U1m newbie. Why the clock does not run?

[manterola]

I updated the U1M right away after installing it to my new 600xl. I follow the instructions and  latest FJC firmware. I also followed very closely FJC videos. By pausing the video I noticed that the RTC was running in his 600xl. 

Is this a problem with my U1M or I have not done something? 

[+DrVenkman]

It’s hard to know if you’ve done something wrong or not without photos. There’s an outside shot the battery is dead on the board, but not very likely. They last for years. 

 

Post some good, in-focus photos of your installation and chances are we can find or rule out any installation issues right off. 

[_The Doctor__]

not only that, but make sure you are using the proper RTC driver for yours

[flashjazzcat]

RTC driver is totally irrelevant if the clock is not running in the setup menu (where no driver is required). Check first that the coin cell battery is not flat and that is correctly inserted (i.e. that the two metal fingers are not squashed underneath the cell). One that's eliminated, you have an installation problem or (highly unlikely) a dead RTC chip.

Edited May 16, 2020 by flashjazzcat

[manterola]

Thanks all, I'll check the battery and post pictures of the installation. But, I remember using Sparta in the 89 or so and using TIME to set the time and even without RTC the time was running, but obviously lost after a reset, so last night I tried setting the time in the SDX and it was also set but stuck. 

Anyway, I'll recheck everything, the rest looks like is working fine, I updated twice, just in case last night.

Thanks

[_The Doctor__]

didn't realize the conversation was about what's happening in the menu... oh well.. such is life. It's interesting in that the follow up shows stalling in Sparta as well though, it must be polling the RTC ever so often then. So a battery it may be. If the clock is still dead, using other config and driver to get around it, I tested the theory using RT8 and it works even if u1m is dead same for SuperSDX cart.... which I will see if a conflict will occur as the clocks may occupy same space, SDX cart vs u1m...

might be good to have an RTC table for all our various gizmos and carts...

Edited May 16, 2020 by _The Doctor__

[manterola]

OK, this is not looking great. I first removed the battery and checked the voltage with and without charge. It looks OK. I also checked the contacts I the battery holder.

Once I reinstalled the battery, all my settings were gone (as you might expected). 

Then I checked the continuity in the connections  between the p2 connector and the Sally pins. I discovered that the two connections: phi2 and RW to the two vías close to the PBI are NOT directly connected to the Sally pins...(!?!? Are they buffered? ). In my pictures those are the cables yellow and black respectively. 

The next two signals I connected close to the antic chip do have direct continuity to reset and halt pins in the Sally chip. Those are the red and blue cables respectively. 

So my next experiment was to move the yellow and black cables next to the Sally chip. That did not make any change, the system clock in the ultimate 1M conf screen is still stuck. I can change the time and date, and other settings, everything is saved and survive many power cycles, but the clock does not move. 

I might try to move all my connections now to the connection points close to the upper part of the Pcb as recommended by FJC. I will also check the ribbon cables for continuity... 

Any other suggestion? 

BTW, this is a NTSC Chelco 600XL

Edited May 16, 2020 by manterola

[_The Doctor__]

what are the chances a complete redo of the cpld and flash will fix the interaction between the clock and firmware... if everything else is working? maybe somethings borked bitwise in the works? reprogram from the ground up so to speak.

Edited May 16, 2020 by _The Doctor__

[flashjazzcat]

Are BIOS setup menu settings saved (surviving a power cycle)? The settings use the same NVRAM as the clock.

[manterola]

Yes, all my settings are saved. But the system clock is still at the same time I configured 2 hours ago

Is it OK that the connections points you mentioned in your video for phi2 and rw close to the PBI are not connected to the Sally pins? Are they buffered, maybe? 

Edited May 16, 2020 by manterola

[_The Doctor__]

the real time clock runs even when the Atari is off... so phi doesn't matter... if the Atari was holding the real time clock from changing, turning off the Atari would allow the real time clock to change, or if there were some wiring that could cause that... disconnecting the upgrade and then re inserting it later would allow the time to tick along... and you would see the time had changed during the period of time it wasn't in the machine....  You are left with something wrong on the upgrade pcb, one of it's chips, the programming of the cpld, or flash. Not much else going on there.

