Reverse Engineering the Macintosh SE PCB & Custom Chips for 1:1 reproduction

[Kai Robinson]

Well it's a success, reads consistently as a 27C020 - no matter the manufacturer - although the AM27C020 seems to work best. 

So - here's the final code revision for the GLU logic - anyone able to verify this against the GLU's own description in Apple's own hardware manual?

 

Name     glu ;
PartNo   00 ;
Date     07/11/2020 ;
Revision 01 ;
Designer Kai Robinson ;
Company  Atmel ;
Assembly None ;
Location  ;
Device   virtual ;

/* Dedicated input pins */

pin 1   = CLK;       /* Input */
pin 2   = PB6;       /* Input */
pin 3   = IRQ;       /* Input */
pin 4   = WRDATA;    /* Input */
pin 5   = ENABLE1;    /* Input */
pin 6   = PA4;       /* Input */
pin 7   = 16MCLK;   /* Input */
pin 8   = PA3;       /* Input */
pin 9   = RTXCB;    /* Input */
pin 11  = OE;       /* Input */

/* Programmable output pins */

pin 12  = RTXCA;    /* Combinatorial output */
pin 13  = OUTA;      /* Fixed high output */
pin 14  = BBUIRQ;   /* Combinatorial output w/ output enable */
pin 15  = ENABLEL;  /* Combinatorial output */
pin 16  = ENABLEU;  /* Combinatorial output */
pin 17  = FLOPPYWR; /* Combinatorial output */
pin 18  = nc;        /* Fixed high output */
pin 19  = FCLK; /* Combinatorial output w/ output enable */

/* Output equations */

!FCLK     = 16MCLK   & !OE;
 nc       = 'b'1;
!FLOPPYWR = WRDATA;
!ENABLEU  = !ENABLE1 & !PA4;
!ENABLEL  = !ENABLE1 &  PA4;
!BBUIRQ   = !PB6     &  IRQ;
 OUTA     = 'b'1;
!RTXCA    = !PA3    & !RTXCB   & !OUTA 
          #  PA3    & !RTXCB   & !OUTA 
          # !PA3    &  RTXCB   & !OUTA 
          #  PA3    & !RTXCB   &  OUTA;

/* End */

 

[quorten]

`OUTA` and `RTXCA` look suspect, `RTXCA` will always be one with that logic so apparently you cannot enable synchronous modem use?  Not like you could ever do it in earlier Macintosh computers... I'm guessing the hardware manual's passing mention of this is because the feature was cancelled late in the product development.

 

Otherwise, I'd say everything else looks good when also consulting the VIA register layouts in the hardware manual.

 

Also, I've looked a little into implementing a BBU clone and started coding up some Verilog.  I'd say implementing  the embedded asynchronous DRAM controller feels like it's the most complicated part, that being said most of the rest of the BBU functions are simple combinatorial logic or counting loops for things like scanning the video buffer, sound buffer, etc.

 

Mentioning the BBU, one thing I don't quite understand is how the Macintosh SE controls the boot-time memory overlay.  Earlier Macintoshes used a VIA register bit to control the switch, but there is no such bit reserved in the Macintosh SE's VIAs.

 

Also, perusing emulator source code seems to have reached the limit of its helpfulness at this point, looks like a lot of how Mini vMac, MAME (formerly MESS), and PCE implement the finer details of Macintosh SE BBU emulation is through borrowing an emulation of a Macintosh Plus and patching ROM hacks in place for SCSI support.

[360alaska]

Look,

 

I found N.O.S Maxell, looks like it might work my IIFX or a newly created SE logic board:

 

https://www.diytrade.com/china/pd/15028592/Maxell_3_6V_1_2_AA_Lithium_thionyl_chloride_battery_ER3.html

Edited November 9, 2020 by 360alaska

[quorten]

Well if those Maxells aren't already exploded from sitting in inventory they must have something good about them!  I have to admit, it's tempting to see leaded solder-in batteries being sold... yeah you could just solder in another battery for a more authentic looking board, but it's totally punitive to not use a battery holder or connector of some sort.

[LaPorta]

The batteries in and of themselves do not seem to explode, it's that they explode when depleted.

[quorten]

7 minutes ago, LaPorta said:

The batteries in and of themselves do not seem to explode, it's that they explode when depleted.

If the batteries are really as defective as we may believe, they would have self-discharged to 10% or less of original capacity in 35 years.

[Phipli]

The original SE soldered axial batteries often do well compared to batteries in other Macs. I checked one and it was almost 3v after 31 years.

 

Why use a NOS battery when you can buy a New battery? This is the same brand as the original.

 

https://www.ebay.co.uk/itm/262860459865

 

Or get a socket and put in a removable one.

 

Question : the Soldered SE batteries are labeled 3v. Can you fit a 3.6v without causing harm? With my SE I fitted a socket but went to the trouble of buying a 3v version.

 

https://www.ebay.co.uk/itm/280394202604

[cheesestraws]

I think the last round of components for this have arrived now!  IC sockets and some more capacitors.  So I'm going to try to get the other half of the soldering done soon...

[Kai Robinson]

@cheesestraws Pics pls

[cheesestraws]

I will post some my workbench is currently full of floppy drives and oscilloscope probes (urgh) but when I've finished measuring exactly how depressingly broken this thing is then I'll get onto it...

