NimH AAA to 5V, suggestions?

That's what I did as a test... paste into UltraEdit, reformat paragraph width, then add the >'s.

As I use UltraEdit extensively to conform strange Spice variants to PSpice, I suspect the whole thing can be with a macro... all automated.

Actually I have several macros in Macro Express that trim quotes, for example, any lines with more than 4 >>>> go bye-bye automatically ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

                   Spice is like a sports car... 
     Performance only as good as the person behind the wheel.

Yeah, I don't often care enough to do that, unless I'm breaking up a paragraph to respond to points raised within.

Years ago I used to call out to an editor frequently, when writing code under dos or windows, but since writing software on Linux became my 'normal' environment, only time recently I'm calling out to an editor in recent times is to snag LTSpice example and drop it into a name.asc file. Using Notepad++ these days, I did use ultraedit years ago. And, to view unix text files.

Grant.

I'll stay with add newlines, then cut paste as quote to get the '>', when it occurs to me ;)

Grant.

On a sunny day (Wed, 04 Aug 2010 10:08:30 +1000) it happened Grant wrote in :

(inside the keyboard).

Here is a test with a 'store bought' inductor, well, I took it from some old PCB, ftp://panteltje.com/pub/1.2_to_5V_converter_img_2252.jpg This is that coil in detail, I added a 5 turns secondary for feedback :-) Also changed the diode for a smaller Schottky. The effect is that the frequency drops a lot, and efficiency goes up a bit. I measure this: In 90 mA at 1.222 V -> 0.10998 W Out 13.87 mA at 5.53 V -> 0.07670 W This gives an efficiency of 69.74 %. The frequency is 37 kHz. With a 2300 mA NiMH rechargeable the thing will work for 25.55 hours exactly. I am satisfied, It should also work with high efficiency LEDs or LEDs that drop close to 3 V. That is why I use a 5V supply, and not 3.3.

I forgot to measure this inductor, in fact I should also design an inductance meter, but it could be as much as 160 uH, if they followed the application note in the thing I took it from. (A L4962A based switcher), but it could be far less.

An efficiency of say 70 % does not seem much, but is difficult to improve at

1.2V, due to the Vce sat of the transistor I think. Maybe things could be tweaked a bit to 75 %, but likely not much more.

On a sunny day (Wed, 04 Aug 2010 16:31:50 GMT) it happened Jan Panteltje wrote in :

Forgot lthe link: ftp://panteltje.com/pub/1.2_to_5V_converter_coil_detail_img_2256.jpg

Nice chip, MOS switch, abd that at such a low voltage. It sure is smaller then the discrete circuit I have now, Efficiency seems about the same for the A and a bit better for the B version. For sure something to keep in mind if this ever becomes a big series.

Yes, I saved a few of that style from dismantling old gear.

Grant.

fix=3D

ock=3D

be =3D

=A0Yo=3D

uld=3D

ent=3D

s.

my version.

NiMH batteries.

The switching transistor isn't saturating very well--that's costing you a few % in efficiency. FMMT617 is quite good. What kind are you using?

If the output's 5.5V, then that collector waveform must be about 1.2V/ division. What's the deal, is the output set low, or are those metric volts?

-- Cheers, James Arthur

That's like the constant current buck switcher I made,

but I only needed two turns. BJTs sure have high transconductance.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

On a sunny day (Wed, 4 Aug 2010 20:36:09 -0700 (PDT)) it happened snipped-for-privacy@yahoo.com wrote in :

Just a BC547.

I think in that version it was 4.85 V output. See my other post for the improved version, different coil.

e

Yes, I just monitored Vce on the scope and kept adding turns until I no longer did see any improvement. Also tried a synchronous detector with a BJT, but the Schottky was better and simpler. Even tried it as a switch 'upside down'. About 70% efficiency is OK at 1.2 V I think. What I am thinking about now is a simple low battery detector to switch it off when battery voltage drops below say 1.05 V. At .95 V the circuit no longer hacks it, and the output drops to 3V, and The PIC stops working correctly. It is easy to detect with a BJT using the 0.7 V Vbe as reference, , but it needs some flip-flop action too. To protect the expensive rechargeables.

I've observed hysteretic operation in these before. Obviously they turn on at 0.7-0.8V, but they turn off slightly below due to bias generated by base rectification.

