The mysterious AMD chip

A project log for Z80 Reverse-Engineering And Hacking Adventures

What else can I say?

Eric HertzEric Hertz 06/22/2022 at 05:090 Comments

Update at the bottom...


Well, shoot... I thought I had a dedicated log for this. And Right Now I can't do linking or copy-pasting, lest I lose my train of thought.

So this'll be short, then maybe reorganized, later.

Several pins are N/C, the 'scope shows various non-TTL waveforms on them, less than 1V. I'm guessing they're just floating.

 But, pin 5 looks a bit like a sawtooth wave (or more like a capacitor charging slower than it's discharged).

It's between about 1V and 4V, and at roughly 5MHz. Since it's right next to the Z80-CPU-Clock input to the AMD (pin 6), my guess is that it's for the case where the AMD chip might need to drive a crystal of its own... 

So, I dunno, I guess that spoke to me of a microcontroller, but I suppose many devices might work asynchronously to a bus (e.g. a video card's pixel clock). So, who knows.

As @ziggurat29 pointed-out, it seems to be flooded with data/register-configs(?) with no handshaking... It's expected to accept and respond to an IORQ at whatever timing the CPU gives it. That'd be hard for a microcontroller.  So, I guess we're back to some sort of dedicated peripheral chip, or programmable logic. (Hmm, what about a FIFO memory? Seems silly, most its pins tied together).


But here's a weird one... Pin 7 seems to be outputting a square wave at about 308KHz 50% duty-cycle. 

(BTW, likely coincidence, but this is the same frequency put into the DART when set to 9600baud. Yes, I did try a different baud setting, this was unchanged).

Looks like the CPU clock frequency divided by 16... 

I was unable to see any signal on pins 40 or 35/37... I'm not sure if that's because it wasn't configuerd to be outputting anything there, or if it's very brief pulses.

BTW, this thing gets hot if you leave your finger on it long... At first I thought the probing of floating pins was causing the heat-up, maybe putting them in weird inbetween-states, but now I think maybe it was just that my finger was blocking its regular cooling.

OH, big one: The 308KHz signal seems to start *immediately* on power-up. Plausibly before the processor even configures it? It continues during reset. I should've thought to hold reset *during* power-up (TODO?).

TODO: Merge other/past notes about it here, or at least link them...


Update 9/28/22:

I think I found it...

AM9513, basically a beefy timer/counter chip.

Compared the pinout with my earlier measurements/descriptions, here and I think we have a match.

Kinda boring, after all that build-up... but not unexpected. I think I even came to the "counter" conclusion a while-back.

Spose it'd be telling if I actually looked at where it goes, in the circuit. But my guess is it has to do with precisely timing a "shutter" for calibrated light intensity measurements.

Not sure why they couldn't've used another Z80CTC, Maybe it doesn't do one-shot pulses.

And there's also that weird mixture of Z80 PIOs as well as Intel 8255's... suggesting... I guess no particular partiality to the z80 peripheral architecture.

It also has a "Time Of Day" counter (hours, minutes, seconds, fractions, in BCD) that I suppose could be used as a bit of an RTC. Though I gather it's not used like that in the original firmware, as I hadn't seen anything clock-like, and don't recall the backup battery being connected to it. (And, I guess, would be a bit difficult to use that way without CPU intervention at least once a day, unless maybe it can be daisy-chained with another counter, internally... hmm...)

Actually, though, that makes me wonder... I had noticed a distinct whine even when the power switch is off... thought maybe it was redundant for the backup battery, but maybe it also keeps the "RTC" powered and sync'd via 60Hz. Hmm...


Next day...

Nah, the battery backup is only for the SRAM... And I'm rather surprised how simply it's implemented. A 3.6V NiCd, diode, and 100ohm resistor. Guess that's all yah need.

Be easy-enough to bulk ig up or do another for the "RTC" idea.

Oh, turns out, too, there is no circuitry before the power switch except the EMI filter... which surely isn't designed for a cheap inverter. So I think the whine can't be from anywhere else.

No always-on low-power for keeping the battery charged. No always-on 60Hz to clock the "RTC". But, again, an easy-enough mod to consider, I suppose.


Kinda funny how I found this; I was thinking of an FPU as another someday-maybe... came across intel's 8bit offerings, which, as I understand, were licensed from AMD, the AM9511 and AM9512... 9 and 5 were the missing numbers on this chip "AM__13." (Why On Earth have we gone so far backwards in search technology as to no longer have wildcards?!). Heh, kinda weird the next in the series of two FPUs is a friggin timer/counter.