what's wrong with this circuit?

Or in the extreme theoretical case, it just rings forever and the charge sloshes back and forth. No energy is lost. No charge is lost.

John

While I like that interpretation -- since I thought someone posted an analysis where, with a resistor in series, that's where the power goes, and the amount of power in that resistor remained a continuous function even in the limiting case as r->0 -- I kinda suspect the radiation case is perhaps the more correct one.

The problem, of course, is that everything is a "little bit" of an antenna, but it's not so attractive to run around a circuit model adding little radiation resistance to all your nodes... although I suppose this isn't that much worse than adding, e.g., parasitic capacitances or inductances all over the place (...the real problem is going to be estimating those radiation resistances...). More fundamental it gets down to the question of, "what actually transfers power -- voltages/currents or magnetic/electric fields?" and how, really, either approach is just as valid, you just have to be careful to make sure you account for ALL the fields or voltages/currents (including displacement currents)...

Antennas didn't fit in entirely cleanly with traditional circuit analysis anyway. E.g., while passive, they don't comply with Foster's reactance theorem.

---Joel

A Spice simulation, two caps connected by an inductor, would ring forever. It would be interesting to see the effects of algorithm and floating-point errors.

John

Indeed. Note that one of SPICE's forerunners was CANCER -- Computer Analysis of Non-linear Circuits, Excluding Radiation!

That energy analysis overlooks a crucial fact: the solar cell is a

50mA constant-current source. It'll give you 50mA at .55v, or .35v, it doesn't care, but once case yields 28mW of usable energy, and the other only 18mW.

That is, you get more energy out of the solar cell if you operate it at its optimum voltage.

The circuit as shown is sub-optimal, quite lossy. For one thing, what's this fixation with ripple? The load is a secondary *battery*, e.s.r = 100mA^2 x 1 ohm = 10mW while discharging). Okay, that saves 20% or so.

Using a 100v-rated IRF530 for Q4 is wasteful. Switching to a lower- voltage part lowers on resistance, is easier to drive, and cuts gate, switching and conduction losses all at the same time. And, we can replace that big TO-220 with a $0.23 SOT-23.

The original's cute gate driver isn't needed if we just parallel the gates we saved from simplifying out the one-shot. That saves a transistor and a diode. And, if we make C2 larger we can keep the solar cell operating closer to its optimum voltage for more of the time, further improving efficiency.

Here's a refined, simpler, significantly more efficient version that won't hang (I think I got all the inversions right):

,-----+----------------------------------------, | | | | | )|| L1 | --- solar cell )|| 100uH | ~~> - 0.55v )|| | | | | === ,------------+--------------------|------------, | | | | | | R3 | 2x CD4093B | | | 1M U1a | gates, | BAT54 | | | |\ ,------, paralleled +-->|---+----+ | +-| >O--|1 14 \ U1c,d | | | | | |/ | \ |\ |--'Q4 | |+ | | | U1b 3 )o--+--| >O-|| PMV60EN | --- | | |CD4093B / | |/ |>-. | - | Q2 | .---|2 7 / Rx | | --- | PN2222| | '------' | | | - | R7 |/ | | R4 | | | | +--100k-+--| +---------+--100K-+ | 16v | | | |>. | | | | | zener | C2 R2 | --- C4 | | D4 | | | | 100uF 1M | --- | '--|

te:

e:

t. You

. JL is

served?

battery

ee

the

d by

the

sume a

e

en

to

solar

ge

2 x C x

lar x

tio

l

rgy

x C

making

y

).

Here's one I have trouble with. Suppose there are 2 equal weights connected with a pulley (one on the ground, other at the top) so that as one weight falls, the other rises, and they both end up in the middle at half the original height. Seems the ending potential energy (mgh) is the same as the initial, since each weight has mgh/2. Assuming some friction losses, where did the extra energy come from?

-Bill

The weights won't move from their initial position unless some additional force is brought to bear.

What makes the system move in the first place? If we assume the connecting cable/rope/whatever is massless, the force exerted on both weights is equal. What force upsets that equilibrium to cause the weights to raise & lower to reach equal height?

If the cable/rope/whatever has mass, then the side with the weight on the ground will have greater downwards force than the other side, because of the extra mass of the cable/rope/whatever. What force cause the weights to move to the same height?

Ed

So IOW, rephrase the question in terms of equal weights on each end of a seesaw rather than a rope - and - pulley mechanism?

