+x=usr(1536) Posted July 22, 2021 Share Posted July 22, 2021 Received a brand-new SIDE3 on Monday. Straight out of the box, I upgraded it to version 0.40 of @flashjazzcat's firmware and installed an 8GB Kingston Micro SD card (class 10, U1) in an adapter. Everything seems to work fine in terms of cartridge loading functionality. However, the RTC seems to lose a significant amount of time - about 20 minutes per day, give or take. At this rate, it looks as though a loss of about 2.5 hours / week will be the rough average. The battery is presumed healthy as it reads dead-on 3 volts outside of the cartridge. Use is in a 600XL with a 64K internal RAM upgrade and an 800XL with both an internal 256K RAM upgrade and UAV. Both machines are powered by 5V 2.5A USB power supplies (ex-RasPi) through cable adapters from the usual suspects. Any thoughts? I'm at something of a loss to explain it. Quote Link to comment Share on other sites More sharing options...
flashjazzcat Posted July 22, 2021 Share Posted July 22, 2021 Weird. No time loss here, at least when the loader is managing the RTC. Are you observing time loss using the SDX SIDE3CLK driver, or the SIDE3 loader? 1 Quote Link to comment Share on other sites More sharing options...
Wrathchild Posted July 22, 2021 Share Posted July 22, 2021 could be a dodgy crystal? 2 Quote Link to comment Share on other sites More sharing options...
flashjazzcat Posted July 22, 2021 Share Posted July 22, 2021 Seems plausible. I was reminded by the designer that calibration settings exist in the RTC to combat drift, but I'm sure this is to pedantically account for losses or gains of seconds per day, not 20 minutes. 1 Quote Link to comment Share on other sites More sharing options...
_The Doctor__ Posted July 22, 2021 Share Posted July 22, 2021 It's worth a try though, maybe some odd or off value made it's way into it. Lots of strangeness in this brave new world. 2 Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 22, 2021 Author Share Posted July 22, 2021 5 hours ago, flashjazzcat said: Weird. No time loss here, at least when the loader is managing the RTC. Are you observing time loss using the SDX SIDE3CLK driver, or the SIDE3 loader? SIDE3 loader. To be completely honest, I haven't yet used SDX on it extensively enough to have even checked the time there. Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 22, 2021 Author Share Posted July 22, 2021 The one thing I haven't yet tried is seeing if the RTC issue happens with the machine turned on. I'm working overnight tonight, so should be able to let it run for pretty close to 24 hours before the next clock check. For reference, the cartridge is plugged into the 800XL. Quote Link to comment Share on other sites More sharing options...
ivop Posted July 22, 2021 Share Posted July 22, 2021 What's the PPM of the crystal used? With 20 PPM you should drift about two seconds per day. Quote Link to comment Share on other sites More sharing options...
flashjazzcat Posted July 22, 2021 Share Posted July 22, 2021 Datasheet here: MCP7951X-MCP7952X-Battery-Backed-SPI-RTCC-20002300C.pdf Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 22, 2021 Author Share Posted July 22, 2021 Just on the off-chance that the battery is the culprit, I'll pick one up on my way to/from work. Since I really don't have a good way to load test a CR1220, it's entirely possible that it's measuring 3V out-of-circuit but is dropping down to something the RTC doesn't like when inserted. Quote Link to comment Share on other sites More sharing options...
ivop Posted July 22, 2021 Share Posted July 22, 2021 The datasheet says: On-Chip Digital Trimming/Calibration +/- 1 ppm resolution +/- 259 ppm range So properly calibrated it would drift 1 second per million seconds, and totally off it would be 259 seconds per million seconds. Worst case is: 259/(1000000/60/60/24) = 22.377600 Slightly over 22 seconds per day. Plus or minus. But nowhere in the vicinity of 20 minutes. 2 Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 22, 2021 Author Share Posted July 22, 2021 4 minutes ago, ivop said: The datasheet says: On-Chip Digital Trimming/Calibration +/- 1 ppm resolution +/- 259 ppm range So properly calibrated it would drift 1 second per million seconds, and totally off it would be 259 seconds per million seconds. Worst case is: 259/(1000000/60/60/24) = 22.377600 Slightly over 22 seconds per day. Plus or minus. But nowhere in the vicinity of 20 minutes. Time on the RTC has been synchronised to exactly 12.00 CDT, so getting an idea of drift with the cartridge plugged in should be easy to calculate. The drift can be measured against pretty much any system clock here: each network has access to the local NTP server, which in turn syncs hourly with the north-america.pool.ntp.org servers. Firewall rules force all NTP traffic to the local server. It's safe to say that I am mildly obsessive about accurate time. Too much of my life was spent in information security staring at logs where I couldn't be certain that timestamps were trustworthy to not have that knocked into me. In any event, it's definitely weird - I've certainly seen RTCs that will drift at different rates with temperature, ones that lose less than a second per month, and ones that will lose a few seconds a day. But never anything like this. 1 Quote Link to comment Share on other sites More sharing options...
