[En-Nut-Discussion] Re: Mass storage, RTC daughter board for the Ethernut-2.

[Grahame M. Kelly]

On Thu, 21 Oct 2004 08:00 pm, en-nut-discussion-request at egnite.de wrote:
Hi Douglas / Evening Mailing List.

> (1) Is this of interest and should therefore be published?

I would be interested in this daughter board and would kindly ask you to
publish it (with your code) under S/W & H/W GPL conditions - for all to
enjoy.

> (2) I am considering using a particular Maxim RTC chip that allows me to
> have a battery (or perhaps a 1 Farad capacitor) of such a size that the
> Ethernut-2 could operate long enough to flush all of the FAT buffers to the
> MMC and close files etc, and maybe send out an alert.

I have used the DS1302 with a 1Farad Super Cap, and 32.768 KHz Xtal.
IMHO its the best way to go, as you can also "store" design/operational 
parameters on the RTC EEPROM area (things like alarms, limits) seperate to
which ever AVR controller you use.

I would also like to see you make provision for connecting DS18B20 Temperature 
packages, as they can be useful in monitoring equipment temperatures 
remotely.

Using a 1F Super Cap to maintain power to the Ethernut is another issue 
totally!  Your looking at an EtherNUT being ~2.7Watts i.e. 2.7W.Sec. A 1F cap 
at 5V DC gives ~12.5W.Sec, so excluding your circuitry, max time before it 
dies is going to be when Cap voltage falls below ~3.5VDC (Assuming all IC are 
from the HC or CMOS families) you will get about 1 to 1.2 second uptime max 
IMHO.

> (3) How much current would a 1 Farad capacitor supply at 5v, and for how
> long?

It can hold the DS1302 RTC in working order, using only 1uW/sec according to 
the spec.

Given Watt sec = (E^2 * C)/2 and E = 5VDC, C = 1F then  W.Sec = 12.5x10^6 sec.
which divided by (60 sec x 60 mins x 24 hours ) approximates to 144 days.
So about 100 to 130 days would be a good expectation given sods law;
i.e. circuit - temperature changes, device leakage, humidity variations etc.

On a DS1302 RTC you can select by S/W the internal charging resistor(s) and 
the number of diode(s) [voltage drop = 0.7v/diode] placed across the charging 
circuit to the super-cap.

Look at the DS1302 PDF = http://www.maxim-ic.com/quick_view2.cfm/qv_pk/2685

at the bottom of page 8/13.

The minimum charging current will be 0.45 mA and the maximum charging current 
is 2.2 mA. So its all dependant on the initial voltage across the super-cap, 
and your charging selection. My experience is has been 4 hours to fully 
charge the super caps (a sample of 3 only) so, you will have to see for 
yourself. But once it reaches its optimum charge, trickle charging takes over 
to keep the RTC in full working order, with and without power. See PDF spec 
for full details. Sometimes I pre-charge them at higher current than allowed
to get first time circuit power ups, working for full in house testing.

> (4) Does the LM1086 have reverse voltage protection so that if my daughter
> board is supplying power to the 5v line (and hence the 3.3v line) with no
> input power to the LM1086, will this damage the LM1086.

Look at the spec:
http://optics.ph.unimelb.edu.au/atomopt/electronics/datasheets/voltregs/lm1086.pdf

Looking at the basic functional diagram in the spec (page 1) it would appear 
not!  i.e. The device should not be bothered by it.  But, try it on a number 
of samples (say 10) and check your results - it is the only way to be sure.
I do likewise on my designs where the EtherNUT card power regulator is not 
being used as an external power source via my own card provides the necessary
voltage/current for the Ethernut.

> (5) I may also add a reset button that can be accessed from the same side
> as all the connectors are on so that the unit can be reset without having
> to pull the power from the unit (when it is in an enclosure).

Connect your reset button to Ethernut J5 Pin 9 - MR (Master Reset Input). This 
way you will keep all circuits "in-sync" which is very important. AND make 
sure you allow for very good grounds to the Ethernut, do not depend on just 
using the ground pins on the J5 connector.  Also, keep +5VDC noise/ripple to
less than 10mV if at all possible.

> (6) In a later revision I was thinking of allowing the Ethernut to Boot and
> self program from the MMC card, meaning that customers could be sent a new
> MMC card (e.g. a cheap 8MB one) that reprograms their Ethernut without
> requiring an engineer or on site visit, a good idea?  

Why not just provide the clients with an available online upload file so that 
your clients EtherNUT can simply check for a update existence.  If an update 
exists, let the clients Ethernut auto-magically upload, store, check and 
re-boot itself - just as you can do by using tftp boot now. 
See the: Ethernet Boot Loader  @ http://www.ethernut.de/en/software.html
and its s/w readily available at http://www.ethernut.de/en/software.html
Ethernut Boot Loader.

This way your client Ethernut can upgrade themselves at any time.
You only have to place the upgraded s/w onto a web or ftp site. 
You could also program client Ethernut to do interval checks (be it daily, 
weekly, monthly whatever) to test for new download version availability and 
if you wanted you could capture the IP address of those Ethernuts that self 
loaded because there maybe problems with the other unots or they maynot
have connectivity you originally requested/required.

Hope this assists you somewhat.
Cheers. Grahame