--Winston

Provided it is constructed so that its centre of gravity is exactly on the axis of rotation and the beam is truly rigid it will be stable at any angle. In practice deformation of the loaded beam moves the centre of gravity down slightly and the system tends to either horizontal or vertical equilibrium. Balances are rigged so that the beam is as rigid as is practical and centre of gravity is slightly below the beam centre by design. Trimmers are provided for fine tuning this.

A more interesting question is if you connect a small partially inflated balloon to a larger one with a glass tube and a closed tap. When you open the tap connecting the two balloons what happens?

Regards, Martin Brown

You

JL is

rved?

ttery

he

by

e

e a

r

C x

x

ng

That has nothing to do with the simplified analysis I gave. If the circuit switches into flyback just when I=3DImax, then it looks like the solar cell forward conducts and clamps the voltage across C to a diode drop.

=A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 =A0 =A0 =A0 =A0 )|| =A0L1

=A0 =A0 =A0 )|| 100uH

=A0 =A0 =A0 =A0)||

=A0 =A0 =A0 =A0 =A0 =A0|

-----,

=A0 =A0 =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 =A0 =A0|

93B =A0 | =A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 | BAT54 =A0 =A0 =A0|

+-->|---+----+

=A0 | =A0 =A0 =A0 | =A0 =A0|

\ =A0 =A0|--'Q4 =A0 =A0 | =A0 =A0|+

=A0PMV60EN | =A0 ---

=A0|>-. =A0 =A0 =A0 | =A0 =A0-

=A0 =A0| =A0 =A0 =A0 | =A0 ---

=A0 | =A0 =A0 =A0 | =A0 =A0-

=A0 =A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

=A0 =A016v =A0 |

=A0 =A0 =A0 | =A0 =A0 zener =A0|

=A0 | =A0 =A0 =A0 | =A0 =A0|

Isn't your circuit basically the same thing except that instead of discharging the capacitor in one cycle you're just chopping it up and increasing switching losses. The problem with circuits like this and the blocking oscillator is that you don't make allowance for the battery ranging from say 1.2V out to 24V and then look at your performance.

(...)

So then a "balance scale" (with the fulcrum point above the center of gravity of the beam) would tend to have relatively few stable angles when weighted equally on both sides then? These angles would tend to group tightly; with it's beam perpendicular to the axis of the local gravitational field?

Eventual pressure equalization? The larger balloon inflates to accommodate the mass of (air?) representing the difference between the mass that can be accommodated by each balloon at a specific pressure relative to ambient? Would there be some 'ringing' around the equilibrium point as the air chilled by expansion into the larger balloon heats via conduction through the walls? Would we see counterflow back into the smaller balloon, where the process would repeat until the pressure *and* temperature of the gas in both balloons would be fairly equal?

--Winston

Most references put the optimal load voltage at 0.45 to 0.48.

Current flow doesn't reverse in the inductor, so Il is still in the direction that pulls down the solar cell and its bypass cap. The circuit doesn't pump energy back into the solar cell.

John

Is the surface area of the smaller partially inflated (and therefore partially expanded) balloon greater than or less than the surface area of the larger balloon?

Ed

t. You

. JL is

served?

battery

ee

the

d by

the

ume a

lar

x C x

r x

o y

king

I meant your energy analysis takes no notice of the solar cell's MPPT. Sucking C2 down significantly greatly decreases solar cell output.

I don't think so--turning the FET on always sucks C2 down by more than the flyback charges it back up. If not, that's a disaster!

-

=A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 =A0 =A0 =A0 =A0 )|| =A0L1

=A0 =A0 =A0 )|| 100uH

=A0 =A0 =A0 =A0 =A0)||

=A0 =A0 =A0 =A0 =A0 =A0|

-------,

=A0 =A0 =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 =A0 =A0|

4093B =A0 | =A0 =A0 =A0 =A0 =A0 =A0|

, =A0 =A0 | BAT54 =A0 =A0 =A0|

=A0 +-->|---+----+

=A0 | =A0 =A0 =A0 | =A0 =A0|

|\ =A0 =A0|--'Q4 =A0 =A0 | =A0 =A0|+

=A0PMV60EN | =A0 ---

=A0 =A0|>-. =A0 =A0 =A0 | =A0 =A0-

=A0 =A0 =A0| =A0 =A0 =A0 | =A0 ---

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0-

=A0 =A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

=A0 =A016v =A0 |

=A0 =A0 =A0 | =A0 =A0 zener =A0|

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

a

Switching losses usually run 2-3% in most of my designs. That's recouped several times over by operating the solar cell at MPPT.