ivop Posted July 22, 2021 Share Posted July 22, 2021 (edited) 11 minutes ago, x=usr(1536) said: Time on the RTC has been synchronised to exactly 12.00 CDT, so getting an idea of drift with the cartridge plugged in should be easy to calculate. What do you mean by plugged in? I understand it's plugged in but is the Atari also powered on, or not? 11 minutes ago, x=usr(1536) said: It's safe to say that I am mildly obsessive about accurate time. I know the feeling. One of my routers used to drift very quickly. Edit: totally wrong test removed. date && ssh router1 date shows less than a second difference. Forgot my typing latency And I'm also mildly obsesive with uptimes: [243 ivo@workstation ~]$ ssh router1 uptime 19:46:16 up 670 days, 19:27, 0 users, load average: 0.00, 0.01, 0.00 Machine names are not real BTW Edited July 22, 2021 by ivop 1 Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 22, 2021 Author Share Posted July 22, 2021 Just now, ivop said: What do you mean by plugged in? I understand it's plugged in but is the Atari also powered on, or not? Read my mind and find out ? The 800XL is powered on and will be for the next 24 hours or so. This should give at least some idea as to whether or not the drift might be battery-related. 3 minutes ago, ivop said: I know the feeling. My router would drift very quickly. Yup. That's one of those things that I go ballistic on as soon as I see it happening. 4 minutes ago, ivop said: And I'm also mildly obsesive with uptimes: Nice 1 Quote Link to comment Share on other sites More sharing options...
ivop Posted July 23, 2021 Share Posted July 23, 2021 (edited) BTW I did not yet take into account the clock source or oscillator (not sure which one SIDE3 uses). That one could also be off by a maximum of +/- 20ppm, although most crystals are way better these days, down to +/- 1ppm. Still well below half a minute per day (24.1 seconds max in total). Perhaps the crystal (if that's used) sometimes has hicups and misses a beat. Does it have the right capacitors? (6-9pF, higher is discouraged). Maybe one is not making good contact with the solder pad? Edited July 23, 2021 by ivop spelling 1 Quote Link to comment Share on other sites More sharing options...
_The Doctor__ Posted July 23, 2021 Share Posted July 23, 2021 (edited) I vote for the perfect storm... check components and joints... then settings/adjustments... an oops here or there can can be cumulative. Edited July 23, 2021 by _The Doctor__ 1 Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 23, 2021 Author Share Posted July 23, 2021 (edited) Having just checked the SIDE3 after it had spent 29 hours straight plugged into the 800XL with the power turned on, two things stood out: Either the loader or the computer crashed. When I turned the TV back on to check on it, it had gone into the screensaver but it was frozen and there was no response to the joystick or keyboard. After power-cycling the machine, the clock was 26 minutes slow. I do have a CR1220 battery sitting here waiting to go into it, so I'll start with that since it's probably the easiest thing to begin with. Given that the cartridge was receiving power the entire time, I'm not convinced that this is necessarily the problem, but it is at least a starting point. 7 hours ago, _The Doctor__ said: I vote for the perfect storm... check components and joints... then settings/adjustments... an oops here or there can can be cumulative. Entirely possible. I'll give it another 24 hours and see where things sit with the new battery. @ivop, I'm also keeping your idea re: crystals and capacitor values in mind. Edited July 23, 2021 by x=usr(1536) 1 Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 23, 2021 Author Share Posted July 23, 2021 (edited) Battery's in. Time of RTC reset: 17.37. The cartridge is back in the 800XL, which is powered on. Edited July 23, 2021 by x=usr(1536) Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 24, 2021 Author Share Posted July 24, 2021 (edited) A little over 24 hours later, the RTC is keeping accurate time. If there has been any drift, it's not noticeable in the loader. Regarding the loader: I can't conclusively say that the battery was the problem because removing the old one also reset the NVRAM, which could also have been a factor. However, it does appear as though the issue was related to something in the vicinity of those two items. Next test will be to unplug the cart for 12 to 24 hours, and see if the RTC is still OK. My suspicion is that it will be, but given that it's been powered on continuously I'd like to see for myself that it is. Edited July 24, 2021 by x=usr(1536) 3 Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted July 25, 2021 Author Share Posted July 25, 2021 Aaaand I think we're good. Just tested the SIDE3 on the 600XL, 800XL, and 1200XL (via cartridge port extender) after it sat for about 15 hours without being plugged into any system. If the RTC has lost any time, it's negligible. Still no way of telling if it was the battery, NVRAM, or both, but everything seems OK. 2 Quote Link to comment Share on other sites More sharing options...
flashjazzcat Posted July 25, 2021 Share Posted July 25, 2021 Good. So we can forego calibration settings in the clock drivers, then, entirely as expected. 2 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.