Boost voltage

[Tugdual]

Interesting reading. Possible solution to make a better use of 1.2V rechargeable batteries in the 50g for example?

[Hlib]

The threshold of shutdown in my HP-50G is equal 3.55 volts, elements spend 85% of a full resource here as well as rechargeable batteries. I improved HP-48gii with the block outside not to violate the internal scheme. It works close to 0.8V per one element. E.g. TI-83plus normally works close to 0.75V per one initially. The HP-corporation will never provide you information on the resistor which needs to be replaced in the scheme for change of a threshold of shutdown (throw out your money for batteries). Galvanic cells up to date don't correspond to specifications. When I received original FX-2.0 in 2000, in a set were firm 4xR03. Now even LR-03 is much worse on quality. I remember those happy days with "Colour graphic FX-9850GB Plus" which consumed current 4ma at maximum.

[Claudio L.]

(06-04-2015 02:31 PM)Hlib Wrote:  The HP-corporation will never provide you information on the resistor which needs to be replaced in the scheme for change of a threshold of shutdown

As far as I understand, the 50g battery indicator is fully controlled by software. It has Vcc connected to one of the A/D inputs, so you can read the voltage through the A/D converter and make a decision whether you want to turn off the device, show an alarm to the user, etc.
However, I don't think it's safe to use the batteries all the way down, as the device could reset at any point and may cause data corruption.
This is also true for the new booster device shown in the original post. If you fool your calculator into thinking you have good batteries when you don't, all you are doing is risking your precious data. Is the extra 20% energy you'll get from your batteries worth the risk? Probably not.

For other devices (toys, remote controls, etc), I think that voltage booster is a great device.

[Tugdual]

(06-04-2015 03:20 PM)Claudio L. Wrote:  

(06-04-2015 02:31 PM)Hlib Wrote:  The HP-corporation will never provide you information on the resistor which needs to be replaced in the scheme for change of a threshold of shutdown

As far as I understand, the 50g battery indicator is fully controlled by software. It has Vcc connected to one of the A/D inputs, so you can read the voltage through the A/D converter and make a decision whether you want to turn off the device, show an alarm to the user, etc.
However, I don't think it's safe to use the batteries all the way down, as the device could reset at any point and may cause data corruption.
This is also true for the new booster device shown in the original post. If you fool your calculator into thinking you have good batteries when you don't, all you are doing is risking your precious data. Is the extra 20% energy you'll get from your batteries worth the risk? Probably not.

For other devices (toys, remote controls, etc), I think that voltage booster is a great device.

I recently purchased one of these "intelligent" battery chargers and see that only 50% is drained when the 50g goes panic. This actually frustrates me a lot... especially when I have purchased pretty good batteries. Rechargeable batteries do not follow the same discharge pattern as normal batteries and nowadays "intelligent" equipment ask you for the type of battery you're using. The 50g always assume it is chemical and if 1.2V is a risk for standard batteries, it is quite normal for rechargeable batteries. Remaining question is what is the actual minimum voltage @ which the 50g can safely operate.

[Dave Shaffer]

Here's yet some more info on this. It's a bit breathless, but more background.

I suspect this is basically a joule thief in a thinned down sleeve. I see no reason why it shouldn't work, although not maybe quite a well as claimed.

[Claudio L.]

(06-04-2015 03:46 PM)Tugdual Wrote:  Remaining question is what is the actual minimum voltage @ which the 50g can safely operate.

I tested this last year. I don't remember exact values, but roughly these were the ballpark numbers:
The calculator idles in a "slow" clock, and jumps to a faster clock on demand.
I think for aklalines (nominal 6V) it was around 4.4V was still "safe", because when the turbo mode kicked in, the voltage dropped to about 3.8V and was still OK.
For rechargeables (nominal 4.8V), 4.0V was "safe", with voltage dropping to 3.5V or so on demand.
When voltage was around 3.3V the CPU would reset randomly.
But this is for bare minimum hardware. I didn't test if for example writing to flash could be achieved with only 4.0V, as it may drain more current. Also I didn't test if an SD card could function properly (voltage wise probably OK since they switch to their own voltages anyway, but current wise... may read OK but not write for example).

That the CPU can function doesn't mean is 100% safe, as I didn't test a worst-case scenario by any means.

I measured this using newRPL firmware, which has "slow" mode at 6 MHz and with the CPU in IDLE mode (Wait for interrupt), so it's supposed to drain less current than the stock firmware (12MHz with the CPU fully awake). On the other hand, the "fast" mode is 192 MHz (vs. 75 MHz stock), that's why the voltage drops are so steep.

[Katie Wasserman]

(06-05-2015 07:16 PM)Dave Shaffer Wrote:  Here's yet some more info on this. It's a bit breathless, but more background.

