Testing 4116 RAM chips out of circuit

[Ed in SoDak]

Would it be possible to test 4116 RAM out of circuit for possible shorts in the power pins? Maybe by using the diode check function of my DMM to the 12v, -5 and +5 pins to ground? I have a bunch of pulls, one of which may have a short. I'd like to be able to cull it out by means other than testing for smoke in an otherwise working console.

Thanks!

-Ed

[Gary from OPA]

You should be able to see a very small amount of ohms between ground and power pins.

 

If there is zero ohms, then most likely it is shorted out.

 

Another strange thing, never figured out logic of why it might work, but I had chips were there was like a bank of 'stuck' bits not going 1 or 0 in the array, but the rest of chip was ok, but here is strange thing I was able to 'restore' fix the chips, by taking the bad ones and wrapping them up in tin foil nice and tight so all the pins were shorted together and then connecting the tin foil to earth ground, waiting a few days, unwrapping the chips and then plugging them back into the pcboard sockets, and zippy do da, they were fully working again.

 

Never figured out why, and thought it was weird one-off thingie, but I did it a few more times over the last few years and it worked each time, as long as chip only had 'stuck' bits and was otherwise reading/writing the complete bank of memory, I guess maybe I can speed up the process by building a bare pcboard with sockets with all the pins wired up to earth ground, but have not tried that yet.

 

Any how back on topic, take a chip that you know is working, check out the amount of ohms you get between the power pins and ground, make sure you use a good digital meter that can correctly read low amount of ohms in the 10 and down range, and then check out your other chips that you not sure if they working, if its anything like zero to 3 ohms it is basically going to smoke when you plug it in.

[Ed in SoDak]

Thanks for the reply! I'm working on a Timex TS1500, but the RAM is the same as in the TI.

 

The tinfoil trick is funny, maybe if I wore a tinfoil hat too it would be even better? A friend was testing them via software in a working ZX81 computer in a socketed RAMpack. He tried "hammering" the stuck bits by repetitively poking a number into specific memory addresses.

 

I have a few pulls from a TI99 console, two of them immediately heat up the power supply transistor. Using the diode check on my DMM, I get different readings among these used chips, but nothing definitive so far. My TS1500 may have more than one bad chip, but poking/peeking the memory seemed to say the trouble was on the D3 dataline, so I started there.

 

Looks like I may end up pulling all the RAM and installing new chips. Once i track down a good source, that is.

 

EDIT: Got 16 RAM chips and other parts on the way from Jameco. Maybe I can get a RAM pack going too.

 

-Ed

Edited August 25, 2014 by Ed in SoDak

[+arcadeshopper]

my eprom programmer will also test ram.. but it requires an adapter for that size chip

[Ed in SoDak]

Nice capability, if you happen to have one! I looked at programmers while I was shopping for RAM at Jameco. The cheapest used a parallel port and DOS, so that means I'd have to drag (yet another) old system out. It's cheap enough at about $300, but still too expensive for the limited uses I'd have. USB programmers probably don't support Mac and also cost quite a bit more. I could skip the testing, buy a lot of parts and sockets and still be way ahead of the game.

 

But what to do with those pulls? Right now, I'm taking the tack my Canadian friend Ian used. He made a purpose-built Timex setup with all socketed RAM just for the purpose of trying out various Unknown Status 4116 chips. A simple added circuit kept the internal 2k RAM active and remapped the RAMpack to a "safe" area for testing with software.

 

I'm learning a lot along the way. While robbing parts from failed boards, I accidentally discovered one merely had a bad transistor, so I got another TS1000 working as a bonus. I'm trying to fix a broken RAMpack now to fit onto it. Hopefully this will spare the kinda scarce TS1500 and my working TS1000 from the ordeal of testing suspicious RAM chips.

 

Such a setup could also be used to test TI99 VDP RAM chips. So far, I'm just having fun getting old stuff working again!

 

-Ed

[Gary from OPA]

Such a setup could also be used to test TI99 VDP RAM chips. So far, I'm just having fun getting old stuff working again!

 

-Ed

Yeah, I have couple of TI99 consoles that have signs of bad VDP RAM, it just so bad that it is not socketed, and the 30 year old crappy tin solder is so hard to melt, plus TI bend over 2 of pins of each chip to make sure they don't fall out. Such a pain in ass to remove.

 

The other causes of bad ram in systems is not the chips but in some cases the power rails supplying the 3 voltages to the bank of chips, the caps in the area can be problem, in the TI99 those white long strips I found to be major issue after 30 years, resoldering or replacing them helps out alot.

[Ed in SoDak]

I hear ya! I removed the RAM and socketed a TI once, and only once, and that didn't fix it! Moved it on as a parts unit and last I heard, it was still dead, though he tried mightily to get it going, as I had years ago.

 

Gets me to wondering... I remember seeing test adapters that gripped an IC and brought all the contacts up for easy connection of test probes. Now, what if you could make a plug and short ribbon cable to fit a RAM socket installed in a Timex ts1000's RAMpack. The other end of the ribbon has that clippy IC testy thingy on it. (Highly technical terms!) The big IF is: if an IC can be tested "live" this way while in-circuit in another machine.

 

The Timex is small and light and could be placed near or even on the TI motherboard to keep the cable short. Testing without desoldering would be a very cool thing for TI-ers and Timex or other confuser owners.

