FREDDIE chip seems suddenly dead

[ijor]

I was performing some tests on my 130XE and suddenly it stopped working. After some diagnosis seems to be that FREDDIE is dead. At least there is a good 14.3 MHz input signal, but no 3.57 MHz output to GTIA. As a matter of fact FREDDIE pin 37 (OSC output) doesn't seem to be fully stable (guess it is floating).

 

PSU seems to be fine. It works on a different computer. And trying a different PSU on this 130XE still gives the same, FREDDIE dead, behavior.

 

A couple of weeks ago on of the DRAM chips died. But being an "MT" chip, known to be problematic, I didn't pay too much attention. But now I'm a little bit suspicious to get two chip deaths one after the other. Might still be the PSU, even when it seems to be working fine on another computer? Or may be it was just a coincidence and bad luck? Or this behavior is not so uncommon?

 

Thanks

[+bob1200xl]

You should measure the output voltage of the PSU, both loaded and unloaded. It may be high (>5.25 v), which will stress all the chips. They may not fail right away, but over the long run, you might get failures. It took two weeks for Freddie to fail after the DRAM died, right?

 

Worse scenario is the PSU randomly spiking over voltage. Put a scope on it and see if it triggers once in a while. I set mine to a very slow time, like 1s/DIV. Or, you can use Single Sweep.

 

Two failures is a co-incidence, three would be a Bug!

 

Bob

 

[ijor]

2 hours ago, bob1200xl said:

You should measure the output voltage of the PSU, both loaded and unloaded. It may be high (>5.25 v), which will stress all the chips.

...

Worse scenario is the PSU randomly spiking over voltage. Put a scope on it and see if it triggers once in a while. I set mine to a very slow time, like 1s/DIV. Or, you can use Single Sweep.

 

Good tip. I measure 5.20-5.23V without load, and 5.17-5.20V with load.

 

Detecting spikes is a little more complicated for me. Will try to hook a reasonable instrument later.

 

2 hours ago, bob1200xl said:

It took two weeks for Freddie to fail after the DRAM died, right?

 

More or less, but according to the calendar! But I barely use this computer. So if you ask how much time it was actually turned on between both events, I'd say just minutes, may be half an hour. Sounds too bad?

 

Thanks again

[ijor]

Btw, this is a 220V PSU. So no Ingot model, that I just read they are dangerous

Edited December 19, 2020 by ijor

[BillC]

49 minutes ago, ijor said:

Btw, this is a 220V PSU. So no Ingot model, that I just read they are dangerous

I don't think the INGOT was restricted to 110V, the post linked below has a 220V power supply that looks similar. 

 

How heavy is the power supply?

 

Edited December 19, 2020 by BillC

[ijor]

On 12/19/2020 at 5:59 PM, BillC said:

I don't think the INGOT was restricted to 110V, the post linked below has a 220V power supply that looks similar.

How heavy is the power supply?

 

I stand corrected, I wasn't aware. Indeed, my PSU is precisely that mentioned in the post you linked. And yes, it is quite heavy. Surely it's filled with epoxy. But not all epoxy filled PSUs are that bad, are they?

 

However doesn't look that my PSU suffered yet the dangerous death as mentioned in other INGOT threads. At least it seems to output a, more or less, reasonable voltage. And at least the first time, it didn't fry my computer because only one RAM chip was damaged. Of course, I didn't test yet for spikes, so the PSU might still be the culprit. And now with FREDDIE gone, I don't really know if the other chips are working or not. Difficult to test when FREDDIE doesn't produce the main system clock.

 

Is the Ingot PSU know to produce spikes as well? Or, as reported in other threads, it's ok until something breaks inside and start outputting a crazy voltage level? Just asking to have a reasonable idea about what produced the damage in my case.

Edited December 22, 2020 by ijor

[shoestring]

Machine is over 30 years old so I wouldn’t always suspect the PSU but considering it’s an ingot, I’d not be using it in the long term. It’s definitely not uncommon for more chips to fail on your bench one after the other as you’re working on it, I see it constantly with arcade stuff I’ve worked on. My understanding is that FREDDIE is CMOS as opposed to the other chips which are NMOS,  so they won’t take as much of a beating and are usually prone to being one of the first chips to go a long with the DRAM. DRAM failures were a huge problem even back in the day when these chips were still brand new.

 

You could find a NOS replacement fairly easily as I did. But with new tech available it would be much easier to get an SRAM replacement or even the Antonia board as everything required is on there. But I’d still be checking with Simius first to confirm that it will run happily without the FREDDIE chip.

 

 

[BillC]

4 hours ago, ijor said:

 

I stand corrected, I wasn't aware. Indeed, my PSU is precisely that mentioned in the post you linked. And yes, it is quite heavy. Surely it's filled with epoxy. But not all epoxy filled PSUs are that bad, are they?

