130XE memory chips

[ylph]

Hi,

 

I am a new member here and a new owner of a 130XE with a faulty memory chip (boots straight to memory test). I used a 4164 from one of my working old socketed 800XL's to find the bad chip by piggybacking it over the soldered chips on the 130XE one by one until the memory test passed, and I could boot to BASIC.

 

The 130XE has a mixture of different kinds of memory chips on it, some are NEC 04164C-15 (I assume 150 ns ?), some are Micron 4264-15 (two different types it seems, a 411 and 427, also 150 ns ?) and some are Micron 4264-20 (three different types, 446, 434 and 435, 200 ns?) One of the 4264-20's (bottom right) has been replaced in the past (at least the solder joins look reworked on it, somewhat poorly actually) and the bad chip is also a 4264-20 next to it - bottom left. Was it common for the 130XE's to come with such variety of chips on the same board ? This one was made in Taiwan btw. All of my 800XL's (HK and Taiwan made) come with a set of matching identical chips (older ones are 300ns and newer ones are 150ns)

 

What is the difference between the 4164's and the 4264's - from everything I found they should be interchangeable, but I am still curious why the different designation, and also the 4264's seem much harder to find than the 4164's ? Also is there an ideal speed for the chips ? I see 4164's at anywhere between 100ns to 300ns, I would assume that the lower the better, however is there any harm in using 100ns chips in the 130XE, combined with other 150 and 200 ns ones ?

 

Finally what do you guys recommend as a best source to buy small quantities of these chips inexpensively ? I see them sold on-line in various places anywhere from $0.22 per chip to $5 per chip.

 

Any suggestions/insights are greatly apreciated ! This seems like a very knowledgeable group of Atari-heads

 

Thanks.. ylph.

[+Guitarman]

The 4164 and 4264 are, spec wise, the same. The difference in number was the manufacturers doing. The reason 4264 is hard to find is because it is specifically a Micron (MT) number. Most of the other makers wil be labeled 4164.

 

Someone correct me if I'm wrong.....

 

I also had a 130XE with bad memory but was quickly fixed. I would suggest socketing all of them if you can.

 

As far as speed goes, try to get the smaller number. Most Atari's I have seen have been loaded with 100-200ns chips. Try to match speeds if you can because the memory bus will only run as fast as the slowest chip in the bank.

Edited July 20, 2006 by Guitarman

[Rybags]

250 ns is supposedly the slowest memory that will work on the Ataris.

[MEtalGuy66]

YEah that "replacement" chip your talking about was probably actually done on the assembly line. Ive seen quite a bit of that on 130xes.. ATARI used the absolute cheapest garbage memory they could find on the 130xe.. And if it didnt work at burn-in time, they isolated which chip, and changed only that one. And that would have been soldered by hand, versus the rest of the board which had already been mass-soldered in a solder bath on the assembly line.

 

When TRameil took over ATARI, the 8-bit line got some MAJOR cost cutting.. hence also the lack of sockets.. hence the garbage ram chips... etc..

 

Whenever I get a 130xe, the first thing I do is remove ALL of the ram, and throw it in the trashcan..

 

The 4164 and 4264 are interchangeable... But by far the most common replacement chip you can find (and ussually the cheapest) is the 41256 which is actually a 32kx1 chip (instead of an 8kx1 like the 4164) but it wont hurt anything.. THey will plug right in, and the 2 extra adress lines just wont be used.. And if you ever want to do a 320k memory upgrade, you will already have the RAM.. JUst hook up the extra adress line (one is alrady hookd up) and the additional MUX logic..

[+wood_jl]

Bump to an old thread!

 

 

 

I have an old 130XE I hadn't tried, until tonight.

 

In all my years of XL/XE, I've never seen this screen come up automatically like this. Well, we know what the problem is.

 

I'm thinking about giving a repair a try; I have some old Commodore and TI/99 boards that I "practiced" on and while I'm a ham-fisted idiot, I don't see how I'm going to gain any experience unless I try.

 

This 130XE: She's an old dog, and not the best to look at, anyway. She'd probably clean up OK with some peroxy treatment on the keyboard, but honestly, I don't think it's worth the expense of paying someone to fix it. The el-cheapo desoldering iron thing I have seems to work pretty well....at least on those other boards. I understand there's a great risk I'll delaminate a trace or tear one out pulling the chip out without getting all the solder off of it. It's a chance I'm willing to take. If I destroy the board, I'll either (1) never be inclined to try this again, or (2) learn something and actually get closer to being able to do this sort of thing. I know the risks.

