Falcon DSP pinout

[sqward]

Hi there
I've tried to find a pinout for Falcon DSP but it seems quite a challenge. You'd think that Motorola manual would do it but if you look closely you'll notice that 56001 in there has 132 pins while the one if Falcon has only 80!

Internet knows nothing about the 80 pin version, and other then the incomplete pinout which has no labels for ~14 pins (http://tho-otto.de/hypview/hypview.cgi? ... &index=476) there's nothing out there.

How is that possible? What am I missing?
Regards

[mpattonm]

Atari_Falcon030_Service_Guide.pdf
page 55
its missing power pins tho, so these have to be reverse engineered from Falcon PCB data. I may have a complete pinout, incl power, somewhere...
BTW those ~14 pins will be either NC, or power.

[sqward]

Nope. Sorry, I forgot to mention. There's discrepancy in that document which I discovered earlier: page 55 lists pins for 132 pin package which is not correct.

[mpattonm]

Oh, right, sorry. Anyway, here is the complete pinout for SSP4240220FJ33-QFP80

[sqward]

Interesting, thank you very much.

I wonder if there's official Motorola doc to confirm this? I find it hard to believe that's no mention of a cut down 56001 anywhere. By cut down verison I mean the removal of the serial port pins (RXD, TXD) in port C.

[czietz]

Welcome to the world of custom (or semi-custom) chips. For a lot of the ICs in the products that I develop at work, you won't find any public documentation, either.

[sqward]

Is it a custom chip though? For 20k units? Would it save them any money?

[mpattonm]

There is no Motorola datasheet AFAIK. One of the options to cross-reference my attachement is Falcon schematic, for example.

[czietz]

Is it a custom chip though? For 20k units? Would it save them any money?

It's probably the same silicon as the standard 56001, just in a custom package. (That's what I was trying to say with "semi-custom".) Also, we don't know what sales figures Atari promised to Motorola.

[sqward]

There is no Motorola datasheet AFAIK. One of the options to cross-reference my attachement is Falcon schematic, for example.

I tried that. Your document makes sense but I was hoping it wasn't correct Falcon schematic doesn't reference VCC and GDN pins as far as I can see so cross-referencing is not fully possible.

[czietz]

If you were really, really desperate, you could trace the signals using the Falcon PCB layout: https://www.chzsoft.de/storage/falcon-rev4-board.pdf. However, it's absolutely no fun following a trace through vias and multiple layers.

But, considering that mpattonm has spent considerable time making a better Falcon PCB (Beginning of very long journey), I'd be quite certain that his pinout is correct.

[sqward]

Maybe I'll just explain what I was looking to achieve: regular 56001 has a async serial interface (SCI) which woudn't be connected in Falcon if only regular 56001 was used. I was hoping to run that serial port with a pair of wires another unused async serial port in MFP. This would help me with debugging my DSP software.

The diagram from mpattonm would suggest that SCI is cut out of the part that is used in Falcon. However, Falcon schematic says nothing much about the location of VCC and GND pins. So I kind of hope that TXD and RXD pins are still there undiscovered

[krupkaj]

please remove the post, I found you already mention what I though.

[mpattonm]

However, Falcon schematic says nothing much about the location of VCC and GND pins. So I kind of hope that TXD and RXD pins are still there undiscovered

The extract I have sent you includes power pins mapping. If you just take a look at it and count all the pins, you will realize not a single pin is left unconnected. So where do you think the ASP port pins would be hiding? :O

[sqward]

By mistakenly being labeled VCC or GND

[mpattonm]

Well they are all _physically_ connected on VCC or GND and the track width of each of their connection on Falcon PCB confirms they are indeed power pins, not just a (RX/TX) data lines tied say to ground. That would not make sence anyway.

[sqward]

Thanks. My hopes are gone now

[czietz]

OK, then use the original PCB layout I posted above, look up all pins that are labeled VCC or GND in mpattonm's schematic symbol and check whether any of them is not connected to the respective power planes in the PCB. Or use a continuity tester on the actual HW.
EDIT: mpattonm already confirmed that.

Another thing that makes "hidden" pins unlikely: If we assume it's the standard silicon in a custom package, the order of pins will be the same. (You want to avoid bond wires criss-crossing.) Therefore, any potentially hidden serial port pins would need to be between SC0 and H0. Those are pins 54 and 55, respectively, in the 80 pin package. Nothing in-between.

[Eero Tamminen]

Maybe I'll just explain what I was looking to achieve: regular 56001 has a async serial interface (SCI) which woudn't be connected in Falcon if only regular 56001 was used. I was hoping to run that serial port with a pair of wires another unused async serial port in MFP. This would help me with debugging my DSP software.

Have you tried Hatari DSP profiler? It has even support for profiling spinloops to find out where one could remove waits between CPU<->DSP communication (which I added when Douglas was optimizing those for Bad Mood).

[sqward]

Hatari's DSP emulation in general is in quite poor state for my needs. My last demo doesn't work for several reasons (https://www.pouet.net/prod.php?which=78884):
* 030 emulation timings are off
* SDMA emulation is sketchy (handshake mode seems to only work with Norig mp2 player?)
* No DSP timer emulation
* No host port interrupts on DSP side

I Hatari can get there though.

[tyk]

Is DSP port pinout compatible with Nextstation??

[Rustynutt]

Welcome to the world of custom (or semi-custom) chips. For a lot of the ICs in the products that I develop at work, you won't find any public documentation, either.

Hehe. Had this for years.
Looks darn close to a 68000.

[Rustynutt]

Hatari's DSP emulation in general is in quite poor state for my needs. My last demo doesn't work for several reasons (https://www.pouet.net/prod.php?which=78884):
* 030 emulation timings are off
* SDMA emulation is sketchy (handshake mode seems to only work with Norig mp2 player?)
* No DSP timer emulation
* No host port interrupts on DSP side

I Hatari can get there though.

Back burner project, I too had DSP hardware talks with Doug back in the Afterburner/Nemesis daze.

We discussed having to make a pin out adapter due to Atari's implementation of the 56000 going over to the 56004 with a 50mhz rating.

The idea of desoldering-resoldering the DSP/adapter drove the upgrade idea over to the expansion bus, with the benefit of using a DSP with much higher rated frequency w/ backwards compatibility.

Unfortunately, never layed out a PCB for an expansion pin DSP interface before he disappeared again.
We were talking along the lines of redirecting I/O via GALS, and using the 56000 as a phraser to interface legacy hardware.

I'm sure others could code something like that, but that man is/was pretty sharp

Someday, will toy around with this developer board, and document benchmarks to see if it's even worthwhile. A 100mhz DSP on the expansion bus of the CT cards might stirr coders dreams.