User avatar
Zippy Zapp
CA, USA

Posted Sun Feb 05, 2017 11:01 am

That is essentially what you would do with a phone battery pack only it would be rechargeable and the proper voltage. In that picture I don't see a diode and resistor to negate the charge circuit though. I wonder if it is on the backside of the pack?

That is pretty much how I wired my 486 and my older Macintosh Performas/Power Mac that took the 4.5 square battery. Only I used a 3 AAA holder to get 4.5 volts. Of course the Macs don't have a charging circuit to worry about.

User avatar
intric8
Seattle, WA, USA

Posted Sun Feb 05, 2017 11:14 am

I suppose you could use rechargeable AA batteries, too.
In that picture I don't see a diode and resistor to negate the charge circuit though.
I don't know about that. Here's the listing I saw for it.

I ordered a multi-meter today, either way. I hope to get to the bottom of my issue one way or the other (even it if isn't fixable).

User avatar
intric8
Seattle, WA, USA

Posted Sun Feb 26, 2017 6:27 pm

I pulled out the new Multimeter today and learned how to test DC voltage. I got the battery out and it was fine (unfortunately - I was really hoping the guys in the UK somehow sent me a dud when I got the new coin battery component).

Oh well. I think that pretty much means I've got a bad trace somewhere, because SysInfo is able to find my IC just fine and thinks it is working.
So, if the battery is OK and assuming the little component I got for it works (seems very likely it does) then one or more of the traces that lead from the battery posts to the IC must be bad.

I need to try and hunt down a tutorial on how to bypass the traces altogether and just use wires. I feel like I've seen that somewhere before but can't remember where I saw it. It would only be 1-3 wires and a little hot glue. Anybody else seen that anywhere?

User avatar
intric8
Seattle, WA, USA

Posted Thu Mar 16, 2017 8:24 pm

UPDATE: I got it working!

After working on a completely different hardware project - adding an SCSI2SD card with 4GB of RAM - I discovered what was going on and how to fix it. I think it is entirely a KS/WB 1.3 phenomenon that was rectified in future versions, since I can't seem to find others online really talking about it. I'll explain soon!

User avatar
Zippy Zapp
CA, USA

Posted Fri Aug 17, 2018 9:42 pm

After reviewing the datasheet for the OKI RTC chip M6242B, which seems to be a common IC found in many of the Amigas I have seen, a CR2032 battery should be fine for a decent amount of time. The datasheet states that battery backup is still enabled down to 2.0 volts.

For those that have installed the CR2032 mod, how long has it been since you first installed? I am curious to know how long one of these batteries can last in the Amiga. I know in the PS2, mine are still original from 2000. 17 or 18 years is amazing.

User avatar
jdryyz

Posted Fri Jan 11, 2019 10:01 am

Speaking of clocks...

Does anyone have the Y2K compliant version of SetClock? From what I have been able to find, it is included in the WB1.3.3 v34.34 adf bundled with Amiga Forever. Does this mean there is no freely available version of a patched SetClock program?

User avatar
jdryyz

Posted Fri Jan 11, 2019 10:38 am

I must partially take back what I said. Having just replaced my Varta battery with the AmigaKit battery holder, I was able to reset the clock, set the proper date and time and save the data. It seems the correct date/time is still there after I powered it off! Yay! PCB was not as damaged as I thought.

The part that I am taking back is about Y2K compliance. I noticed the correct day of the week showed up after setting the date so it must understand that 19 is not the year 1919. I was under the impression Workbench 1.3 was not Y2K compliant at all. Further research indicates it is partially compliant and will still have trouble some thirty years from now? Heh heh.

I think I will be on Workbench 2.x or even 3.x by then. :lol:

User avatar
intric8
Seattle, WA, USA

Posted Fri Jan 11, 2019 12:17 pm

Yeah, I didn't respond to your initial question because all of my clocks work just fine. But there is a point in the future - as you said, in a few decades - where that will cease to be the case. But that's where my A1000 clocks have been for years. Most of those clocks were not written to work correctly beyond 20 years. But the clocks still "work" just with the wrong year.

So, I can see the time of day, and it'll stamp dates with unique dates and times. The years will just be off. I'm not using any Calendar programs with my A1000's so it doesn't bother me. (mostly)

User avatar
Shot97
Detroit, MI, USA

Posted Fri Jan 11, 2019 4:33 pm

Latest version of Workbench 1.3's setclock http://obligement.free.fr/files/setclock_v34.3.lha

I believe all Amiga operating systems are capable of understanding the correct date until the year 2078, that's when it reverts back to 1978. However, Workbench 1.3 has a problem "setting" the time after the year 2000, where it can be slightly or largely off, this is where the updated setclock is required. Now I don't have a 1000, so I lack real word experience with that exact model, however until someone explains better why a 1000 would have issues with the exact same operating system a 500 uses, like Workbench 1.3, I'm inclined to want to believe even Intric8 could have the proper date and time, that there is a way for that.