 

However doesn't look that my PSU suffered yet the dangerous death as mentioned in other INGOT threads. At least it seems to output a, more or less, reasonable voltage. And at least the first time, it didn't fry my computer because only one RAM chip was damaged. Of course, I didn't test yet for spikes, so the PSU might still be the culprit. And now with FREDDIE gone, I don't really know if the other chips are working or not. Difficult to test when FREDDIE doesn't produce the main system clock.

 

Is the Ingot PSU know to produce spikes as well? Or, as reported in other threads, it's ok until something breaks inside and start outputting a crazy voltage level? Just asking to have a reasonable idea about what produced the damage in my case.

I was going through some stuff I got a while back and found 4 INGOT power supplies, 3 are within spec @ 5.1 to 5.2V but the last was @ 5.95.

I recently took the bottom off it, leaving it otherwise intact to investigate how it failed, it appears that it is the 78S05 voltage regulator that has failed and not a resistor in the voltage divider/boost circuit as discussed in another thread.

 

I measured the output voltage of the 78S05 @ 5.6VDC to the ground terminal, plus another 0.35VDC between the ground terminal and the transformer center tap. The center tap of the transformer connects to the ground wire of the DIN7M plug, I think the fuse is installed in this path. The transformer output is 9.35VAC, after the diodes it is 7.65DC to the center tap or 7.3VDC to the 78S05 ground terminal.

 

I believe the worst case failure mode of this type power supply could see 7.0 to 7.65VDC at the output.

 

I had another INGOT fail back in the late 80s which killed the 8 x 41256 chips in a RamboXL upgrade, the rest of the chips were fine but the failed RAM may have consumed enough power to keep the voltage low enough to save them. Since the 41256 were higher density I believe they were more susceptible to be damaged by the higher voltage than the rest.

 

 

 

[shoestring]

25 minutes ago, BillC said:

I was going through some stuff I got a while back and found 4 INGOT power supplies, 3 are within spec @ 5.1 to 5.2V but the last was @ 5.95.

Interesting failure mode for a 7805 if that Ingot is indeed using one, they all use those right ? Some of the bad Commodore power supplies use a different regulator which are more prone to over voltage but the 7805 seems to be quite safe to use.
 

I’ve had someone jump in on one of my threads saying they’d never seen that type of failure mode in their 25 years of experience. 
 

Was that under load conditions? Either way 6v is too high.
 

 

 

[ijor]

12 hours ago, shoestring said:

Machine is over 30 years old so I wouldn’t always suspect the PSU but considering it’s an ingot, I’d not be using it in the long term. It’s definitely not uncommon for more chips to fail on your bench one after the other as you’re working on it, I see it constantly with arcade stuff I’ve worked on.

 

Thanks for the advice.

 

12 hours ago, shoestring said:

My understanding is that FREDDIE is CMOS as opposed to the other chips which are NMOS ...

 

Interesting. I'm not so sure, where you read that? If you look at the values at the preliminary engineering datasheet, it seems to be an NMOS device. And 1983 was probably too early for a fast CMOS LSI anyway. Of course, the final product might be different and by 1985 CMOS fab was much more advanced. But FREDDIE was already produced for the 1400XL which was released earlier, and seems the 130XE used, at least initially, exactly the same NCR part. There is a later FREDDIE version with a different part number, conceivable could be a CMOS redesign.

 

The chips that I'm pretty sure are CMOS gate arrays are MMU and EMMU. May be this caused some confusion?

 

FREDDIE engineering datasheet is here: https://web.archive.org/web/20060214093449/http://www.atarimuseum.com/ahs_archives/archives/pdf/computers/8bits/freddie-mcu.pdf

 

 

[ijor]

9 hours ago, BillC said:

I was going through some stuff I got a while back and found 4 INGOT power supplies, 3 are within spec @ 5.1 to 5.2V but the last was @ 5.95.

I recently took the bottom off it, leaving it otherwise intact to investigate how it failed, it appears that it is the 78S05 voltage regulator that has failed and not a resistor in the voltage divider/boost circuit as discussed in another thread ...

Thanks again for all the info.

[shoestring]

12 hours ago, ijor said:

 

Thanks for the advice.

 

 

Interesting. I'm not so sure, where you read that? If you look at the values at the preliminary engineering datasheet, it seems to be an NMOS device. And 1983 was probably too early for a fast CMOS LSI anyway. Of course, the final product might be different and by 1985 CMOS fab was much more advanced. But FREDDIE was already produced for the 1400XL which was released earlier, and seems the 130XE used, at least initially, exactly the same NCR part. There is a later FREDDIE version with a different part number, conceivable could be a CMOS redesign.

 

The chips that I'm pretty sure are CMOS gate arrays are MMU and EMMU. May be this caused some confusion?

 

FREDDIE engineering datasheet is here: https://web.archive.org/web/20060214093449/http://www.atarimuseum.com/ahs_archives/archives/pdf/computers/8bits/freddie-mcu.pdf

 

 

The XE was released in 85. So manufacturing FREDDIE in CMOS was not a problem. I read it was manufactured in CMOS here, I’d trust that it’s correct info seeing that Jurgen is an engineer and makes some great add ons for the A8.

