• Please review our updated Terms and Rules here

MM8I Replacement

vrs42

vrs42

Veteran Member
Joined
Nov 24, 2009
Messages
1,010
Location
Beaverton, Oregon
Related to a discussion start in the EAE thread, I have ordered prototypes of a proposed MM8I replacement:
NVRAM.png
The idea is that these cards snap apart and insert into the slots that you would use to cable up a real MM8I. These could be in your PDP-8/I, your PDP-12, or in an existing MM8I.

The leftmost card has a jumper block that allows you to enable specific fields, up to 32K worth. (The first two fields should be in your base machine, so they probably shouldn't be enabled.)

Hopefully the project can be got working without too much trouble, and will be of interest.

Vince

I uploaded a PDF schematic, but I'm not sure if that will get included.
 

Attachments

  • 20230322.pdf
    165.9 KB · Views: 13
m_thompson said:
Which one of the DS1230 parts are you planning to use? They vary from $1.07 each to $31.75 each.
Click to expand...
They are all basically non-volatile 62256 equivalents in wide-dip packages. I was just looking at them on Digikey, and I see that no matter what, they want $200-$300 for the 10 I wanted.

I thought that was really spendy. I plan to use my on-hand FRAM parts for the debug. If and when it's working, I can order a giant pile of the DS parts at $1.07 each. (As it is, I don't have the delay lines, so I had to buy $300 worth of those.)

Really any 62256 compatible chip should be fine. (I'm still avoiding Aliiance brand though, after the Omnibus memory experience.) I'm not even sure the non-volatile feature is important unless your system handler is two pages.

I didn't put the coin cell battery back-up on the PCB, though I could have.

Vince
 
Interesting. I'm told by the Digikey representative that my order for the delay lines has been declined by the supplier. I'm supposed to call back tomorrow during business hours for more clarification.

Vince
 
That means they don't have any in inventory and don't want to make a small run. They would probably make 10000 of them for you. It will be interesting to hear what they tell you.
 
I was pretty sure I had checked both "In stock" and "Normally stocked", and even now it says "2,618 in stock". It does ship from another warehouse, so perhaps one hand isn't fully in touch with the other.

Vince
 
Likely those are parts then that are supplied by Rochester Electronics, which is a supplier of obsolete parts. DigiKey uses them as a vendor for out of production by the original vendor parts.
 
Well, they claim to have stock too:
www.rocelec.com

Part DS1000M-250

Rochester Electronics is the world's most trusted solution for end of life semiconductors. Search our large inventory of semiconductors and buy now.
www.rocelec.com www.rocelec.com

In Stock:2,737
There is a minimum quantity of 102 and a maximum quantity of 2,737 for this product.
100 – 499 $2.46

I ordered some on eBay at $0.50 each. They are gull wing, but they should do for now.
 
vrs42 said:
They are all basically non-volatile 62256 equivalents in wide-dip packages. I was just looking at them on Digikey, and I see that no matter what, they want $200-$300 for the 10 I wanted.
Click to expand...
This is a characteristic of “marketplace” products. Marketplace items are always $300 minimum. I always check the “exclude marketplace” box in Digikey search.
vrs42 said:
Really any 62256 compatible chip should be fine.
Click to expand...
Jameco has some available, both slim and wide.

vrs42 said:
(I'm still avoiding Aliiance brand though, after the Omnibus memory experience.)
Click to expand...
I recall reading about this issue in a post a while back and would be curious to investigate the issue.

vrs42 said:
I didn't put the coin cell battery back-up on the PCB, though I could have.
Click to expand...
I used to think that was important, but it seems that many of us use a bootloader, so I’m not sure that non-volatile memory is needed.
 
gwiley said:
This is a characteristic of “marketplace” products. Marketplace items are always $300 minimum. I always check the “exclude marketplace” box in Digikey search.
Click to expand...
I'll have to watch for that in the future. If I check the box to exclude them, I may not be able to find stock at all, though.
gwiley said:
Jameco has some available, both slim and wide.
Click to expand...
A while back I ordered a boatload of little adapter PCB, so that I can switch between wide and narrow 28 pin without worries.
gwiley said:
I recall reading about this issue in a post a while back and would be curious to investigate the issue.
Click to expand...
I beat on it for a while, but all I was ever able to learn was that the Alliance SRAM didn't pass the memory test (or various simplifications of it that I wrote), but most other SRAM did. It was never reliably failing just wasn't reliably storing either.
gwiley said:
I used to think that was important, but it seems that many of us use a bootloader, so I’m not sure that non-volatile memory is needed.
Click to expand...
The older machines don't really have a boot card, so it might matter there. OTOH, if your first 8K are core, you'll likely have enough OS/8 to reboot still in memory there. Unless your system handler is two pages, in which case I think you'd also need the last page of field 2 to be non-volatile.
 
Hijacking my own thread a bit...

I've been exploring an idea similar to the MM8I idea, where the three connector boards of the non-DMA Posibus interface carry an implementation of the PT08. The idea is roughly that the MAC connector gets a card conatainng roughly a variation the M707 Teletype Transmitter, the AC Input connector gets a variation of the M706 Teletype receiver, and the BMB connector gets the equivalents of the M452 Baud Rate Generator and a couple of M103 Device selectors. The equivalent of the M706/M850 EIA Interface Connector would squeeze in wherever there was the most room left over. Hopefully the inter-card communication would mostly be a few control signals and the baud rate clock.

If that existed, then that and an MM8I replacement would be sufficient to boot an 8/I to OS/8 from SerialDisk.

The PDP-12 has wiring for a second serial port already in the backplane, so probably doean't need this.

Most PDP-8/L haven't got enough memory, at least until I can get most of the logic for the BM8/L to fit in a CPLD. If it had the memory, it could benefit from this.

Doug's also working on his variation of SerialDisk which uses the console serial, so I don't know how much actual interest there is for a second serial port for Posibus.

Vince

Oh, and another issue keeps cropping up about what speed to run the serial ports at, since 115.Kbaud is closest to the top recommended speed for these machines, but is a multiple of 110 baud, not 150. So a crystal change or seldom used extra divider chains end up cluttering the baud rate generator design.
 
This MM8I-replacement sounds very cool; once it works I might try to build one for myself!

vrs42 said:
I've been exploring an idea similar to the MM8I idea, where the three connector boards of the non-DMA Posibus interface carry an implementation of the PT08. The idea is roughly that the MAC connector gets a card conatainng roughly a variation the M707 Teletype Transmitter, the AC Input connector gets a variation of the M706 Teletype receiver, and the BMB connector gets the equivalents of the M452 Baud Rate Generator and a couple of M103 Device selectors. The equivalent of the M706/M850 EIA Interface Connector would squeeze in wherever there was the most room left over. Hopefully the inter-card communication would mostly be a few control signals and the baud rate clock.

If that existed, then that and an MM8I replacement would be sufficient to boot an 8/I to OS/8 from SerialDisk.
Click to expand...
That would be similarly interesting... except my 8/I is a negibus machine, heh.
 
Nevets01 said:
That would be similarly interesting... except my 8/I is a negibus machine, heh.
Click to expand...
Well, I had a similar idea which involved paddles which did level conversion, building out the "Y" cables that were sometimes used with converted PDP-8/I. So there'd be 6 paddles at the Negibus end end and three at the Posibus end (or 11 and 5, if DMA were needed). Each Negibus paddle just needs 9 level converters, and the Posibus paddles can be just straight-through.

Then, a connector block or two to create a home for the Posibus device.
 
You must log in or register to reply here.
Back
Top