10 - Send buffer empty |SD

$00012C|ST-MFP-11 - Receive error |SD

$000130|ST-MFP-12 - Receive buffer full |SD

$000134|ST-MFP-13 - Timer A (STe sound) |SD

$000138|ST-MFP-14 - RS-232 Ring detect |SD

$00013C|ST-MFP-15 - GPI7 - Monochrome Detect |SD

$000140|TT-MFP-0 - GPI 0 |SD

$000144|TT-MFP-1 - GPI 1 |SD

$000148|TT-MFP-2 - SCC-DMA Controller |SD

$00014C|TT-MFP-3 - Ring Indicator SCC B |SD

$000150|TT-MFP-4 - Timer D |SD

$000154|TT-MFP-5 - Timer C |SD

$000158|TT-MFP-6 - (Reserved) GPI 4 |SD

$00015C|TT-MFP-7 - SCSI DMA Controller |SD

$000160|TT-MFP-8 - Timer B |SD

$000164|TT-MFP-9 - Send Error |SD

$000168|TT-MFP-10 - Send buffer empty |SD

$00016C|TT-MFP-11 - Receive error |SD

$000170|TT-MFP-12 - Receive buffer full |SD

$000174|TT-MFP-13 - Timer A |SD

$000176|TT-MFP-14 - TT Clock (MC146818A) |SD

$00017C|TT-MFP-15 - TT-SCSI Drive Controller NCR 5380 |SD

$000180|SCC Interrupt |SD

$0001BC|SCC Interrupt |SD

$0001C0|User Defined, Unused |SD

: | : : : | :

$0003FC|User Defined, Unused |SD

-------+----------------------------------------------------------------+-----

Might explain why its been redirected and driven by the GSTMCU...

[1024MAK]

Jumping in without any research, (I.E. a wild guess ), could it have anything to do with the switching of the 32MHz clock when using the Genlock? (linky)

Mark

[ralcool]

You're right.. The Mono detect & Previously GP0 output is exchanged for 'Genlock input', they are fed to a 74S257 chip, clocked with 32Mhz.

Generating a new 'MonoMon' signal sent almost to IO7 (like a normal ST mono detect pin.. but different, also altered with interupts from XSint).. And still with added logic for TA1 problems.

Assuming I'm right.

Far too late.. I'm off to bed.

ST.