The brains of the operation are not in the real time clock boards. All the boards do is keep track of years from 0 to 99. They were very simple designs in that aspect, it was the Amiga itself, rather Workbench, that took those simple 0-99 digits and translated them to the correct year and time. This is what makes me get a little wide eyed when Intric8 notes issues beyond the year 2000, as at first glance it makes little sense why a 1000 with Workbench 1.3 would not be able to translate the simple 0-99 in terms of the years that the clock board gives it. The Amiga would have had a serious Y2K issue if it had not been based on Unix originally, which in the case of the Amiga, set "year 1" as 1978. That's where the PC had its issues, the clock boards themselves were of little difference compared to the Amiga, the PCs also kept track of years 0-99, only in the case of DOS/Windows at the time, "year 1" was 1900, which meant the rollover was going to happen in the year 2000, where it's 2078 with the Amiga due its year one being 1978. DOS/Windows also were programed to show dates in the exact same way they read them from the clock, which I'm sure is the real reason why "year 1" is 1900. DOS showed the year as 85/89/94, possibly even Windows 95 was showing dates in that way. The Amiga on the other hand, while it may have at times displayed the year in two digits in programs or even in the operating system itself, always understood the year as 4 digits. It took the two digit the clock board kept track of, did some math with the loadclock command at startup, and from that point on it was 1988, 1995, and even 2019.

Unless the 1000 clock boards were even more simple than the other Amiga models and only kept track of digits from 0-30 or something similar. That sounds possible but not exactly logical. The 100 year roll over was a digit limitation in the cheaper boards used. The Amiga could very well keep on trucking on and understand it's 2178 if the clock boards did not have that two digit year limitation. The next logical digit limitation prior to 100 years would be at the 10 year mark, so a 1000 would stop knowing the correct date in 1988 if that was the case, and that would of course make no sense as it was still being sold then. It wouldn't make sense for a cheaper clock board to only count 20 to 30 years before rolling over, it would have to pass a digit limitation. The PC market was simply forced to deal with these clock counting limitations sooner than they should have because it was decided that year 1 would be 1900, 78 years before personal computers were a thing, But it was not the PC clock boards, it was DOS, it was Windows. Workbench was simply more intelligent with how it dealt with the cheap two digit clock board limitations, programing year 1 as 1978, when PCs were actually a thing! But the Amiga itself, Workbench, rater smart about this whole situation, which makes me wonder how a 1000 could have issues the other models do not. No real world experience there, so I could be wrong, but I know if I had a 1000 it would take me a long time to give up on that one.

Despite this, all Workbench versions will stop being able to load the correct date in the year 2046. At that point, it's problems with the loadclock program in all versions of the operating system that will cause the Amiga to mistranslate the clock board digits, being randomly off by a little or a lot. All Workbench versions will have this issue in 2046, and as far as I know, Hyperian does not not seem to be working on that issue with new versions... So as long as you never turn off your Amiga after 2046, you'll be good until 2078, but 2046 is the effective useless date for the Amiga, which is actually rather mind blowing considering all the issues the PC was having back then..

Though a badly written program could very well mean you're best off to still keep the date prior to 2000. The Amiga may have been able to understand the correct date, that does not mean a program necessarily does, which might cause issues.

User avatar
intric8
Seattle, WA, USA

Posted Fri Jan 11, 2019 6:14 pm

until someone explains better why a 1000 would have issues with the exact same operating system a 500 uses
Because it has nothing to do with the operating system at all. The issues are the janky clocks and their drivers.

My current clock was made by Microbotics way back when. And their software - the drivers - don't support going beyond 2008. Their code simply doesn't support it, and trying to make it current makes it crash and not save. I have a few others that I'd like to test eventually, but most of them are broken.

On the 1000, some of the clocks are so old, the OS can't see them properly. They had custom software that you had to use (and call in your startup sequence) to activate and use. Not Workbench's problem at all, and not really Workbench's fault, either.

My A2000's run everything just fine.

In theory, someone could write new drivers for some of the really old RTCs, but that's beyond my personal skillset.

I'll put it this way - if you use the standard (A1000 specific or beyond) Workbench lines in SS to set the clock, it won't even find it. For these really old clocks, you have to use programs that shipped with the clocks, and you have to hit those programs during launch. Not a typical situation back in 1986 on an A1000, as most didn't have hard drives, either.

I have one setup that - back in the day - assuming you owned 1 of 20 possible hard drive configurations, you could use special Microbotics software to write either a customized Workbench disk, or hard drive setup with custom Workbench, where it would pack its clock software into the thing for you. But unless you have one of those ancient hard drives, it doesn't work at all. The best route (and I'm not the only one who has done this) is to install it to a hard drive and just point to it in the SS. Works well enough for what I want it to do.





Return to “Hardware”