Edited May 17, 2020 by _The Doctor__

[flashjazzcat]

8 hours ago, manterola said:

Is it OK that the connections points you mentioned in your video for phi2 and rw close to the PBI are not connected to the Sally pins? Are they buffered, maybe? 

They're buffered. The machine wouldn't even boot if the connections were wrong.

 

So: if your settings are being preserved, we know the battery works and that the computer can communicate with the RTC chip. Next thing to suspect is the clock crystal on the U1MB, soldering on the DS1305, or some problem with the chip itself.

[+tf_hh]

2 hours ago, flashjazzcat said:

So: if your settings are being preserved, we know the battery works and that the computer can communicate with the RTC chip. Next thing to suspect is the clock crystal on the U1MB, soldering on the DS1305, or some problem with the chip itself.

 

I would agree. If BIOS settings are saved and hold, then the CPLD can´t be the culprit. Also the DS1305 RTC works fine (address and data storage logic). Take a look with a magnifier at the little small crystal:

 

 

Check if one of the thin wires from the crystal is broken. And then check the bottom side here:

 

 

It´s a long time ago, but I have two times got U1MB´s delivered from Lotharek, where one of these capacitors was missing. It looks like they were soldered, but not long enough, or a cold-joint, whatever. When one of these capacitors is missing, then the RTC didn´t get it´s needed frequency for the internal counter. So everything works, but the time will not go on.

 

[manterola]

I checked the crystal connections to the capacitors and to the pins 3 and 5 of the ds1305, everything is correct. I applied a bit of solder to the crystal and capacitors just in case. 

Capacitors are connected to ground in the other side, so that should be right. 

I also checked continuity between every pair of legs of the ds1305 and inspected using a magnifier glass, everything is correct. I don't know how returns work, but it sounds complicated, given the current state of affairs.

Maybe I should try a new crystal since it look easy to replace given it is a thru-hole component. 

Not sure, but it look like everything else is working, the only other weird thing that happened was that uflash crashed when I tried to update side2 using the sio2pc (it get stuck in the roms file selection). 

Ah BTW, yesterday I also tried downgrading the firmware, the SDX, upgrading again, removing and replacing the battery, reload factory settings, etc. But at this point everything point to the crystal or the ds1305... Right? 

 

[flashjazzcat]

What with the unrelated crashes, unless you have a serious stability issue or some weird problem with the ribbon cables, I start to wonder if it's time to send the board back. I received an U1MB a while back and the clock just displayed junk. There was also a missing capacitor thanks to a rare manufacturing error. Lotharek handled things superbly and I had a replacement in days. So: problems do happen, but they are rare.

 

Perhaps it's time to send Lotharek an email.

 

[manterola]

Finally I realized the Crystal was the source of the problem: as soon as I removed it, the clock seconds started to advance. The advance was first  like 2 seconds, but by touching the board the seconds keep advancing, probably by picking up noise. there was point in which the clock advanced by itself without crystal every 5 or 6 seconds (very slow). Then I connected a new (old) crystal (same frq) I found in a box of random electronic stuff and the clock worked immediately, even without the load capacitors which I removed to see if there were the source of the problem (I replaced them after the crystal was in place).

So, I did all of this because returning this to Poland looked very complicated and time consuming and I wanted to give it a last try. 

In conclusion: a happy ending.

 

[_The Doctor__]

Glad you explained exactly how you fixed it and what the fault was, since you're not the first to have this happen, either the crystals are defective at install or damaged in shipping... either way lotherik can act on it with a little burn in test, giving a couple tap tap taps or change his shipping to pad the crystal/pcb a bit better. For the rest of us... we will know where to look first... TSB

[flashjazzcat]

3 minutes ago, _The Doctor__ said:

since you're not the first to have this happen, either the crystals are defective at install or damaged in shipping... either way lotherik can act on it with a little burn in test

Have there been multiple instances of Ultimate 1MB boards shipped with faulty crystals?

[_The Doctor__]

With clocks that didn't work yes... with the crystal being bad?  Some people just write it off using it without the clock or ship it back. He took the time to run it down himself, so I thank him, add a note, and when another shows up... this will be one of the first things to check. Although I must admit, I've done as others and touched up solder joints (needed or not) and not checked the heartbeat for the clock. So it's good to have more information and to act on it.

