Power management and deep sleep

[Tim Wessman]

Hello,

I was thinking about power consumption and so on earlier this week and realized that I really don't have much knowledge on just what earlier calcs used to be so good at sipping power. Does anyone have any links or useful info on any of the techniques that were used? Custom chips helps a lot, but I know there was more than that.

For example, I just found out that the 41s had a deep sleep mode (yeah, I know. . . I'm a young'un). Is that similar to the 48 deep sleep? In what ways was it different? When was the deep sleep first introduced?

I know that was used to ship the thing in the box to make sure it would have batteries when you pulled it out, but are there any links or info anyone can point me to about how/when this was implemented?

Thanks!

TW

Edited: 8 Jan 2011, 12:03 p.m.

[Katie Wasserman]

Tim,

I don't know about "deep sleep" modes, but if you're talking about the current HP ARM-based designs I believe that the only way to drop power consumption enough for really long-term storage is to power off the chip. In power-off mode the AT91SAM7L128 will consume 100na verses 3.5ua in the next lowest power mode (which I think you use now). Waking up form power-off mode can only be done using the FWUP pin which could be tied to the "ON" button on the calculator through a diode switch.

Alternatively, you could use a "hardware" solution by putting a small piece of Tyvek between the bottom of the battery and the contact on the calculator. Have a short tab stick out of the battery compartment telling the new owner to "pull this out" when ready to use the calculator. --- I just tried this on the 12C+ and it works quite well.

-Katie

[Tim Wessman]

Yeah, I'm familiar with what can and can't be done with the current ARMs. I am just feeling more interested from a historical/technical/personal interest.

TW

[Andrés C. Rodríguez (Argentina)]

Just for the record, my HP-41C (1980) came with a small box (located under the calculator) with four "N" batteries; so no batteries came installed. I think my HP-42S (1990) came with the batteries already installed.

[Garth Wilson]

My 41cx's alkaline batteries would go about two years until I got the double extended memory module, and now they only go about six months. Obviously one of the memories is a bit leaky. CMOS circuits theoretically take absolutely no current at all however when they are not switching. Even in the mid-1980's I remember having a 10uF capacitor holding a SRAM's memory intact for 24 hours or more. So if the processor has a stop instruction, and it has an on-board clock oscillator so it can even shut that down, and then you can wake it up with an interrupt, it can sleep on virtually no power. The 41 can wake itself up with alarms, an in fact my 41cx is my only alarm clock (although it was more cool 20 years ago when I was using it to wake itself up and take data periodically, log it, and put itself back to sleep). I don't know how much current the timer module takes, but it's obviously peanuts, like a wrist watch.

[Didier Lachieze]

Hi Tim, the HP-41 power management is described in details in the HP- Service Manual under Theory of Operation - System Operation.

In summary there are 3 different modes:

RUN: CPU is working and display is active

STANDBY: CPU clock is off - display is active

SLEEP: CPU clock is off - display is off

When you press a key the HP-41 goes in RUN mode, executes the associated code and goes back in STANDBY mode, after ~10mn without activity or if you press ON the HP-41 goes to SLEEP.

[Håkan Thörngren]

I believe hunting for low power was with the HP-41, LCD and use of CMOS. The processor just shuts off, the power consumed is to hold the RAM which at least from the beginning was separate circuits.

I once had the batteries out of a HP-41C (no extra memories) for a week, then realized that I had not saved the program I wrote on it. I put the batteries back and the memory contents was intact!

The HP-10 series uses basically the same CPU, but running at half voltage and lower clock frequency.

There were no real time clock in the 10-series, the HP-41 introduced it later.

I am not so familiar with the Saturn that came after it, but remember that the HP-28S could hold memory for amazing periods of time by going into coma, dropping the clock update. I had one for years in a drawer with batteries going flat before I resurrected it by putting in new batteries and the memory was still there. The clock was off though. I ended up selling it and eventually got a HP-48GX.
I believe there is some key sequence at least on HP-48 to put it into coma. This was to allow it to be shipped with batteries installed without affecting the batteries much.

The HP-41 shipped with batteries in a separate box. I do not remember whether the 10-series or the early Saturns shipped with installed or separate batteries.

Quote:

---

Hello,

I was thinking about power consumption and so on earlier this week and realized that I really don't have much knowledge on just what earlier calcs used to be so good at sipping power. Does anyone have any links or useful info on any of the techniques that were used? Custom chips helps a lot, but I know there was more than that.

For example, I just found out that the 41s had a deep sleep mode (yeah, I know. . . I'm a young'un). Is that similar to the 48 deep sleep? In what ways was it different? When was the deep sleep first introduced?

I know that was used to ship the thing in the box to make sure it would have batteries when you pulled it out, but are there any links or info anyone can point me to about how/when this was implemented?

Thanks!

TW

---

[Christoph Giesselink]

Power Management on Saturn CPU based calculators

Normally whe distinguish between 4 modes, but most of the power consumpting units can be separately switched on or off.

1) Running Operation

The calculator is doing normal operation at nominal crystal speed, the CPU decodes opcodes, display is on, timers are on, 1ms keyboard scanning on and the Clarke/Yorke chip the card detection updating every 122uS.

2) Idle State

This is the state where the calculator is on and is waiting for input. In this case the Saturn CPU is powered down by the SHUTDN opcode, display is on, the timers are on, 1ms keyboard scanning is on and on the Clarke/Yorke chip the card detection updating every 122uS.

3) Light Sleep

This is the standard off mode. In this case the Saturn CPU is powered down by the SHUTDN opcode, display is off, the timers are on, 1ms keyboard scanning is off and on the Clarke/Yorke chip the card detection updating every 1mS.

4) Deep Sleep (aka Coma mode or storage mode)

In this case the Saturn CPU is powered down by the SHUTDN opcode, display is off, the timers are off, 1ms keyboard scanning is off and on the Clarke/Yorke chip the card detection is off.

Enter HP28S/HP48 Coma/Storage Mode

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

High End Pioneers: <ON> <+> <SK6>

HP28S:             <ON> <ENTER> <<->

HP48SX/GX:         <ON> <SPC>

Beware that the HP28S has a bug in the hardware design around the Lewis chip, so the calculator is consuming more power in coma mode than in light sleep mode.

Here's an extract of

Path: lth.se!sunic!uupsi!rpi!dali.cs.montana.edu!uakari.primate.wisc.edu!sdd.hp.com!caen!uflorida!haven!umbc3!gmuvax2!peraino

From: peraino@gmuvax2.gmu.edu (Bob Peraino)

Newsgroups: comp.sys.handhelds

Subject: HP28 power study

Message-ID: <3017@gmuvax2.gmu.edu>

Date: 30 Nov 90 21:59:50 GMT

Organization: George Mason Univ., Fairfax, Va.

Lines: 290



       Power Consumption Statistics for the HP-28 Scientific Calculator

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

                                Robert Peraino

                               Scott Hutchinson

                               January 18, 1990

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

...



Standard Modes

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

	 Off           -   13.5  uAMPS

         "Storage mode"-  106.2

	 On/idle       -  446.0

	 On/running    - 2412.0

Christoph