I^2R loss in the inductor usually makes it advantageous to minimize peak inductor current, which in turn implies continuous mode operation and lots of switching.

No, of course I absolutely look at that--the optimum circuit strongly depends on input and output voltages. Mine may not be absolutely optimum (especially at high output voltage), but it's simpler, smaller, cheaper, and a bunch more efficient than the original under all conditions.

I've designed a bunch of boost converters that run off

nt. You

k. JL is

nserved?

battery

ree

t the

ed by

the

sume a

e n o

olar

e

x C x

ar x

io

gy

C

aking

Right, good point.

James

Speaking of battery range, consider what happens if you put a freshly charged NiCd battery pack into that design, even if you limit the pack to say 14.4 volts.

The zener must either be protected, or rated above maximum battery voltage. The original shows a 16V zener across a 16V battery. Even with a 14.4 V NiCd pack that 16V zener can be in trouble, if the pack is installed in the solar charger immediately after being charged by a mains powered charger. At 12 cells, Vbatt could be over 17 volts.

Ed

For the purposes of the experiment two perfectly identical balloons but with different amounts of air in them. One with a small amount in and one nearly fully inflated and then connected together.

Regards, Martin Brown

ent. You

e g

ck. JL is

onserved?

e battery

gree

at the

sed by

o the

ssume a

he

e

en

to

solar

ge

2 x C x

lar x

tio

l

rgy

x C

making

, e *,

nd-

-

=A0 =A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 =A0 =A0 =A0 =A0 =A0 )|| =A0L1

=A0 =A0 =A0 =A0 )|| 100uH

=A0 =A0 =A0 =A0 =A0)||

=A0 =A0 =A0 =A0 =A0 =A0 =A0|

---------,

=A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 =A0 =A0|

CD4093B =A0 | =A0 =A0 =A0 =A0 =A0 =A0|

es, =A0 =A0 | BAT54 =A0 =A0 =A0|

=A0 +-->|---+----+

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

=A0 |\ =A0 =A0|--'Q4 =A0 =A0 | =A0 =A0|+

|| =A0PMV60EN | =A0 ---

=A0 =A0|>-. =A0 =A0 =A0 | =A0 =A0-

=A0 =A0 =A0| =A0 =A0 =A0 | =A0 ---

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0-

=A0 =A0 =A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

=A0 =A0 =A016v =A0 |

=A0 =A0 =A0 =A0 | =A0 =A0 zener =A0|

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

d e

o a

If you run C2 down during the burst mode so that the switching shuts down for a recharge then I don't see much advantage to it, it's just different, and if you don't run C2 down then the solar cell is saturated and operating at less than max current. The thresholds on CMOS Schmitts are wildly variable too so you really can't design anything precise around them.

ent. You

e g

ck. JL is

onserved?

e battery

gree

at the

sed by

o the

ssume a

he

e

en

to

solar

ge

2 x C x

lar x

tio

l

rgy

x C

making

, e *,

nd-

-

=A0 =A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 =A0 =A0 =A0 =A0 =A0 )|| =A0L1

=A0 =A0 =A0 =A0 )|| 100uH

=A0 =A0 =A0 =A0 =A0)||

=A0 =A0 =A0 =A0 =A0 =A0 =A0|

---------,

=A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 =A0 =A0|

CD4093B =A0 | =A0 =A0 =A0 =A0 =A0 =A0|

es, =A0 =A0 | BAT54 =A0 =A0 =A0|

=A0 +-->|---+----+

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

=A0 |\ =A0 =A0|--'Q4 =A0 =A0 | =A0 =A0|+

|| =A0PMV60EN | =A0 ---

=A0 =A0|>-. =A0 =A0 =A0 | =A0 =A0-

=A0 =A0 =A0| =A0 =A0 =A0 | =A0 ---

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0-

=A0 =A0 =A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

=A0 =A0 =A016v =A0 |

=A0 =A0 =A0 =A0 | =A0 =A0 zener =A0|

=A0 =A0 | =A0 =A0 =A0 | =A0 =A0|

d e

o a

It needs something like this where the upper comparator threshold is cell saturation voltage and lower threshold is somewhere around MPP.. Please view in a fixed-width font such as Courier.

. . . . . . . . +------------------------+ . | | . | | . | SD | . ---------+-----------LLLLLLLL----+--|>|---+ . | | | | | . + | | | | | . ----- | | | | . ---