Edited May 26, 2020 by _The Doctor__

[flashjazzcat]

13 minutes ago, _The Doctor__ said:

So it's good to have more information and to act on it.

Certainly. I just sent Lotharek a link to the thread.

 

[lotharek]

30 minutes ago, _The Doctor__ said:

With clocks that didn't work yes... with the crystal being bad?  Some people just write it off using it without the clock or ship it back. He took the time to run it down himself, so I thank him, add a note, and when another shows up... this will be one of the first things to check. Although I must admit, I've done as others and touched up solder joints (needed or not) and not checked the heartbeat for the clock. So it's good to have more information and to act on it.

actually, i rememer only one mail reporting faulty crystal - recently....

so, if no reports to me in past, no actions were taken..

 

ps. tap tap in crystal ????? burn in tests of rtc ? lotharik ?

 

 

 

 

[_The Doctor__]

burn in, just leave it running live in a machine for some hours make sure the clock is working, tapping in and around the crystal when initially done setting up.

I'm not the first to encounter it, as it looks like FJC, manterola and myself have had an occurrence. Having been mentioned by at least 3 others over time and manged to deal with one other. I'm glad for the heads up.

 

My thought was if you didn't observe a problem during burn in, maybe the crystal was intermittently working and during shipping would fail further. So it's definitely something to check. Parts aren't always the freshest or best these days... depending on where they originate.

[lotharek]

44 minutes ago, _The Doctor__ said:

burn in, just leave it running live in a machine for some hours make sure the clock is working, tapping in and around the crystal when initially done setting up.

I'm not the first to encounter it, as it looks like FJC, manterola and myself have had an occurrence. Having been mentioned by at least 3 others over time and manged to deal with one other. I'm glad for the heads up.

 

My thought was if you didn't observe a problem during burn in, maybe the crystal was intermittently working and during shipping would fail further. So it's definitely something to check. Parts aren't always the freshest or best these days... depending on where they originate.

i really appretaiate that suggestion "just leave it running live in a machine for some hours make sure the clock is working" will not work for me.. making batch of 100 ultimates would take 100X2 hours--> 200 work hours -> over a month. sorry, cannot be done. clock set, config set, power off and after few seocnds power on... unless ultimate price will 5 times higher.. sorry -not possible..

 

since 2012 i have produced much more than 100 ultimates. if there is issue, i try to get it solved in next batch. but faulty crystal is not design issue - it is part related. no change to design can fix that. shit just happens

 

tapping around crystal ? ... i will leave that uncommennted.. 

that gives only finger pain after tappin 100 ulitmates 3 times...

 

 

Edited May 26, 2020 by lotharek

[_The Doctor__]

Many people leave a machine on while doing other work or overnight after fixing them or making an item, they may even turn items off and turn them on...checking them right before they ship them out as just about anyone I've known does the same that make home-brew items, fixes or sells stuff... it doesn't run up any cost, well maybe a penny in electricity and they get sleep while it runs overnight.... but certainly not any real extra time. Why even a video our dear flashjazzcat tests his work for a bit, some mention of it.... sometimes longer sometimes less... It made me feel more confident in his work.

 

I did state that the problem was the part... as I've bought chips from china with a higher failure rate... and right now chips that haven't even arrived.

 

I quick vibration or tapping of a board won't kill anyones fingers, but any vibration source would do. diaphragm pumps, and mechanical buzzers work great, if a little jostling or tapping by hand would be a problem, I've got friends with carpel tunnel issues and just typing is painful for them sometimes. But if we are going to be funny, there always a cheap and effective way of doing it... again it wouldn't cost nearly anything.

 

Yeah my tappy finger hit an I instead of an E.... kind of funny actually.... I agree it already costs more than enough to get one, so I hope the cheap and easy way hobbiests and factories have been doing this sort of thing effectively over the ages is useful.

 

I really appreciate that you've taken some time to look into the issue. It really is a nice bit of kit. Good stuff.

Edited May 27, 2020 by _The Doctor__