You could tweak the turn-on voltage with a high-ohm divider (not exactly open circuit, but arbitrarily low leakage anyway), and turn-off voltage by boosting base current.

You can generate more base current by rectifying the flyback, which is negative going at the base:

Worst case of this circuit is an E-B diode, which clamps base at -0.7V, forcing maximum base current back into the circuit, making it run at full throttle. An optoisolator, of course, is an excellent controlled current source.

This goes for Vebo(max) breakdown, too, which is why it's a good idea to clamp the flyback voltage on these things (maybe instead of RCD snub, a diode + TVS), and to use a series base diode, to prevent runaway action.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

iMH=3D

hen=3D

ty =3D

u

I saw the other post--70% efficiency is decent. I get 80% commonly,

85% with great care in similar circuits. You have to use a low Vce(sat) transistor--otherwise 150mV out of 1.2Vcc in a discontinuous- mode converter works out to about a 8-10% loss.

-- Cheers, James Arthur

tje

:

No need to protect a single NiMH cell--you can't damage it here. Full discharge is fine--damage comes from reverse charging the cell. This circuit can't do that.

In practice you won't fully discharge the cell, because the PIC will quit before you ever get there anyway.

-- Cheers, James Arthur

On a sunny day (Fri, 6 Aug 2010 09:35:10 -0700 (PDT)) it happened snipped-for-privacy@yahoo.com wrote in :

That iwould be good, that would simplify things. I was conidering using the analog comparator in the PIC, combined with some I/O and RC time.

Yes it will stop working normally, but all LEDs seem to go on then :-) A brown out detection and / or watchdog would simply keep restarting it. and the power conversion circuit will keep running and consuming some power of its own.

Just looked up NiMH : Wikipedia seems to disagree with you:

A fully-charged cell measures 1.4?1.45 V (unloaded), and supplies a nominal average 1.25 V/cell during discharge, down to about 1.0-1.1 V/cell (further discharge may cause permanent damage, and the risk is increased with multi-cell packs).

On a sunny day (Fri, 6 Aug 2010 09:28:00 -0700 (PDT)) it happened snipped-for-privacy@yahoo.com wrote in :

It is just a BC547, 'made in China' nothing special. You also lose .5V in the Schottky, at 5V that is an other 10%.

On a sunny day (Fri, 6 Aug 2010 09:35:10 -0700 (PDT)) it happened snipped-for-privacy@yahoo.com wrote in :

PS I just had a cool idea how to do the auto shut off: -------------> to converter | + 1.2V NiMH _____| - | d === g ---------------------------< from Vout (5V nom) --- s IRLZ34N MOSFET |C1 | /// ///

The MOSFET is in series with the battery minus connection. At switch on C1 is discharged, and the circuit will produce 5V output for a few milliseconds before C1 charges. That is enough to make the MOSFET conduct. When output voltage drops below Vgs of the MOSFET (say 2.5 V, else use a divider), the MOSFET Rds increases, output voltage falls further, and zero current will flow. I mean really ZERO. Now that is a minimum component solution of the ultimate kind :-) The Rds-on of this MOSFET is a few milli-Ohm, so 90 mA current causes millivolts drop, no losses to speak of,.

I should patent this :-)

Still have to try it though :-) But I *have* this MOSFET.

On a sunny day (Fri, 06 Aug 2010 17:52:51 GMT) it happened Jan Panteltje wrote in :

milliseconds before C1 charges.

divider),

flow.

millivolts drop,

OK tried it, of course the intitial idea was OK, but the polarity wrong. So now it looks like this, and I replaced the cap with a momentary start-button. start button | --- ----0 0---- | | -------------- s d -------> + to converter | g IRLZ32N + | --------< +5 from converter NiMH | | - __| 0-------->[ ] cut off level | stop | 0 | 100k /// buton | /// /// A stop button contact could be made by shorting the gate to ground. Solves the expensive switch problem at the same time, foil swicthes? hehe

On a sunny day (Fri, 06 Aug 2010 18:59:46 GMT) it happened Jan Panteltje wrote in :

few milliseconds before C1 charges.

divider),

flow.

millivolts drop,

Typo, reverse drain and source! It works:-)