32k RAMPAK

[Martin]

Andrew you are correct they are not as 'volatile' as these posts would have you believe but as with all things 'computers' if you haven't backed it up and it crashes - well we have all been there! I have one of Olivier's 256K packs and I've never had a problem with it, but I still back it up when I've added something new. I have two 32K ones that I acquired from Peter and you put in the replacement battery adaptors. The one I use for 'development' resizes itself sometimes while I'm plugging in the RS232 Comms Link (never with the USB one). The one that is in my workshop LZ64 to back up the memory A: I've never had any trouble with but I hardly ever plug in the comms link.

Interestingly PSION knew there as an issue as they put this 'warning' on their information leaflet supplied with all the packs.
.

RamPackLeaflet - Warning.jpg

.
Jaap has 'result' of 36 as ROM version 2.4 (CM/XP)

Sincerely
Martin

[Lostgallifreyan]

The main problem with RAM packs is that the backup battery has a low internal resistance, and I think the protection against reverse bias or sudden discharge (I assume this exists) wouldn't help to prevent anything that makes the main power line voltage drop from causing a sudden drop that causes corruption and loss of data. Switching on hungry devices like older GPS modules directly wired to the top slot can do this, and even plugging and removing a RAM pack when the Organiser is powered can do it. (I think the manual advises against doing this).

I think some people have tried adding a resistor and capacitance as LPF on the RAM pack supply to protect against the risk, but I didn't do it, I just avoid using RAM except as a fast access to things I'm working on, or non-critical logging. So long as I don't remove packs or let the main battery weaken, the RAM on a pack is as safe as the internal RAM, and I extract data via a powered comms link if I need to rescue data from an Organiser I think might be about to lose power. This was one of the main reasons I wrote ORG-Link. I wanted a tool I could use quickly when I had other things on my mind..

I didn't know about the deletion of the MAIN file and its effect on a RAM pack, but I did notice that deleting the internal MAIN file can cause interesting weirdness with SAVE and FIND.

[MartinP]

That missing link in the schematic between pins 1 and 2 of IC5A is only on the version from Jaap's site. The copy in Martin's Technical Ref. Manual is ok. I guess Jaap got an early version from Olivier, the corrected one came from Olivier's site, so must be a later one.
Martin P.

[Lostgallifreyan]

Whatever the reason for a RAM pack getting an unexpected 'resize' and data loss, it's worth considering how to prevent it. The obvious advice of 'backup' is useless if the act itself causes the data loss! If the internal battery is too low, and the backup battery on the pack is also low (likely, given human nature, time, and circumstance), then power MUST be applied safely before any backup can occur, and plugging a pack to put the backup on it might be impossible without causing the event that triggers the data loss. If it does not, it might anyway the moment you try to write to a Flash or EPROM pack.

The risk is so complex, so hard to solve, and the stress of trying to avoid it is bound to hit when you most want that NOT to happen.

I've considered this at length, and come up with what I think is a definitive safe sequence of events.

First, consider a few things:
1. If RS-232 is plugged while powered, there may be damage, general advice is NOT to do this. Do not power any device till the line is firmly connected.
2. Even if damage does not occur, a PC will often cause a control line to wake up the Organiser the moment the link is plugged in. If the onboard power is very low, this might be enough to destroy RAM pack data.
3. Some Comms Links have 4 pins shorter than others, aimed at reducing risks like these during plugging and unplugging, but not all! Those based on new USB circuits and built into power adapters definitely do not have this protection.

----------------------------------------------------------------------------------

There is a safe sequence, and if it really matters to you, stick to this, rigorously:

Do NOTHING to the onboard packs. Leave them as they are.

With the Organiser AND the PC switched off entirely, connect the serial link. I doubt it matters which end first, but PC-end first gives you a system ground right from your first move. This could help reduce risk from small static discharges.

Connect power to the Comms Link, then start the PC. If this does not switch on the Organiser too, switch it on yourself from the keypad.

Launch ORG-Link and read the RAM pack to a file on the PC.

At that point you have safe salvage and can do anything you want with it.