 

He helped me with my 65XE repair, reach out to him. He is very helpful.

 

 

[ijor]

2 hours ago, shoestring said:

The XE was released in 85. So manufacturing FREDDIE in CMOS was not a problem.

 

The XE was released in 85, but again, FREDDIE was released in 83 for the 1400XL. And some 130XE computers, including mine, use exactly the same FREDDIE chip that used the 1400XL.

 

2 hours ago, shoestring said:

I read it was manufactured in CMOS here, I’d trust that it’s correct info seeing that Jurgen is an engineer and makes some great add ons for the A8.

He helped me with my 65XE repair, reach out to him. He is very helpful.

 

I'll ask Jurgen in the thread you linked. He might have some info. But again, there are at least two different versions of FREDDIE, so it is very possible that the newer version (and only the newer version) is CMOS.

 

I'm sure he could repair my computer. Unfortunately I'm not located in Europe. Thanks again.

[shoestring]

I really don't know why that even matters tbh.

[ijor]

On 12/24/2020 at 10:29 AM, shoestring said:

I really don't know why that even matters tbh.

 

I just find the issue very interesting. True, it is not very relevant to the original post, except may be for the purpose of understanding why FREDDIE was specifically affected by a potential PSU problem.

 

But as somebody that reverse engineers chip die layouts, and specifically I reverse engineered the whole Atari chipset, I am very interested in all the fab and MOS process details. Sorry if I got carried away here and I got a bit off topic.

[Rybags]

What's the deal with overvoltage tolerance for NMOS vs CMOS?  From what I've heard CMOS is way more sensitive to static but does the same apply for the general power supplied to the chip?

[BillC]

On 12/22/2020 at 7:09 PM, BillC said:

I measured the output voltage of the 78S05 @ 5.6VDC to the ground terminal, plus another 0.35VDC between the ground terminal and the transformer center tap. The center tap of the transformer connects to the ground wire of the DIN7M plug, I think the fuse is installed in this path. The transformer output is 9.35VAC, after the diodes it is 7.65DC to the center tap or 7.3VDC to the 78S05 ground terminal.

I decided to recheck this power supply with a load, I used a KPR113 flashlight bulb(4.8V@0.75A) and socket from an old 6V flashlight. I found that the 78S05 output that had been ~5.6V is reduced to ~5.1V. The voltage to the transformer center tap, used for the DIN7M plug negative, drops from ~5.95V to ~5.4V which still exceeds 5% tolerance. Without the voltage divider boost circuit the voltage would still be slightly high but within tolerance.

 

I still believe this 78S05 is starting to fail, due to the slightly high(102%) output voltage under load and much higher voltage(112%) without a load, but could still be considered within tolerance at this time. The voltage divider boost circuit is a problem, 5.1V is OK but 5.4V is high. I am considering putting a jumper from the 78S05 ground to the transformer center tap to eliminate the voltage boost, then leaving a load on it and checking the voltage periodically.

[BillC]

As a followup to my previous post I load tested another INGOT that is still within tolerances, I connected the 6V flashlight bulb and voltmeter to the DIN7M plug while readings on the first were taken at the PCB.

 

This Ingot measures 5.18V without and drops to 5.06V with load, both within 5% tolerance. 

[ijor]

Interesting info @BillC , thanks.

 

Btw, in the linked thread above, Jurgen @tf_hh claims that he never saw a 220 PSU behaving like the "Ingot". May be it was just by chance, but he probably checked lots of damaged 220V PSUs.

[BillC]

3 hours ago, ijor said:

Interesting info @BillC , thanks.

 

Btw, in the linked thread above, Jurgen @tf_hh claims that he never saw a 220 PSU behaving like the "Ingot". May be it was just by chance, but he probably checked lots of damaged 220V PSUs.

It could be that 220V Ingots are rare, the 110 degree thermal fuse also reduces the risk of failure due to operating at extreme temperature.

[BillC]

On 12/26/2020 at 5:39 PM, BillC said:

I decided to recheck this power supply with a load, I used a KPR113 flashlight bulb(4.8V@0.75A) and socket from an old 6V flashlight. I found that the 78S05 output that had been ~5.6V is reduced to ~5.1V. The voltage to the transformer center tap, used for the DIN7M plug negative, drops from ~5.95V to ~5.4V which still exceeds 5% tolerance. Without the voltage divider boost circuit the voltage would still be slightly high but within tolerance.

A followup, I decided to check the AC ripple of this INGOT and the result was very surprising!

~3.1VAC at 120Hz, which would actually be pulsed DC through the diodes.

 

This reading likely means the main capacitor has failed or become disconnected, these pulses could also be skewing the DC voltage reading. I don't know what the peak DC voltage is, this would easy to determine with an oscilloscope but I don't have access to one at this time.

Edited January 12, 2021 by BillC