Another U1MB Bites the Dust

[+mytek]

I'm reminded of Queen's song: Another One Bites the Dust

 

And although I love that song, in my case it's an all to familiar occurrence in my relationship with the marvelous Atari product called the Ultimate One Megabyte (U1MB).

 

Disclaimer: The U1MB is one of the best upgrades you can do for your Atari 8-Bit bar none. And the quality of this product as it comes from Lotharek is truly the best, as well as @flashjazzcat's custom U1MB BIOS. No the real reason that so many of these have died under my care is entirely due to user error as in letting me get anywhere near them. Well the reason for the 4th board dying is still to be determined.

 

A couple of days ago the U1MB in my Monster 600XL died with the Xilinx CPLD being hot as a pistol, and the normally 3.3V supply running down around 1.78V due to the excessive load on the regulator being too much for it to keep up with.

 

 

This makes for the 4th board where I ended up with a fried CPLD. The first three are 100% entirely my fault, being the result of accidentally applying wrong voltages in various spots - usually into one of the accessory control pins.

 

 

Now being known as the U1MB Killer doesn't make me feel particular warm and fuzzy, but I've learned to live with the reputation, likely being the one and only person to send so many U1MBs to their death. However if this was WWII and I was a Flying Ace I'd be off to a great start 👍 👍 👍 👍 (4 thumbs up for sure).

 

But as I mentioned earlier only the first three can I take total credit for causing their demise, with the 4th being kind of a mystery to me, although i guess all things come to end at some point in time.

 

So why was I so careless to cause so many U1MB deaths?

 

It's actually quite simple, I'm an Atari 8-Bit developer that likes to integrate the fantastic functionality of the U1MB into so many of my alternative Atari 8-Bit Mother Board Projects, and R&D can have deadly consequences. Usually it just comes down to forgetting to power off the system when moving wires or setting up to probe a signal. And sometimes it would happen when I had plugged something in with the header being off by a few positions. And in another case I was using the wrong voltage power supply applying 9 VAC instead of 5 VDC. This last case required a slight redesign on the 1088XEL to use a different size barrel connector than the one commonly used for Atari disk drives and the 1200XL (funny how I'd have one of those 9VAC power bricks in the vicinity???).

 

Certain Death

 

I also discovered early on to never ever apply a source of 5V to any of the U1MB Control Signals. Although that lesson came a bit too late for one poor unfortunate soul (RIP).

 

Why did the Last and Most Recent U1MB Die?

 

Yes that is the mystery which brought me here today. I had built my Monster 600XL a few years ago and it had become my daily driver, with me turning it on almost each and every day to check something out or to flash a few PIC chips via the JOY2PIC adapter. But after being gone for a week, with the unit sitting in an off state, I came in a few days ago and turned it on to test a TBA R-Time 8 Cart to see if its battery was still good. And that's when I was met with the dreaded RED screen of death. WTF!!! How could this be? I had just used this unit a week ago without issue.

 

So I opened it up and took a look inside...

 

 

Nothing looked out of place. Yes I know there is a lot going on inside this thing, but after all I am quite the tinkerer so no surprise. It actually has three of my upgrade boards in play.

 

I checked all the ribbon cables going underneath the U1MB to see if something got pinched and pierced, but all was good. And as can be seen none of the U1MB Control Signals are even being used, so no chance of foul play there. Also checked the wires from the TK-II-PBJ board that are snaked underneath, and saw no evidence of anything shorting out to a pin under the U1MB.

 

I also checked out the R-Time 8 board I had been wanting to test for anything amiss, but everything looked good and it worked just fine in another machine. Besides as far as I could tell there would be nothing one could do from the cartridge port that would be the cause for the U1MB CPLD frying. Well other than jamming some 120VAC power leads into the cart port, which I try not to do

 

So what we have here is quite the Sherlock Holmes Mystery as to what took out the U1MB's CPLD

[Geister]

I'm sorry for your loss.  However, if it provides the community with a bit of knowledge that helps prevent further deaths their lives will not have been lost in vain.  

 

I do understand feeling jinxed with things like this.  My own son was/is known as the angel of death for tires.  At least he's getting better, he now knows to stop driving when he notices the tire has gone flat.  There's at least some forensic evidence left when I see the tire & wheel.

[+scorpio_ny]

How hot does the Xilinx CPLD get under use? If it gets really hot, it's positioning is not ideal for heat dissipation. Could it be possible is suffered a thermal death?

 

[+mytek]

26 minutes ago, scorpio_ny said:

How hot does the Xilinx CPLD get under use? If it gets really hot, it's positioning is not ideal for heat dissipation. Could it be possible is suffered a thermal death?

