PDA

View Full Version : Do all 2114 SRAMs have the same pinout?



GiGaBiTe
June 22nd, 2017, 03:15 AM
Long story short, I have an original centipede (copyright 1982 on the ROMs) arcade board that a client wanted me to repair. The entire cabinet inside and out has been neglected for probably well over a decade and has led to several of the 74xx series logic chips to get corroded and sad, as well as the main memory, which comprises of two TMS2114-45NL SRAMs. I've been able to source replacement chips for all of the 74xx series logic chips and everything else I need, except the 2114 SRAMs.

I'm not able to find an exact replacement for those chips, but I'm able to find other manufacturers of 2114s, but no datasheets for them. My question is, are 2114s pin compatible between manufacturers? The existing chips are 450ns, so I'm guessing the same or faster parts will work if the pinouts are the same.

The problem with the board is that it won't output a video signal. It does boot and run in attract mode, you can hear the sound playing. It does eventually crash though, which I think has something to do with the corroded chips and duff capacitors, 35 years is pretty long in the tooth for them.

glitch
June 22nd, 2017, 05:35 AM
Yes, they all have the same pinout, but they also can have different properties -- speed (as you said, same or faster), and current consumption (I don't know if the L suffix on the TI part makes it low power). Current consumption is most important when the RAM is being used as battery-backed memory, like for saving high scores.

I'd imagine 2114s are plentiful from the usual suppliers, but if you have a hard time finding them, I do have some spares.

It seems that the TI logic from that era is more prone to having the legs rot off than normal. The black oxide found on many of them is OK -- it's silver oxide, from the silver plating on the legs, and actually has near the same contact resistance as the silver plating (this is a big reason why it's used in contacts, especially high current contacts).

GiGaBiTe
June 22nd, 2017, 06:34 PM
According to the only datasheet I could find on the TMS2114-45NL, the "L" means that it draws 330 mW vs a standard 550 mW for a normal TMS2114 chip.

So basically I need to find a 2114 that:

- Uses a 1kx4 organization (4096 bits.)
- Is 450ns or faster.
- Has a current draw of ~330 mW or less.

The legs on the corroded ICs do have a black oxide on them, but they also have something that looks like brown rust and one or two chips have an additional white fluff on them. There are no batteries on the PCB or anywhere in the cabinet as far as I could see, so I have no clue what that is. White fluff to me usually means magnesium oxide or heavily corroded aluminum.

If you have any 2114s that fit the bill and don't mind parting with a couple of them, I'd rather buy them here than fleabay or China IC mart.

gslick
June 22nd, 2017, 06:45 PM
Unicorn Electronics has 2114L for around $2.50 each. Only issue there is their $25 minimum order.

http://www.unicornelectronics.com/IC/STATIC.html

GiGaBiTe
June 22nd, 2017, 07:21 PM
I don't see myself getting into the arcade repair business, so I don't need that many :p

Thanks for the link though. The centipede board only has two 2114s next to the 6502 CPU.

glitch
June 23rd, 2017, 04:52 AM
I'm sure I can find a pair that fit the bill, they'll probably be faster/lower power, but that doesn't hurt things! I'll send you a message.

Dwight Elvey
June 23rd, 2017, 08:05 PM
Do realize that these chips are NOS. They were made when a clean
room was vacuumed once a week ( well that is an exaggeration ).
Still, a lot of these part may be bad. I once did some testing for
a local outfit and found 20% of there old 2114s were bad.
Do realize this as a possible.
The places that sell these old parts rarely have any method to test
them. The best they can do is replace any that are defective.
Dwight

gekaufman
June 23rd, 2017, 08:14 PM
Although I've had some failures, my experiences with 2114's hasn't been that horrid, perhaps 2-3% bad in the last batch I tested.

2102's and 4116's have caused me much more grief.

In any case, NeoLoch makes a very nicely done tester for most common SRAM types (including 2114).

http://neoloch.com/products/Inquisitor.php

- Gary

GiGaBiTe
June 24th, 2017, 06:59 AM
Do realize that these chips are NOS. They were made when a clean
room was vacuumed once a week ( well that is an exaggeration ).
Still, a lot of these part may be bad. I once did some testing for
a local outfit and found 20% of there old 2114s were bad.
Do realize this as a possible.
The places that sell these old parts rarely have any method to test
them. The best they can do is replace any that are defective.
Dwight

This is why I was wary of getting chips from random suppliers, they're obviously either NOS or used pulls from old gear.

