Fit-PC3 BIOS AC loss set to power on & it works...sometimes

[jimfred]

We have a couple of Fit-PC3 with the BIOS configured to Power On upon AC loss and it works sometimes and sometimes not. The 12 VDC supply looks clean - doesn't sag.

[mbirger]

I suspect it is a HW issue, thus please consider an RMA if this feature is important. Also please note if it is easy to reproduce the issue.

[jimfred]

mbirger-

Hey! I think I have an on-off-on sequence that's fairly repeatable on at least 1 'good' unit used in the field:
[0] unit is powered-up and running normally
[1] remove power for 2 seconds
[2] apply power for 1 second
[3] remove power for 3 seconds
[4] apply power and leave on.
At this point, I'd expect the unit to power up but instead I see no logo light, no Enet lights and I see a red LED (thru a side mounting hole) near the power section. To recover from this condition, a power-down doesn't seem to help - even if I leave power off for 10 seconds. To recover, the power button must be clicked.

I haven't explored variations in timing. Although this sequence is somewhat contrived, power bumps could be common without some sort of UPS or power conditioner. I can reproduce these results by plugging & unplugging either the 110 VAC-side or the 12 VDC side of the power supply that came with the Fit-PC. Timing is approximate and not critical - I didn't use a stopwatch.

We've joked about putting a watch-dog-activated solenoid on the power button but I'd guess that there's a more elegant solution like a UPS etc.

I hope you've seen something like this and can offer some insight or suggestions. I'm thinking that you could reproduce these results with almost any Fit-PC3.

-Jim

[jimfred]

I've attached a scope trace showing this problem in more detail.
There are 4 frames in the attached image:
[1] Shows a normal power-up sequence where, when power is applied, the Logo LED turns on and the Fit-PC3 boots normally.
[2] Shows a double-power-up. In this case, the Logo LED 'blips on' and then turns off. The Fit-PC3 doesn't boot. The timing is not at all critical and is easily reproduced. Power is being turned on&off at the 110 VAC side of the power adapter rather than the 12 VDC plug.
[3] Shows subsequent power-ups where the Logo LED 'blips on' for 120 msec but then turns off. No number of power-ups recover from this condition.
[4] The only way to recover from this condition seems to be a manual button press.

For an un-attended appliance, this would require that a technician be deployed to the appliance and press the button.

I've been able to repeat this problem on every Fit-PC3 - new or old. This is a serious problem that will require either adding a UPS or, if less expensive, selecting a new PC.

-Jim

[irads]

A DC sequence that interferes with fit-PC power sequence is definitely possible. With normal AC switching we have not observed power up issues - this is the recommended method.

[jimfred]

The problem is seen with AC switching rather than DC switching. Consequently, power bumps can cause the Fit-PC to not boot. Since I can repeat this with any Fit-PC3, you would easily be able to reproduce the problem.

[davelanton]

I have the same problem with several Fit-PC3's at unattended client sites. The systems are often unplugged and moved and then powered back up. Several times they don't boot up as expected, and they are not easily accessed.

Are you looking into a solution to this?

[jimfred]

Are you looking into a solution to this?

Yes - Compulab (the mfg) suggested either a UPS or the Intense PC.

This link http://www.nslu2-linux.org/wiki/HowTo/F ... erAlwaysOn addresses a similar problem with a totally different device by adding a tiny $0.35 Microchip MCP120-450DI/TO. I don't think it's cost-effective for us to apply such a modification to a Fit-PC3.

I suspect that timing needs to be tuned in an RC circuit on power circuit on the Fit-PC - but reverse engineering the circuit might not be cost-effective either.

We are looking into [1] a UPS-like add-on, [2] the Intense PC and [3] other similar devices. I don't have an easy answer yet.

...not easily accessed...

I feel your pain. The whole idea of an un-attended automated device is to reduce or eliminate operator intervention.

-Jim

[jimfred]

We discovered today that the critical time is an ON-time of roughly 2 seconds or less i.e., if the Fit-PC3 sees power for 2 seconds or less, upon the next power-up the Fit-PC3 will not start.

[opencompany]

We discovered today that the critical time is an ON-time of roughly 2 seconds or less i.e., if the Fit-PC3 sees power for 2 seconds or less, upon the next power-up the Fit-PC3 will not start.

Have you found a solution for this?
For us the 3 seconds limit is a major issue as well.

Thank you.

---
From CompuLab:
Current power supply disconnect/connect specifications for fit-PC3:

In order to avoid unexpected system behavior with current fit-PC3 HW design, power supply disconnect/connect operation should be applied as follows:

1. Power disconnect/connect to be applied on the AC side.
DC side is the output from the transformer and rectifier circuits, and cannot guarantee stable behavior when disturbed.

2. Time interval between power disconnect and next connect should be longer than 3s.
---