 

Remember AA user "Syfo-Dyas" did quite a number of "mods" before he ended up trashing his board. I'm encouraged enough by that to try.

 

I just wanted to ask where people recommend getting RAM chips and sockets, in the small quantities this machine requires, and roughly the cost. I see what I *think* may be the ones on Ebay, but I thought one of you big experienced brains on here might be able to point to something more specific. I don't know the terminology....for the proper sockets, for example.....so it's a bit hard for me to search for it.

 

One thing for sure; the Tramiel era 8-bit computers finally *did* create a need for that Memory Test that I never thought I'd "use." If the machine calls it up automatically as it does, what would be the point in doing it manually?

 

Thanks for any tips.

 

 

 

EDIT: I haven't looked inside yet (very shortly) but this is an OLD "Made in Taiwian" model so I'm expecting it to be the model with the greater number of RAM chips.

Edited February 21, 2012 by wood_jl

[sloopy]

TIP#1, that was the FIRST Atari mod he performed... and it was a XL mobo NOT an XE, XE are much more fragile...

 

TIP#2, the self test comes up on mem error as a coincidence, not intentional... not all mem errors will bring it up...

 

TIP#3, if $35 is too much to get a 130XE fixed, then how much do you get 130XE's for?

 

TIP#4, crazy legs in the 5th ;')

 

sloopy.

[Rybags]

The second 16K is screwed, which happens to be where extended bank switching occurs, so possibly a problem with EMMU or PIA.

 

Ignore the 2 "good" squares - Self-Test lives @ $5000 (2K) and instead of having some code in the normal OS area to test that RAM, it just gets marked as passed with a delay loop to simulate that 2K being tested.

[Bryan]

The easiest way to remove all RAM from an XE is to get some fine tip diagonals and snip all the pins off the RAM (a Dremel with a small cutting wheel works too). Then you can remove the pins one at a time and clear the holes with a solder sucker. Use as little heat as possible, since the traces come off the board easily. Then put in sockets.

[Rybags]

I don't think the RAM is at fault.

 

Every chip in a column contributes a bit to each RAM access (or 4 bits for such models).

 

For the failure to happen to be that entire 16K area is pretty unlikely. Much more likely is the Ram's OK but the selection logic isn't.

[Defender II]

The easiest way to remove all RAM from an XE is to get some fine tip diagonals and snip all the pins off the RAM (a Dremel with a small cutting wheel works too). Then you can remove the pins one at a time and clear the holes with a solder sucker. Use as little heat as possible, since the traces come off the board easily. Then put in sockets.

That is the best way to remove them and prepare to remove the pins. Just remember to cut near the chip to leave enough to grip with a needle-nose pliers or the diagonals when you go to remove them.

Use a fine chisel tipped iron so you can have better control of where the heat is applied. Don't try to work on a 130XE with one of those cheap irons or you're begging for trouble.

You would be better off practicing on a newer broken PC motherbaord as the traces are pretty fine on those. If you are successful on that than you may be successful on the 130XE.

Mount the PCB in a holder so you can access the front and back of the board at the same time.

If you have a desoldering iron just apply it to the back of the board.

If you have a solder sucker and iron then position the board so you can apply the sucker to the pin on the back side and the iron to the same pin on the component side.

Prepare the solder-sucker and position it on the pin on the back side, pretin the tip of the iron and then apply it to the join between the pin and the pad creating a solder bridge with the solder on the tip of the iron.

When the solder in the joint flows all the way through to the back pull the trigger on the sucker, then remove the iron.

Once the solder is remove check if the pin is free. If it is remove it. If not wait for the joint to coll and then re-apply the tip to the pin and try to remocve it.

 

If you want to try a short-cut use a needlenose plier to hold the cut off part of the pin then apply the tip of the iron to the back side.

When the solder flows all the way through to the front gently pull out the pin. If there is resistance don't continue to pull or you will lift a pad and possibly a trace.

Some pins may require straightening on the back before trying to remove them.

