PDA

View Full Version : GW-12887-1: A DS1287/DS12887 et al. Replacement



glitch
September 6th, 2017, 04:24 AM
By request, and because I too have a few that need replaced:

http://i.imgur.com/0PM5i7Ph.jpg (https://imgur.com/0PM5i7P)

You bend up the legs on a DS12885 (the bare IC used inside the DS12887 module), glue my little board on top, and solder the leads down:

http://i.imgur.com/LUqSagYh.jpg (https://imgur.com/LUqSagY)

I had a little 486 ISA industrial SBC that needed a DS12887 replaced, so that was the test candidate:

http://i.imgur.com/ywHug1mh.jpg (https://imgur.com/ywHug1m)

Works fine! I need to tweak the castellated holes on the board a bit. There's not enough copper to hold the plated-through hole in, at least for OSH Park's castellation process. I'll release the files once it's ironed out. But, it works, and it'll be easier for hobbyists to assemble than the version that uses surface mount bits. I believe it's cheaper as well -- the DS12885 is cheaper than the surface mount version, and you don't need to acquire module pins since the DIP takes care of that.

I'll probably do a run of panelized modules and make up a bunch of these, so if people want a fully assembled, ready-to-go version, that's coming.

One of the best aspects (aside from never having to rework it again!) is that all of the parts are current production and available from e.g. Mouser. No need to have an old IC and tear it down, as with the 48T02/48T08!

eeguru
September 6th, 2017, 04:51 AM
...at least for OSH Park's castellation process.
I was unaware OSH Park officially supported castellation or any form of edge platting? Or is this an unofficial hack?


...and you don't need to acquire module pins since the DIP takes care of that.
It is worth noting Mill-Max makes some nice double tapered break-away headers that work nice for this purpose. I keep bins of them around. But I do like your solution better! :)

Very nice project and idea. Thanks!

glitch
September 6th, 2017, 05:48 AM
I was unaware OSH Park officially supported castellation or any form of edge platting? Or is this an unofficial hack?

They support it officially: http://docs.oshpark.com/tips+tricks/castellation/

It's obviously a little rougher than what you'd expect from a production panel with castellated holes, but it gets the job done for prototypes. Apparently it's somewhat sensitive to the size of the annular ring around what's remaining, I had one of the through-plating sections pop out. So, the new design includes a large rectangular pad that extends well into the board, top and bottom.


It is worth noting Mill-Max makes some nice double tapered break-away headers that work nice for this purpose. I keep bins of them around.

Yep, and Aries makes flat pin DIP headers for a reasonable price, too -- I keep both around for prototyping, I tend to use Batten & Allen pins for production modules, if there's room. The Aries DIP headers have the bonus of being stamped flat pins, much like the Batten & Allen pins. They won't ruin crappy single-wipe sockets.


But I do like your solution better! :)

Very nice project and idea. Thanks!

Thanks! I figure it minimizes cost and assembly difficulty for most hobbyists that need to replace only one or two DS12887s and want to make their own boards.

Chuck(G)
September 6th, 2017, 08:40 AM
Nice work, glitch! But would it be easier to use through-hole pins for the mouting (as mentioned)? Those legs look like a pain to set up.

The gold-plated breakaway headers are pretty cheap on ebay. I use a lot of them when going PCB-to-PCB.

glitch
September 6th, 2017, 08:46 AM
Nice work, glitch! But would it be easier to use through-hole pins for the mouting (as mentioned)? Those legs look like a pain to set up.

The gold-plated breakaway headers are pretty cheap on ebay. I use a lot of them when going PCB-to-PCB.

I just bent them up with needle nose pliers, after the module was glued on, and pressed them into the castellated half-hole in the board. Solder, then trim the end off. I'm not sure it's super clear from the picture, but the rest of the legs are just the IC's normal legs.

One could of course solder jumpers from the castellated holes to the IC legs.

Chuck(G)
September 6th, 2017, 09:25 AM
Ah, I was thinking that you were using the TSOP version of the chip. Now things are clear, thanks.

glitch
October 30th, 2017, 06:31 AM
Revision 2 prototypes are in and tested!

http://i.imgur.com/s420rJKh.jpg (https://imgur.com/s420rJK)

Note the large, rectangular pads around the castellated holes. These held the through-hole plating much better than the small annular rings. Should be easier if someone wants to do a toner transfer too!

Here's a view with the module flipped over:

http://i.imgur.com/4dmKNITh.jpg (https://imgur.com/4dmKNIT)

You can clearly see it's a DIP chip with some legs bent up from this angle. No headers or pins to line up, just glue the module on top of the DS12885, wait for the glue to cure, then bend the leads up and solder them to the castellated holes.

Trixter
October 30th, 2017, 09:42 AM
This is a great little idea! I look forward to shoving my money at you.

ibmapc
October 30th, 2017, 08:39 PM
Me TOO!! Please let us know when and how to order these!!


Greg

glitch
October 31st, 2017, 05:34 AM
I've got these panelized and added to my next PC board order, so they should be in within 2 weeks or so.