I suspect this is basically a joule thief in a thinned down sleeve. I see no reason why it shouldn't work, although not maybe quite a well as claimed.

I'm sure you're right, I'm equally sure that you'll never see anywhere near what they claim. I find the most interesting thing about this is being able to squeeze any useful circuitry into something 0.1mm. I suspect this is just a conductor and the components are all at one end of the device making it likely too long for certain applications. Still it might be useful in older or poorly designed electronic devices that don't work below 1.3 volts per cell. I also wonder what the quiescent current draw of this thing is -- if it's more than a few microamps that would make it pretty useless.



p.s. The patent it says "The height increase of the battery due to the sleeve is about 1 mm". That's not too bad, but will still be a problem in some devices.

[Graan]

Hello
Dave Jones at EEVBlog made a video about this 2 days ago , quite interesting

https://youtu.be/4iEshd6izgk

/Andreas

[Hans Brueggemann]

Let's assume that the calculator drains 50 mA from the battery pack. then, that device would try to squeeze about (1.5V x 50 mA / 0.9V / 0.7) = 120 mA from an already flat (0.9V) cell, to maintain 1.5V at the terminals.
no way that this is going to work.

hans

[Graan]

Since the device will always try to maintain 1.5 V the built in battey gague in devices will show "battery full" until the device will drop dead suddenly. Not really nice.

[Tugdual]

(06-06-2015 06:09 AM)Graan Wrote:  Hello
Dave Jones at EEVBlog made a video about this 2 days ago , quite interesting

https://youtu.be/4iEshd6izgk

/Andreas

Interesting indeed but Lord, I couldn't hear that guy for 30mn.

(06-05-2015 09:17 PM)Claudio L. Wrote:  I tested this last year. I don't remember exact values, but roughly these were the ballpark numbers:
The calculator idles in a "slow" clock, and jumps to a faster clock on demand.
I think for aklalines (nominal 6V) it was around 4.4V was still "safe", because when the turbo mode kicked in, the voltage dropped to about 3.8V and was still OK.
For rechargeables (nominal 4.8V), 4.0V was "safe", with voltage dropping to 3.5V or so on demand.
When voltage was around 3.3V the CPU would reset randomly.
But this is for bare minimum hardware. I didn't test if for example writing to flash could be achieved with only 4.0V, as it may drain more current. Also I didn't test if an SD card could function properly (voltage wise probably OK since they switch to their own voltages anyway, but current wise... may read OK but not write for example).

That the CPU can function doesn't mean is 100% safe, as I didn't test a worst-case scenario by any means.

I measured this using newRPL firmware, which has "slow" mode at 6 MHz and with the CPU in IDLE mode (Wait for interrupt), so it's supposed to drain less current than the stock firmware (12MHz with the CPU fully awake). On the other hand, the "fast" mode is 192 MHz (vs. 75 MHz stock), that's why the voltage drops are so steep.

Thanks Claudio. This means we can operate safely until about 1V which is aligned with the video and is a very good news. Do you by chance know how the battery indicator reacts to voltage?

[Claudio L.]

(06-06-2015 02:29 PM)Tugdual Wrote:  Thanks Claudio. This means we can operate safely until about 1V which is aligned with the video and is a very good news. Do you by chance know how the battery indicator reacts to voltage?

I completed a test run with NiMH rechargeables and my own "safe" battery warning values. With the calculator continuously "on", running at 6 MHz, this morning the battery indicator started showing continuously (asking me to change the batteries). Once the indicator shows, the calculator locks the slow mode to prevent failure, so we should assume current stayed at 12 mA more or less throughout the test.
On light use but without turning it off for a couple hours, the battery indicator finally went critical (blinking), and almost exactly one hour later the calculator began resetting itself in a loop.

That's what my "safe" voltages will get you: a couple of hours of operation at slow speed, and then about one hour more once it goes critical.
Now I don't think these voltages are so "safe" anymore, perhaps I need to raise them a little more.
Later, I will measure the voltage (with no load, as the calculator can't function) on these batteries and report back. This will be the reference voltage for "fully depleted" batteries.
Stay tuned.

[Claudio L.]

(06-08-2015 09:55 PM)Claudio L. Wrote:  Later, I will measure the voltage (with no load, as the calculator can't function) on these batteries and report back. This will be the reference voltage for "fully depleted" batteries.
Stay tuned.

The NiMH batteries read 1.18 V without load, all four show a surprising 4.68 V on my (dubious quality) multimeter.
I put the batteries back and the calculator is functioning, so it looks like they bounced back after the reset loop. But with the critical warning on, under load I'm getting 4.21V (slow mode). I started calculations, and voltage dropped to 4.10V. It stayed there for 10 seconds, while the machine kept doing calculations in a tight loop. After that, started descending slowly (around 0.1V per second) and it went downhill from there all the way to 3.5V and started resetting.