 

Even very skilled electronics techs can lift traces on the 130XE.

Good luck on your project.

Edited February 21, 2012 by Defender II

[+wood_jl]

Wow, thanks for the replies, folks!!! I very much appreciate it. As corny as it sounds, I am really honored to share the forum with such knowledgeable people! I only wish I were 1/10 as capable. I have other A8 machines, so I guess I'll just have to think about this.

 

 

I found this other thread.....(1yr ago)

 

http://www.atariage.com/forums/topic/177355-replacing-bad-ram-in-130xe/

 

....where there are Ebay links to RAM, etc.

 

I think the thing to do is socket the entire motherboard, not a job for the hamfisted.

 

I have Ultimate 1MB I've been afraid to install. I wonder if it would work in this machine? Since it has its own RAM and the MMU comes out (uh, it's soldered too, of course). Trouble is, I am reluctant to install (or have installed) an expensive mod like that into a piece of crap machine. I'd rather get a nicer one to do that with. Dilemma. I suppose I'll think about it.

 

But thanks again for the food for thought on this subject. I don't feel confident, but more informed.

[Rybags]

I still think your Ram is probably OK.

 

AFAIK Ult 1 Meg still uses the existing Ram in any configuration.

 

But it replaces the MMU, so might fix your problem. The installation is easily reversable, the extra soldered in wires are the only thing that would have to come off.

[RodLightning]

Bumping the thread again. I recently got a 130XE with the same memory test screen as described by wood_jl. This machine unfortunately has all 16 drams of the MT variety.

 

I will add that it loads a Star Raiders cartridge, but otherwise will go to the memory test. The piggyback trick to test bad chips didn't work. No effect except when piggybacking a good dram onto position U10 or U11, more errors were generated than in the pictured screen. Piggybacking U11 caused the ROM blocks to turn red.

 

Before I start trying to replace memory chips, is it worth replacing the EMMU or Freddie? Needless to say, nothing on this board is socketed.

 

[Rybags]

Star Raiders is a "diag" type cart and gets control almost straight up. Also it only uses the first 8K of Ram as it was a very early release.

 

The second 16K row - like I said before, ignore the 2 "good" squares as the Self-Test doesn't bother testing Ram under where it resides $5000-$57FF.

The failing row corresponds to the $4000-$7FFF area which is candidate for 130XE extended banking.

Your 130XE uses 8 Ram chips for each Ram address access, each provides 1 bit. Failing only in that row indicates more a likely memory select problem than a Ram chip problem.

Swapping out the EMMU, good idea. If that's no good then maybe try the PIA.

[Joey Z]

Star Raiders is a "diag" type cart and gets control almost straight up. Also it only uses the first 8K of Ram as it was a very early release.

 

The second 16K row - like I said before, ignore the 2 "good" squares as the Self-Test doesn't bother testing Ram under where it resides $5000-$57FF.

The failing row corresponds to the $4000-$7FFF area which is candidate for 130XE extended banking.

Your 130XE uses 8 Ram chips for each Ram address access, each provides 1 bit. Failing only in that row indicates more a likely memory select problem than a Ram chip problem.

Swapping out the EMMU, good idea. If that's no good then maybe try the PIA.

additionally, for testing, (if you don't have one to easily swap) you can completely disable the EMMU like the PAL 65XE w/ 130XE motherboard.

[RodLightning]

Star Raiders is a "diag" type cart and gets control almost straight up. Also it only uses the first 8K of Ram as it was a very early release.

Thanks for the good info. I knew to try Star Raiders, the part about first 8k of ram being used in new to me.

 

The second 16K row - like I said before, ignore the 2 "good" squares as the Self-Test doesn't bother testing Ram under where it resides $5000-$57FF.

The failing row corresponds to the $4000-$7FFF area which is candidate for 130XE extended banking.

Your 130XE uses 8 Ram chips for each Ram address access, each provides 1 bit. Failing only in that row indicates more a likely memory select problem than a Ram chip problem.

Swapping out the EMMU, good idea. If that's no good then maybe try the PIA.

I like the idea of trying another EMMU before pulling drams, even if they are from the dreaded Micron Technology bad batch.

In another thread, I read that not all 4164 MT drams out there are necessarily defective, just the ones sold to Atari. It makes me

wonder how many 130XEs are destined to fail due to bad ram and if any good MT memory is out there in any of them.