If anyone needs one right now I do have several fully functional prototype modules ready to go.

ibmapc
October 31st, 2017, 06:23 AM
How much do you want for a prototype module?

glitch
October 31st, 2017, 06:37 AM
It's about $10 in parts (the DS12885 alone is around $7), $20 + shipping assembled+tested sound OK?

ibmapc
October 31st, 2017, 06:56 AM
It's about $10 in parts (the DS12885 alone is around $7), $20 + shipping assembled+tested sound OK?

Sounds good to me! Set one prototype aside for me, and I'll send you a PayPal tonight. Will $6 cover shipping to 97321?

gslick
December 17th, 2017, 07:58 PM
Anyone know for certain if these would work as a DS1287 replacement in a VAXstation 3100 M76?

glitch
December 18th, 2017, 04:39 AM
Anyone know for certain if these would work as a DS1287 replacement in a VAXstation 3100 M76?

I can test, I have one. They do work fine in a VAXstation 4000, so I suspect it'll be OK. I also now have a GW-1287-1 which is the same board with a different part number, but meant to mount to a DS1285, for those few systems that won't work with a DS12885.

RWallmow
January 1st, 2018, 08:26 AM
Want!!! 42789

glitch
January 22nd, 2018, 09:47 AM
Anyone know for certain if these would work as a DS1287 replacement in a VAXstation 3100 M76?



KA43-A V1.2

F...E...D...C...B...A_..9...8...7...6...5...4...3_ ..2_..1?..


? C 0080 0000.4001
? 6 80A1 0000.4001
?? 1 00C0 0011.700C

>>>


Well, that yak shave is done! Ended up getting NetBSD going just so I could set the time :) Seems the M76 is different enough from the early revisions that my MV3100 disk with ULTRIX wouldn't boot. But yes, it does seem to work -- keeps time, D and E errors (NVRAM, RTC) gone after setting the time.

Great Hierophant
January 22nd, 2018, 11:50 AM
What is the purpose of the crystal next to the battery holder? Doesn't the motherboard typically provide a clock signal?

VERAULT
January 22nd, 2018, 11:54 AM
I think this is great! about time someone came up with a solution/replacement.

glitch
January 22nd, 2018, 12:25 PM
What is the purpose of the crystal next to the battery holder? Doesn't the motherboard typically provide a clock signal?

It's a 32.768 KHz clock source for the RTC. If the clock module relied on a clock from the motherboard, the clock would stop when you powered the machine off, so, it has its own local low-power oscillator.

glitch
January 22nd, 2018, 12:26 PM
I think this is great! about time someone came up with a solution/replacement.

Thanks! I like it a lot better than paying Maxim for an expensive module that a) may not work in the original circuit anyway and b) will still die in around 10 years. Much nicer to be able to just change the CR1225 cell!

VERAULT
January 22nd, 2018, 12:30 PM
Thanks! I like it a lot better than paying Maxim for an expensive module that a) may not work in the original circuit anyway and b) will still die in around 10 years. Much nicer to be able to just change the CR1225 cell!

speaking of which why did you go with the CR1225 over the CR2032?

glitch
January 23rd, 2018, 04:39 AM
speaking of which why did you go with the CR1225 over the CR2032?

It fits in the outline of the module. More important in the GW-48T02-1 and such since those are often down in a carrier (e.g. in Sun systems), but I do have some systems with devices very close to the RTC module where a CR2032 in a holder wouldn't fit. I did a lot of measuring to settle on the largest standard coin cell I could get in there.

There's also a height restriction on some systems (some of the PS/2s, many of the Sun systems). The CR1225 was the best overall solution, and I'd rather only have to buy one kind of battery holder. It's also the cell that was present in several of the different kinds of modules I disassembled for the design phase of this project, so it should provide comparable life to a factory module, discounting of course cheap cells with poor seals (I think it's Mostek that claimed a 50% increase in cell life due to the potting compound making up for poor cell seals). To avoid the problem, I just order high-quality brand new Renata cells :) They're $0.85 each in QTY 1 from Mouser.

VERAULT
January 23rd, 2018, 05:01 AM
It fits in the outline of the module. More important in the GW-48T02-1 and such since those are often down in a carrier (e.g. in Sun systems), but I do have some systems with devices very close to the RTC module where a CR2032 in a holder wouldn't fit. I did a lot of measuring to settle on the largest standard coin cell I could get in there.

There's also a height restriction on some systems (some of the PS/2s, many of the Sun systems). The CR1225 was the best overall solution, and I'd rather only have to buy one kind of battery holder. It's also the cell that was present in several of the different kinds of modules I disassembled for the design phase of this project, so it should provide comparable life to a factory module, discounting of course cheap cells with poor seals (I think it's Mostek that claimed a 50% increase in cell life due to the potting compound making up for poor cell seals). To avoid the problem, I just order high-quality brand new Renata cells :) They're $0.85 each in QTY 1 from Mouser.

Makes Sense, thanks for letting me know.