 

Or maybe just a logic probe is all that might be needed but that probably requires the TI to be powered up for testing, which may not be easy or possible due to an undetermined fault on the motherboard.

 

Probably easy enough to implement for a tryout. Hmmm.

 

-Ed

Edited August 26, 2014 by Ed in SoDak

[Ed in SoDak]

Got the 1500 going again! New member TMA-1 (Ian) gave me his RAM tester program for the 1000 which was well-suited to use on the 1500 with a RAMpack to give it operational memory while the internal RAM lay dormant. That gave repeatable errors that pointed to the chip on data line D3. Traced that down to the suspect chip, removed and socketed it. When my new RAM chips arrived, I plugged one into the socket and the patient has a pulse!

 

That doesn't answer how to test the RAM without power or out of circuit, but at least the 1500 is working again.

 

-Ed

[sjt]

To test the ram, use a known good chip ,bend the pins in and place over the ram in question.

Look at the screen of the computer under test. If the display changes, replace that chip. If it still doesn't look right continue testing.

This is not 100% effective, but will answer the question more often than not.

 

Steve

[Ed in SoDak]

Thanks for the reply. Sure, a known-good chip can help troubleshoot RAM that's soldered in the circuit. But how about for checking unknown pulls?

 

When I was piggybacking some old chips removed from a TI99, I managed to blow a power supply transistor twice, because a couple of the chips were apparently shorted. But shorted in a way that I could not determine that fact beforehand with a DMM. Or I just missed finding the shorted pins.

 

That discouraged me from using the piggyback technique for checking RAM when the working status of the pull was unknown.

 

I wasn't trying to find the bad RAM in the Timex 1500, we had already determined that with a software program. I was hoping to learn how to test some old RAM chips which were already removed from a TI99 console. Hopefully without blowing more parts in the 1500. So I was looking for an Out of Circuit method that did not risk damage to the computer itself.

 

Moot point now, since one new chip got the 1500 working, but I still have old RAM pulls it would be good to be able to test without risk to the computer.

 

In this instance piggybacking using RAM of unknown working status just got me into trouble. I was able to determine that two of my pulls were shorted chips, since they heated up the transistor very quickly and would blow it in a matter of a couple seconds. These I discarded.

 

So yes, piggybacking can help, but only if the piggy on top is in good health to start with, lol.

 

-Ed

[sjt]

It's certainly possible to have shorted ram chips. most often they are just "stuck" However when in circuit the excessive current draw should give them away easily, they will cook. I've found that it's not hard to slip when placing chips over others, bridging pins. That's why you must exercise extreme caution when using this method. Check it carefully before turning on the power.

Out of the circuit, you should be able to detect a shorted condition. Simply take a known good chip and make a resistance chart.

Any abnormality from the known good chip would indicate an issue. However, there may be variances between chip manuf.

If your sure you put a chip over another and the pins were not bridging,and it blew a regulator, it stands to reason that you must have low resistance to gnd on pins (pretty much a dead short) . As you know those chips have 3 supplies. If you socketed a board and placed 1 chip at a time in, unless the chip has a dead short, it should not take out a regulator. If it does you should have been able to read that on your meter

Another thing you can do is run the board using a bench supply and current limit the +5,-5,+12 supplies.

[Ed in SoDak]

I tried resistance checks of the power pins and didn't catch anything. Either I missed it or it's to different pins.

 

A couple months ago I bought a 1500 that would not boot. Voltages were good and subbing out the main chips didn't improve it, so RAM was suspected, but I didn't know which chip.

 

I had eight used pulled 4116s in my junk drawer and I was piggybacking the used chips onto each of the 1500's and that's when I blew the transistor. I think this transistor provides the -5v. I'm pretty sure I wasn't shorting any pins, but that made me nervous about proceeding with that method.

 

A friend provided a short software routine that tested each memory location by writing 0 and 255 and reading it back to see if it worked. The 1500 had one stuck chip discovered when the write/reads gave an error only on the bit that one particular chip controlled. I had one socket on hand, so I pulled that lone 4116 and socketed it.

 

Now I didn't need to piggyback to test my used chips, so I tried them all in the socket. When I tried these two shorted chips, the transistor would immediately heat up. When I installed the socket, I had also substituted a much beefier transistor, and it was better able to handle the short. I kept one finger on it as I tried each 4116. When it became hot in mere moments, I shut it down and got the shorted 4116 the heck out of there. While "safer" than before, it was still a risky proposition.

 

Turns out none of my pulls "fixed" the 1500, but they did produce a different screen display, so I knew I was on the right track.

 

I ordered a spare transistors, a full set of sockets and sixteen 4116s, but as it turned out, only one chip was bad in the 1500, the one we had already removed and socketed. So now I have plenty of new spares, but I'm still curious to see if I can determine the status of my pulls with out of circuit tests.

 

I plan to compare one of the new chips to the pulls and see what transpires. I don't have a current-limited supply, but do have a bare board Timex 1000 with RAM pack to act as the test base. The 1500 is too scarce and cool to risk any further, especially now that it is working.

 

I do have a Radio Shack electronics trainer with a small breadboard that runs on batteries. I could probably rig up a test circuit on it to do a short check more easily than hand-holding chips and meter leads.

 

-Ed

Edited September 17, 2014 by Ed in SoDak