Joule Ringer!

[JouleSeeker]

     I have continued experiments with the SJR 2.0 using an air-core xformer (my thanks to LynxSteam -- good air-core design!).  Note that this is the straightforward 2.0 of Lasersaber and Lidmotor -- there is no bias from the + rail to the base.  Yet it generally starts right up and runs well.

  I need to note that since my last report on this air-core 2.0, I managed to short something out and thus I fried a transistor.  This happened while I was trying to light a bare CFL (i.e., guts removed, just the lamp).  So I had to replace the 2n3055.

PS -- the bare CFL did light up, but not brightly.  I much prefer the LED lamps and I've tested a bunch.  Of these, I still prefer the 1-watt "corn-cob" LED lamp from T-mart that I delineated earlier.  That's what I used today (first one bulb, then two).

  When I replaced the transistor with another 2n3055 that I have, I could no longer get the 53 Lumens/Watt that I discussed earlier.   It is possible some else changed, like connection - resistances; but I tried to keep things constant while I swapped out the 2n3055.  But, by moving the primary windings around a bit, I found today a maximum of 48 Lm/W, which is pretty good I think.

    The experiments today were run mostly at 10 Volts input from a PS.  I had already determined with this set-up that I got optimal light from about 8.5 to 11.5 volts.  10 V DC input makes calculating the input power easy...

  The photo shows 12-gauge wire on the primary concentrated at both ends of the tube.  I had taps on the red wire where I had removed some of the insulation.  I counted the windings and tapped the + voltage in at winding 31, 35, etc.  Note that with fewer windings on the primary, the output current (and voltage) are higher as expected, but the ratio Lumens out/Watts-in = Lm/W is not necessarily larger.  My goal remains to increase Lm/W.

   I'm using the calibrated light-box for measuring Lumens, with the calibration factor 0.08 Lumens/observed-Lux for my box as delineated in an earlier post.

Here are salient results from today's experiments:

One LED bulb, with wires concentrated at left end, tapping at winding number

14   0.8A  2000 lux --> 160Lm, 160 Lm/8W = 20 Lm/W
31  0.60A   2030 lux--> 162 Lm, 162 Lm/6W = 27 Lm/W
40  0.19A  720 lux--> 58 Lm, 58 Lm/1.9W =   30 Lm/W
45  0.10A  310 lux--> 25 Lm, 25 Lm/1.0W = 25 Lm/W

I'm going to shorten the straightforward part, now that you see the pattern, to save typing time...
One LED bulb, with wires concentrated at both ends as in the photo attached, tapping at winding number

31   0.54A  152Lm,  28 Lm/W
35   0.45A  132Lm,  29 Lm/W
37   0.23A    78Lm,  34 Lm/W
40   0.25A   86Lm,  34 Lm/W

One LED bulb, with wires EVENED OUT along the length of the secondary coil; tapping at winding number

31   0.66A  240Lm,  36 Lm/W
37   0.61A  290Lm,  48 Lm/W   154 Khz on the DSO, 58 Vrms (output)
40   0.45A   169Lm,  39 Lm/W

Note the difference in Lm/W with the number of windings (tap) on the primary coil.

TWO  LED bulbs, same type, 1W 220V-AC bulbs, with wires EVENED OUT along the length of the secondary coil; tapping at winding number

31   0.74A  276Lm,  37 Lm/W
37   0.66A  291Lm,  44 Lm/W   72 Khz on the DSO (output)
  Note the drop in resonant frequency with 2 LED bulbs instead of one.  Lm/W drops a little/or roughly the same with 2 bulbs.

40   0.48A  192Lm,  40 Lm/W
45   0.39A  144Lm,  37 Lm/W

Conclusions:  windings spaced roughly evenly on primary is best; tapping at winding 37 for this set-up provides maximum, 47 Lm/W (at least as high as winding 45 -- I did not test higher as the Lm/W was dropping).

[SeaMonkey]

I've just done some Spec Sheet checking on various LED Bulbs/Tubes and
some LED Lamp Driver Circuits.

The input voltage range specification is actually very broad and some
are capable of being powered by both DC and AC.

The listed efficiency of the various driver circuits is >80% which is
reasonably good for a commercial mass produced product line -
but, it could be much better.

Has anyone yet opened up an Led Lamp to investigate the driver
circuit?  With CFL drivers the input is a full wave voltage doubler
and I'm wondering whether the LED Lamp drivers are similarly
configured.

The capacitors in the voltage doublers of CFL drivers are quite
small to enable Power Factor Correction and the drivers are able
to be powered over a fairly broad voltage range as well.  CFL
drivers do not tolerate DC well though - in time the "unused"
capacitor in the voltage doubler will fail.

It takes a pretty high voltage to "strike" a fluoresecent lamp tube
when powered directly from a transformer winding.  For maximum
efficiency a ballast capacitor should be used in series with the
fluoresent lamp tube when AC operated from a high voltage
transformer winding.

[conradelektro]

I did not / do not plan to post here.   I Like your work Conrad and am sorry your upset.  However, I'm not sure how you connect what I say to saturation.  I've learned some about that coil and it's rated for 380va.....its NOT saturating.  I don't believe the "Shacks" are either, as we are well below rated current on secondary (de-rated for higher fo). 
   Now as we approach higher power levels in a given bulb, is the core IN the bulb saturating and losing efficiency....very possible.  I think, at this point we can get an improvement if we can reverse engineer these some. 