So it seems 4.10V is still not as safe as my previous measurements would suggest (this probably is an older set of batteries, with many recharge cycles on them). I'll have to revise my numbers.

[BartDB]

(06-08-2015 11:39 PM)Claudio L. Wrote:  The NiMH batteries read 1.18 V without load, ...

Yes, even depleted batteries will show a no-load voltage close to their normal operating voltage (there is still a small amount of active material in the electrodes, and the electrode potential always remains the same, it is the internal resistance that causes voltage drop and this only happens when a current is flowing. The internal resistance of a depleted battery is higher than that of a charged battery hence the voltage drops more for a depleted battery (there is less active material available in the electrodes thus increasing resistance)).

However, as soon as you put a load on them they "collapse" quickly as you have shown.

(06-08-2015 11:39 PM)Claudio L. Wrote:  I put the batteries back and the calculator is functioning, so it looks like they bounced back after the reset loop. But with the critical warning on, under load I'm getting 4.21V (slow mode). I started calculations, and voltage dropped to 4.10V. It stayed there for 10 seconds, while the machine kept doing calculations in a tight loop. After that, started descending slowly (around 0.1V per second) and it went downhill from there all the way to 3.5V and started resetting.

So it seems 4.10V is still not as safe as my previous measurements would suggest (this probably is an older set of batteries, with many recharge cycles on them). I'll have to revise my numbers.

You are basing your final statement on re-inserting using batteries that are already depleted. This is nugatory. You need to log the voltage of a charged set of batteries from start to finish to see how the voltage drops over time and base your decisions on that.

My personal view is that your settings seem perfect:
1) a few hours once indicator comes on = will probably last to the end of the day giving enough time to get home and put in charged batteries.
2) "critical indication" in my mind would = replace now, so an hour of operation seems acceptable.

Best regards.

[Claudio L.]

(06-09-2015 11:09 AM)BartDB Wrote:  My personal view is that your settings seem perfect:
1) a few hours once indicator comes on = will probably last to the end of the day giving enough time to get home and put in charged batteries.
2) "critical indication" in my mind would = replace now, so an hour of operation seems acceptable.

My concern wasn't that 1 hour might be insufficient, just that statistically we should have some cushion to account for standard deviation in the behavior of batteries. I'm not sure it will behave correctly for *any* set of batteries, as my sample of one set is obviously not enough to draw a final conclusion. Perhaps someone will find a set of batteries where it will last only 10 seconds when it goes critical, and that wouldn't be acceptable.
But I think I'll leave it like that for now, until we get a decent size sample of batteries.

[BartDB]

(06-10-2015 12:19 PM)Claudio L. Wrote:  I'm not sure it will behave correctly for *any* set of batteries, as my sample of one set is obviously not enough to draw a final conclusion.

Well, you can't really cater for *any* set of batteries, particularly as there are now many cheap unbranded and counterfeit branded batteries about. e.g. I purchased a 48 series RAM card from ebay and the seller claimed it had a new battery, it measured 0.12V when I got it a few days later. It only had "Lithium cell" stamped on it - you can't cater for "cheap shit" .

However, I agree about the low sample size. This is for both the batteries and calculator (e.g. does every calculator start resetting at 3.3V?). Also things to think about are e.g. usage of IR (restrict it when low batt is on?).

Do you have a calculator flashable version of newRPL available? I could only find a windows demo on the newRPL page.

Best regards.

[Claudio L.]

(06-11-2015 09:15 AM)BartDB Wrote:  However, I agree about the low sample size. This is for both the batteries and calculator (e.g. does every calculator start resetting at 3.3V?). Also things to think about are e.g. usage of IR (restrict it when low batt is on?).

While I don't think IR consumes a lot, I know some SD cards, especially new "fast" ones, can draw some serious current and get very hot when writing, so a calculator with a battery warning may fail to write to SD card or something like that. But I can't test it until I have an SDHC driver in place.

(06-11-2015 09:15 AM)BartDB Wrote:  Do you have a calculator flashable version of newRPL available? I could only find a windows demo on the newRPL page.

Yes, I do. I'll send you a PM either today or tomorrow.

[TASP]

Has anyone adapted an inductive charger setup to any HP calculators? (or any newer models come that way from the factory?)

Also, would the 'joule thief' circuitry be something to consider wiring internally to, oh, let's say an HP41 hypothetically speaking, LOL ?
My first thought was it would just be a problem with card reader operations, but perhaps a little more circuitry would help; like maybe slowly ramping the voltage from just above to just below the 'BAT' indicator voltage to make it blink, and having onboard sufficient capacitance to complete one card read. (realizing that both techniques there at the same time seems a little tricky to implement)