[quorten]

9 hours ago, Phipli said:

Question : the Soldered SE batteries are labeled 3v. Can you fit a 3.6v without causing harm? With my SE I fitted a socket but went to the trouble of buying a 3v version.

Yes, when powered on, the RTC circuit is actually fed with 5V power, 3.6V will work perfectly fine.

 

Also, nice that you've found the Lithium Manganese Dioxide cells!  Now I've looked at the photos in detail of the original VARTA battery I removed, indeed I see the original just says "3V".  If indeed the original was also Lithium Manganese Dioxide chemistry, that's pretty nice!  My particular copy dropped down to 0.98V in 2019, though, so this is definitely a "your mileage may vary" thing.  Lithium Thionyl Chloride batteries are at 3.6V, those are the ones that are used because they are touted as having a very low self-discharge... but when they do discharge completely, they don't merely leak, they explode.

 

UPDATE: The original battery is labeled with "ER 1/2 AA Order Number 6126", which indicates Lithium Thionyl Chloride chemistry, looks like "3V" was merely an abbreviation of the true nominal voltage of 3.6V.  This is the closest product number datasheet I could find.

 

http://products.varta-microbattery.com/applications/mb_data/documents/data_sheets/DS7126.pdf

 

My particular copy was lot number 0387... appears to be a code for date of manufacture March 1987.

Edited November 10, 2020 by quorten
Original VARTA really was Lithium Thionyl Chloride

[Phipli]

2 hours ago, quorten said:

Also, nice that you've found the Lithium Manganese Dioxide cells!  Now I've looked at the photos in detail of the original VARTA battery I removed, indeed I see the original just says "3V".  If indeed the original was also Lithium Manganese Dioxide chemistry, that's pretty nice!

...

UPDATE: The original battery is labeled with "ER 1/2 AA Order Number 6126", which indicates Lithium Thionyl Chloride chemistry, looks like "3V" was merely an abbreviation of the true nominal voltage of 3.6V.  This is the closest product number datasheet I could find.

 

http://products.varta-microbattery.com/applications/mb_data/documents/data_sheets/DS7126.pdf

 

My particular copy was lot number 0387... appears to be a code for date of manufacture March 1987.

Still have the original, let me know if you are interested in more photos - under "Ord. No. 6126" it says "MH 13654 (N)" - I think that means it is Manganese Dioxide? Mine is date code 0188.

 

http://www.mercateo.co.uk/p/4888-6681/Varta_CR1_2AA_Lithium_battery_6127_UL_MH_13654_N_950mAh.html

[quorten]

That's the UL registration number, apparently the same registration number can be used by a number of different products?  Your battery also uses the "ER" chemistry code so it's also Lithium Thionyl Chloride chemistry.  Interestingly, my copy does not have a UL registration number on it.

[Phipli]

@quorten

 

Interesting. So I'm probably better off dropping a 3.6v in it. I thought I'd gone the extra mile to keep it original (ignoring the socket I fitted).

[Kai Robinson]

@quorten I think i might have found the RTC that's used: https://www.silicon-ark.co.uk/datasheets/pcf8583-datasheet-nxp.pdf

 

The pinout needs re-jigging, but it's used in a lot of other period correct machines (Acorn A3000 series), would it be worth getting a few? A simple PCB adapter is doable. 

[davidg5678]

8 minutes ago, Kai Robinson said:

would it be worth getting a few?

I think so! I think it would be nice to be able to build these new boards with as many or few original parts as possible.

[Kai Robinson]

Ordered 10 to start with. 

[cheesestraws]

14 minutes ago, Kai Robinson said:

The pinout needs re-jigging, but it's used in a lot of other period correct machines (Acorn A3000 series), would it be worth getting a few? A simple PCB adapter is doable. 

 

They seem to be pretty reliable long-haul in the Acorns - at least, I've only had to replace one of them, and I think that was probably invisible battery leak damage rather than anything else.  So probably a good option here, too...

[Kai Robinson]

How it's connected in the Acorn...

[quorten]

@Kai Robinson The NXP chip definitely has a different software interface than Apple's custom silicon.  I'd say the chips definitely look handy for use in other projects (like a Raspberry Pi RTC), but ROM patches would be required to use it for Macintosh computers.

 

One good note, looks like the Acorn schematic is wired up expecting the same load capacitance from the crystal, but the Apple way of using a 10pF and 33pF capacitor together is probably cheaper than using a single 12pF capacitor due to specialized component costs.

[cheesestraws]

Can someone sanity-check me here please?  L1 on the BoM is marked as 33 µH, but on the board I am taking bits off, it seems to be 21 µF (apologies for poor colour reproduction)

 

 

[Phipli]

21 minutes ago, cheesestraws said:

Can someone sanity-check me here please?  L1 on the BoM is marked as 33 µH, but on the board I am taking bits off, it seems to be 21 µF (apologies for poor colour reproduction)

 

 

 

[cheesestraws]

Thankyou!

[Phipli]

3 minutes ago, cheesestraws said:

Thankyou!

Although in my photo it looks more like 27 or 28? Doesn't look brown...

 

Edit : looking again it could be. Sigh, you'd think it would be easier to tell!

Edited November 12, 2020 by Phipli

[Phipli]

I zoomed on the original photo and it is brown 21 it is!