----------------------------------------------------------------------------------

This was not just a plug for ORG-Link. There is a specific reason for this advice. UNMAKE can only read a pack in Slot C. That sucks if your RAM pack is in Slot B because the LAST thing you want to do in this situation is swap packs, the RAM will probably die if you do that. As far as I know, ORG-Link is the only tool ever written that lets you choose the slot to read from. For critical salvage, that matters.

[amenjet]

3. Some Comms Links have 4 pins shorter than others, aimed at reducing risks like these during plugging and unplugging, but not all! Those based on new USB circuits and built into power adapters definitely do not have this protection.

Some of the newer USB adapters do have the longer pins. I have salvaged connectors from old Psion devices to use on newer USB adapters, and I think Olivier might have done this as well. So there are some with this feature, but when the supply of devices to salvage from runs out, there will be no more.

It's also worth mentioning that while the RAM pack is in the organiser, it is powered by the organiser, so should keep its contents even with a flat or no battery.

Andrew

[Lostgallifreyan]

True, and I often use a RAMpack just to test something, and an old one with no battery is guaranteed empty when fitted, ideally clean. So long as there's power to the Organiser, all is good. The context is different for a low backup battery in an unattended Organiser, and this is the context where unexpected data loss hits hardest. The coin cell might last for 10 years, IF the Organiser has enough power to supply the RAM pack all that time, but they rarely do! When the main battery gets low (within a few months), the RAM backup battery has a lifetime not much longer unless it was very new to start with. That's a combination that multiples the risk.

If at that moment the Organiser is plugged to a PC to salvage the RAM contents, the result may be an ON-signal that makes the low main battery start to power not just the Organiser, but the Comms Link. That's enough to cause a sudden drop on the main internal 5V line (because of how a sudden change on one side of any capacitor tries to force the same change on the other), so unless the ground line has an extremely low resistance, the capacitor won't protect the stuff it's meant to from that down-spike, and this is when the RAM dies.

That strict sequence of actions in my previous post should be enough to guarantee that doesn't happen. Changing the main internal battery might work too, but as the Organiser wakes to check its clock every 20 minutes or so, it's like playing Russian Roulette with a wobbly firing pin. Having it do that in the instant when the main battery is out could be unfortunate.. I always assumed that a capacitor held enough charge to withstand that event, but if it was already low, all bets are off. Power to a switched-off Organiser via a Comms Link, after connection to an unpowered PC, seems to be the best way, whatever may be connected at the far end to salvage the RAM content. Not least because then we don't have to take chances on how well-formed the pins are in the top slot connection.

[amenjet]

i still hope to make an MRAM or FRAM datapack, well, get the ones I have already made working. Then you have a RAM pack with no battery, and lots of problems solved.

Andrew

[Lostgallifreyan]

If you do it well, especially for large capacity packs, I'm interested. That would be worth spending well to get. I don't know how often or fast such tech can be re-written, but if it's a lot, then FLASH itself is going to be at risk of obsolescence.

[Martin]

Reassurance

Those new to rampaks might feel uneasy about investing in one after reading through this thread.

Let me reassure any user considering acquiring one that I have had no issues with my 256K rampack in B: and 32K one in C:. You just have to treat them as you would any other 'battery retained RAM' such as that in A:... If you have valuable data - then back it up regularly.

Sincerely and in good faith
Martin


PS... Andrew

"I still hope to make an MRAM or FRAM datapack, well, get the ones I have already made working. Then you have a RAM pack with no battery, and lots of problems solved."

If you do manage with the MRAM or FRAM datapacks Then please put my name against two (or at least one if they are in short supply).. And you know I'm always interested in BETA testing them if you need.

Martin

[thesourcerer]

The only real problem with Rampacks is a shortage of them. The last 3 (32k) that Andrew modified for me sold straight away. Having a replaceable battery makes them very convenient. I also have a number of people wanting the 256k ones.
Why Psion produced the originals with soldered contacts (and a centre mounted pin on the negative) still puzzles me. I am sure they would have been more popular if they had easily replaceable batteries! Even conventional edge mounted solder tags would have been an improvement.