@Peanutbutter:

The crucial experiment:

Take an ordinary incandescent light bulb (it has no internal circuit, just a filament in a glass bulb) e.g. a 25 Watt or a 40 Watt. Put this incandescent light bulb into the circuit (where you normally would put the LED-bulb or the CFL) and watch the power drawn by the circuit.

The power draw does not go anywhere near 25 Watt or 40 Watt, it stays at what I think is the "saturation region" of whatever transformer (defined by core type and number of windings for primary and secondary) one uses.

In my opinion, one crucial point is the number of turns for the primary (which then of course determines the number of turns for the secondary in order to reach a certain step up ratio). The more turns for the primary, the more power can be transferred. But the size of the core limits the number of possible turns. So, a bigger core might be helpful for more power.

Also the frequency is very important, lower frequency allows more power to be transferred. And this circuit runs at a few Kilohertz, so we are limited by the rather high frequency.

I have to dig out some book about transformer design and do the mathematics. The transformers taken out of power supplies are all optimised for 50Hz - 60Hz and the Joule Ringer works at a few Kilohertz. And in case of air cores up to a few 100 KHz, I observed 400 KHz with smaller air core coils.

If a core is rated for higher frequencies (like my E-core for FERROXCUBE) it will support higher frequencies up to 30 KHz, but power transfer is still smaller for higher frequencies than for lower frequencies. Take into consideration that 50Hz is 100 times less than 5000 Hz. So, the power one can transfer is 100 times less at 5000 KHz than at 50 Hz. So, the power supply you took the core from was able to provide 1000 Watt with a 50Hz input, and with the 5000 Hz input you can only reach 10 Watt.

One can see the "saturation" much easier with low inductance cores for pulse transformers (e.g. ferrite grade 3E27). When I used such cores, the power transfer was limited to one or two Watt (similar to the air cores).

The factors which determine the Wattage possible to be transferred:

- core type (material of the core)
- core size
- number of turns for primary (which determines the number of turns necessary for the secondary)
- frequency

And it is not easy to do the mathematics, specially "core type" and "core size" are hard to factor in. Also the number of turns is not as trivial as it sounds when doing the mathematics, because the diameter of the wire and the changing diameter of the windings play a role too. It all comes down to the fact, that my knowledge for doing the correct mathematics is limited. But the tests give a good indication of the limits of power transfer.

This said, I do not want to discourage people to work with this circuit. Just look carefully before drawing conclusions. Read through the posts and you will see that all experimenters observe this "limits for power transfer", specially with air cores (because they have low inductance). The "magic" of this circuit is, that power draw does not rise after a certain number of bulbs is connected. So, power transfer is limited.

Greetings, Conrad

[JouleSeeker]

The latest results may enter into the discussion of saturation.

The experiment I posted above showed quite a lot of Lm/W, but it gets better and gives me hope...  The rating on most LED bulbs is approx 50-65 Lm/W typically.

  First, I went to 9V instead of 10, got nearly the same as at 10 V -- 49 Lm/W @ 170 KHz on the output, air-core.

I had two FERRITE RODS lying around, 20 cm long and approx 9mm diameter.  I simply placed these inside the air-core xformer, laid them in there.  Result:

One LED bulb, with wires EVENED OUT along the length of the secondary coil; tapping at primary winding number 37; 2 ferrite rods lying INSIDE the inner tube:

9V  0.22A  128Lm, [SIZE="5"]64 Lm/W[/SIZE] 111 Khz on the DSO (output)

Taking my dear wife to airport -- gotta run!

[Lynxsteam]

Great work JouleSeeker!  64Lm/watt.  How much do you want?

I am glad the discussion and exploration continues.  The "Magic" is partly that we are exploring, trying different things and having fun doing something useful.  We could all be drinking beer watching basketball (nothing wrong with that).

A couple notes: 
I can't get anywhere near beating the E-Core with the Aircore (to be expected). 
There are some resonant points on the aircore that give really good results
Without the HV hooked up to a load I got quite a shock off the 12v positive lead while hooking up the aircore.  I am pretty sure the battery is part of the LC tank circuit and its charging between cycles.  Put a diode on the negative side of the battery and the circuit doesn't work.  If this is the case the battery may also be part of the limiting factor and part of the tuning.  I get better results with a large lead acid 12 v rather than a small nicad pack.  If the oscillations occur without the LED bulb then perhaps the bulb is not the big factor in the circuit but more along for the ride.

I am where Joule Seeker is with playing with ferrite, but with an enclosed magnetic field around the aircore.  This then is no longer an aircore.   I am playing with completely enclosing the coils to see what difference it makes.

My goal is to get total power up so that the ten bulbs are closer to fully bright and that we maintain the self adjusting ability of the circuit.  I want to avoid core losses and eddy currents in the core material as much as possible.

PB291 says the core is not saturating, Conrad says yes it is.  Could it be the Battery resistance is the limiting factor and confounding everyone's results?  Batteries have resistance and it is very difficult to measure.  It is usually between 1-4 Ohms.