Glitch hooked me up with two 2114s, tested for a fair price so I have the memory chips I need now.

glitch
June 24th, 2017, 09:29 AM
I use various S-100 boards for testing most of my SRAMs and DRAMs. At this point I just about have a board for everything!

gslick
June 24th, 2017, 01:44 PM
This is why I was wary of getting chips from random suppliers, they're obviously either NOS or used pulls from old gear.


I grabbed 3 NOS tubes (63 parts total) of Harris HM1-6514-9 parts on eBay earlier this year at what seemed to be a decent price to me. From the datasheet they appear to share the standard 2114 pinout* with only 35mW active power. Not a part number I knew to be looking for, just came across them and looked them up. I'd have to look to see what the date code is on them to see how old they are. I haven't tried to test any of them yet to see if they all check out OK.

*While they have the standard 2114 pinout, internally the HM-6514 parts have address registers that are latched on the chip enable active edge, so they wouldn't work in a system that doesn't have chip enable transitions on every access cycle. Now that I look at it, that might be a problem on a SYM-1, which simply decodes the 6502 address lines to drive the memory enable signals. Something like access to consecutive memory locations in a given 1K bank might not work correctly. That would be a bummer since one of the systems where I thought I might use these is on a SYM-1.

glitch
June 24th, 2017, 03:51 PM
Correct, you'll need to qualify them with the Phase 2 clock, since addresses are guaranteed to be valid during Phase 2 time. Same thing with RAMtron (now Cypress) Ferroelectric RAM -- it latches on /CE, so it's got to be qualified as it ignores any changes once /CE goes low. That's part of the reason Harris used a different number, there *is* a Harris/Intersil 2114 that's a 100% normal 2114.

GiGaBiTe
July 4th, 2017, 05:55 PM
So mouser must have had the monkeys in the warehouse because they messed up my order legit which is a first.

I ordered 8 x 16p sockets and instead got 9 x 20p sockets. Another was 8 x 14p sockets, but I got 7 x 14p and 1 x 14p with the end snapped off and missing/crushed pins. Lovely. I know it wasn't the shipper because everything was in a hard box and the pins nowhere to be found. I had to make a trip up to Frys, which is like an hour each way just for some 16p sockets and a flux pen ugh.

Here are a few pictures for the curious:

Board Top - http://i.imgur.com/ZYJYmQf.jpg
Board Bottom - http://i.imgur.com/UGffTDn.jpg
CPU Area - http://i.imgur.com/VZIRqvs.jpg
Example of my crappy soldering - http://i.imgur.com/URZ1Un6.jpg

The shiny blobs is conformal coating from a pen, not solder blobs. I decided to leave most of the flux on (it's RA type) because the conformal coating is so brittle that even a light clean causes it to flake off. It's the same reason I'm using sockets for all of the replacement chips, even at very low heat levels the traces and through holes want to go away. I don't want to put any more time than necessary on them, and a socket allows swapping parts if they turn out to not be good. I think this board has sat in an unconditioned garage for a decade too long.

glitch
July 5th, 2017, 04:50 AM
Did the pair of SRAMs I sent do the trick?

GiGaBiTe
July 5th, 2017, 09:10 AM
All I know is they fit where they were supposed to go. I don't have the harness or the proper power supply to get the board running.

It wants +22v, +12v, +5v, -5v and -22v. So I'll have to wait until I can get it back in the cabinet to test it.

Here's another pic with all of the components replaced and circled in red, your supplied RAMs are on the middle left above the two large vertical chips (one is the 6502 and the other is the sound IC):

http://i.imgur.com/TmvB6Sc.jpg

GiGaBiTe
July 13th, 2017, 11:29 PM
So I finally got a chance to put the centipede board back in its arcade cabinet and no surprise, it still doesn't work.

The behavior is different though. Before I reworked the bad looking parts of the board, there was no picture on the CRT and the sound was garbled. Now there is something on the CRT, but no sound. There wasn't any fire or loud bangs, so I assume I got everything in correctly. I made sure to do continuity tests on all of the component legs I replaced to traces on the board, so I don't think I have any broken lines on my parts at least.

It does this when powered on (and excuse the potato camera, it's all I had at the time):
http://i.imgur.com/LM6nWLU.jpg

When you press the reset button on the centipede board, it does this:
http://i.imgur.com/BWWM2hb.jpg

http://www.youtube.com/watch?v=vzdoCiTBuhU

Every so often, I could see a corrupt piece of mushroom or a centipede segment being placed randomly on the screen, but the graphic is "chopped" up and down the screen (side to side since the CRT is mounted sideways with the top pointed to the left side of the cabinet.) I'm pretty baffled at this point and can't really proceed any further because I don't have proper test equipment. I think I'd need either a logic analyzer or a oscilloscope and an EPROM reader to check the six ROMs to see if they're corrupt.