Power Supply Question
- According to Nigel Spon's "Power Switch Schematic", the raw NSLU supply provides +5v @ 2A. But the internal power supplies (+5V, +3.3V, +1.3V) are not allowed to come on until the circuitry is satisied the supply is clean and "respectable".
It is stated that, of this internal +5V, 0.5A goes to provide the stabilized +3.3V amd 1.3V, and 0.5A is available for the USB devices (is that for both of them, or for each of them?)
But how is this limitation to 0.5A achieved? I can find no circuitry anywhere which would cause that limitation. So how is it done?
Now I have the need to drive up to 7 LEDs (maybe 30mA each), an LCD backlight (say 100 mA) and up to 4 relays (30mA each), so maybe 430mA in total. Each driven by its own transistor, of course, so I should be able to use the "raw" +5V for these provided I can be sure that the transistors are all "off" until the "official" power supplies are up and running.
Or should I try to bring out the "clean" +5V (and would it supply that much anyway?) OTOH, I will need to bring out a little of the clean +5V, because my ADC requires a Reference Voltage of 4.096V.
Next, I shall bring out the i2c bus by tapping inbto the clock chip in the customary manner. That purportedly brings out GND (Vss) and a +3.3V (Vcc), but on examining the actual circuits, there seems to be some unidentified resistor in series with that +3.3V. So how much am I allowed to draw?
Originally, I had intended to run the LEDs directly off the i2c bus, but that is obviously out of the question now. Most of the gizmos on the bus use very little. But I shall need 1mA pullup for each LED/relay I need to drive, so would it be safer to use the clean +5V ffor those?
Anf finally, I see the clock chip has an open-drain INT pin on it (pull it down to cause an interrupt). Also, my pushbutton/LED driving chips provide an INT pin, which is pulled down whenever any pushbutton is opened or closed.
Would it be in order to connect that up, so that any pushbutton change would cause an interrupt, and thus save me the trouble of running a polling loop?
And if I did that, how would I catch the interrupt? Can I tap into some facility of some existing driver, or would it need some extra coding at kernel level?
And what is that interrupt from the clock chip currentl;y used for anyway?
Recently I upgraded my NSLU2 to install a WebDAV server. Two things
1) Perl broke. The built-in "-s" always returns 0 instead of returning
the size of a file. As a result, awstats broke as well.
2) Lighttpd is broken. The server cannot reliably serve a file over
1MB. The file gets truncated at some multiple of 512KB.
In summary, I see all the signs of a dying platform. Please advise on
the best course of action:
- How can I recompile Perl and Lighttpd? What environment is needed?
- Is any of the alternative distributions recomended over Unslung,
namely in terms of being better maintained?
Joaquim Baptista, running cacto.no-ip.org