That's a good point, but I don't really know how hot it gets if lets say the board was mounted upside down with nothing above it but free air. Obviously with the way I've got it mounted and all the excess ribbon cable folded up underneath air flow would not be ideal. So perhaps it ran hotter because of that and over time it slowly degraded until it finally couldn't take it anymore.

 

Also having it mounted the way I had it, the battery is totally inaccessible.

 

When I get this U1MB board repaired, I will investigate the idea of flipping it over and see if everything (e.g., keyboard) clears.

 

EDIT: Speaking of repairs, does anyone have either a Digi-Key or Mouser link to a replacement CPLD?

[+CharlieChaplin]

Another U1MB bites the dust...?!?

- We should send you more C64 computers...  😁

 

"And you tell me
Over and over and over and over again, my friend
You don't believe we're on the eve of C64 destruction
No no, you don't believe we're on the eve of C64 destruction."

 

[original lyrics: Eve of Destruction by Barry McGuire]

 

[flashjazzcat]

1 hour ago, mytek said:

EDIT: Speaking of repairs, does anyone have either a Digi-Key or Mouser link to a replacement CPLD?

I just buy them on eBay from China.

1 hour ago, scorpio_ny said:

Could it be possible is suffered a thermal death?

All the U1MB boards in 1088XELs (of which I have one) are mounted upside down, and this never seems to cause a problem.

 

In any case, boiling hot CPLD is usually an indicator that it's toast.

[+mytek]

32 minutes ago, flashjazzcat said:

I just buy them on eBay from China.

Any preferred sellers? maybe a link if you have it

[_The Doctor__]

If you find em order an extra for me ;).     I'm not saying there's murdered one here, but rather it died of natural causes.

Edited August 28, 2023 by _The Doctor__

[+mytek]

2 hours ago, _The Doctor__ said:

If you find em order an extra for me ;).     I'm not saying there's murdered one here, but rather it died of natural causes.

Sounds good Doc - will do.

 

I believe I still have your address info unless you moved since the last time I sent you one of my dead U1MB boards. PM me if it changed.

[TGB1718]

15 hours ago, mytek said:

So what we have here is quite the Sherlock Holmes Mystery as to what took out the U1MB's CPLD

[flashjazzcat]

13 hours ago, mytek said:

Any preferred sellers? maybe a link if you have it

None, and I don't have a link since the last listing I ordered from no longer exists. Here it is, though:

I found that by scouring eBay for the correct part number.

[+Larry]

Here's a related question that I found no answer for -- what is the general life expectancy of a good brand (such as Xilinx) CPLD?  I have read that an FPGA's life is estimated at 10-15 years.  I'm just curious because our 80's hardware has lasted for 40+ years (mostly intermittent use).  I would presume that in general, lower voltage mean less heat and longer life?  Are there any that now run in the 1+ volt range?

[flashjazzcat]

It's a good question, since I have several VBXEs and U1MBs well over ten years old (one VBXE from 2009, for instance, and Incognitos and U1MBs from 2011 or so, some hand-assembled), and they all still work. The bigger problem seems to be sourcing long-deprecated chips which don't arrive with bent or oxidised legs.

[TGB1718]

I had an ICD interface for my ST, had it since the 90's it has 4 GAL chips on it and it works fine, I no longer

have it as I sent it an AA forum member who has reverse engineered it and several other devices for preservation.

 

So maybe older ones have a longer life ?

[Rear Admiral]

Maybe a small heatsink could be of some help.

[_The Doctor__]

all parts except crystals electrolytic caps etc getting ultrasonically solvent cleansed before board stuffing and assembly would be helpful

Edited January 31 by _The Doctor__

[+mytek]

@_The Doctor__ a while back I did get some replacement CPLD chips (several) and they do look to be legimate,. However my abilty to solder these to a board is not good and I ended up tossing the board aside after bending some of the pins when trying to undo some solder bridges. I'll try again when I find myself in a better mood to take it on. Anyway are you still in need of a couple of these? If so I can send them to the address you gave me many years ago where I sent you a U1MB that I had also killed.

 

If your address changed, PM me the new one.

[+bob1200xl]

How long does a CPLD (or any semiconductor) last?

 

140 years? 8000 years?

 

There isn't anything I am aware of that 'wears out'. Thermal cycling may stress mechanical connections and cause a failure internally in the chip. Exceeding design limits may 'blow' the chip. Other than that, nada....

 

Chips that store configuration data in cells (EEPROMs, CPLDs, GALs....) will slowly leak their active particles into oblivion and eventually 'fail'. Only, they didn't fail, their 'battery' just ran down. Data retention is something like 10 years.

 

you pay your money and you take your chances....

 

Bob