 

Also wondering if bad dram shorting out can cause EMMU, PIA or Freddie to fail. My other 8bits are 400/800 and XL so no spare XE

EMMU in the parts bin. I will need to mail order a new C025953 from somewhere. Although consensus on AtariAge says that these Micron

chips will go bad, desoldering, socketing and replacing one 16 pin chip as a first step seems worth a try.

 

Reading about the lifted traces and pads on XE circuit boards is discouraging. I have done a fair amount of soldering over

the years but no idea how I will do with this particular board. I have managed so far with flux, wick and a suction tool

on other projects. I also have a radio shack desoldering iron here somewhere, the one with a big orange bulb attached.

I shudder at the thought of pulling out and socketing all 16 memory chips on such a delicate board. That's 256 chances to screw up a solder pad!

 

I can see this project taking many days of cutting pins and slowly removing pieces. I think I've rambled enough, wish me luck folks.

This may take a while.

[RodLightning]

additionally, for testing, (if you don't have one to easily swap) you can completely disable the EMMU like the PAL 65XE w/ 130XE motherboard.

I don't have a spare emmu. What's the procedure to disable it on ntsc 130xe with 16 drams?

 

I'm also considering trying to swap banks as described in this other thread to further diagnose:

 

http://atariage.com/forums/topic/218035-atari-130xe-ram-test-failure/?do=findComment&comment=2856697

Edited May 7, 2017 by RodLightning

[Joey Z]

I don't have a spare emmu. What's the procedure to disable it on ntsc 130xe with 16 drams?

 

I'm also considering trying to swap banks as described in this other thread to further diagnose:

 

http://atariage.com/forums/topic/218035-atari-130xe-ram-test-failure/?do=findComment&comment=2856697

I'm not entirely sure, I do know it's a handful of resistors and/or wires, that's it. I can probably figure it out of I look up the EMMU pinout real quick....

 

EDIT: Unfortunately, I can't find a pinout I completely trust, nor do I have a 130XE here to confirm it. Is the current EMMU 14 or 16 pins?

 

Anyway, if it's 16 pins, maybe this works:

 

looks to be...

pin 1 to pin 12

pin 2 to pin 13

pin 6 to pin 11

pin 10 to pin 16

 

This will leave you with only 64K of RAM, of course, with the extended bank disabled. You can also swap the role of pin 10 and pin 11 to use the other bank of DRAMs for the 64K.

 

I'm going from jerzy sobola's schematics, which are known to have the occasional error, so I'd appreciate it if someone with a PAL 65XE w/ 130XE mainboard can check this out. I also seem to remember some of these were done with resistors rather than wires, but I'm going from memory.

Edited May 7, 2017 by Joey Z

[Rybags]

The EMMU decides which bank the Cas signal gets sent to (I'm guessing both get Ras regardless).

Casman and Casbnk go through resistors according to the schematic so maybe just disconnect the resistor for Casbnk and tie the output side to GND?

That should turn it into a 64K only machine although switching access to extended banks would just create a memory hole, but for testing that shouldn't matter.

[RodLightning]

EDIT: Unfortunately, I can't find a pinout I completely trust, nor do I have a 130XE here to confirm it. Is the current EMMU 14 or 16 pins?

This will leave you with only 64K of RAM, of course, with the extended bank disabled. You can also swap the role of pin 10 and pin 11 to use the other bank of DRAMs for the 64K.

 

I'm going from jerzy sobola's schematics, which are known to have the occasional error, so I'd appreciate it if someone with a PAL 65XE w/ 130XE mainboard can check this out. I also seem to remember some of these were done with resistors rather than wires, but I'm going from memory.

 

The EMMU I have here is indeed 16 pins. Also, the CAS_MAIN and CAS_BANK signals run through 33 ohm resistors on the board, so probably the same value for jumpering pins will work? Does the EMMU chip need to be removed and pin positions jumpered, or remain in place?

[Joey Z]

 

The EMMU I have here is indeed 16 pins. Also, the CAS_MAIN and CAS_BANK signals run through 33 ohm resistors on the board, so probably the same value for jumpering pins will work? Does the EMMU chip need to be removed and pin positions jumpered, or remain in place?

removed.