Joule Thief 101

[MileHigh]

Brad, your graph is kind of hard to interpret, because your grid lines don't line up with the axis number values. But going just by the positions of the numbers...

Well, there you go.  I actually took it for granted that there would be no issue with the grid lines and the values assigned to the grid lines because I have looked at 1000 graphs before and that was never an issue.  So I didn't even bother scrutinizing that part of the data.  It's an almost unbelievable offense to good practices when presenting data.  Both axes are messed up.

And we can assume that poor Brad is going to put on his thinking and strategy cap and come up with a very creative excuse that puts him in the clear because he is going to cling to some real or imaginary technicality.

[minnie]

tinman,the little r for a 1.5 v AA alkaline cell is about .2.
It should be capable of 7.5 amps.
       John.

[tinman]

Brad, your graph is kind of hard to interpret, because your grid lines don't line up with the axis number values. But going just by the positions of the numbers, it looks like your DC-equivalent value is indicating about 200 Lux at an input power of 20 mW, which translates to a whopping 10000 Lux/Watt.  But in my setup, using the superefficient LED that I've been using all the time, to get 200 Lux at the sensor in my 5x5x18 inch lightbox I have to drive the LED at around 81 mA, 2.96V, which is about 240 mW, and is near the top end of its rating. This corresponds to about 830 Lux/Watt.  For any length of time running this would need a heatsink on the LED. These are just about the most efficient LEDs on the market.

So it seems that there must be something funny about the numbers, either yours or mine, because they are clearly not comparable. Are you using an 18 inch lightbox? How are your 9 LEDs connected, and what kind are they? How are you measuring voltage and current in the DC case?

I have some confidence in my measurements because, as I indicated before, I developed this apparatus and methodology for comparison with values that another lab was getting in some tests we both were running.

I can get around a 200 Lux reading by holding my LED at 4-5 inches distance to the lightmeter sensor and providing about 7.5 mW (3 mA at 2.54 V).

TK

I am still using my small 8 inch light box,as i still have not finished the larger one base around your dimensions,as the paint takes a long time to dry,when the temperatures are around 10*C here ATM. As far as the nine LEDs go,i am not sure how they are connected,as the small PCB they are mounted on is all glued into the ali housing of the torch body,and i cant remove it.

The light beam is focused toward the light meter sensor,and so that would be the reason why the readings are so high-i would think.Once i get the light box all done,and switch to an LED such as yours,i would think it will all drop down to close to your values.

As far as the graph go's,it was just a quick throw together job on windows paint,as i do not get a lot of time after work to get all fancy,but it dose give you a general ideal,and by following a horizontal axis from the lux number,and a virtical axis from the Mw value,each dot represents the results obtained from the tests. I used a 1 0hm CVR to measure current,and the voltage was set at 1.5 volts from my PSU,and also measured with a DMM across a 10 000uF smoothing cap.


Brad

[tinman]

tinman,the little r for a 1.5 v AA alkaline cell is about .2.
It should be capable of 7.5 amps.
       John.

And when the battery voltage is at 1v,little r is how much ?

Brad

[TinselKoala]

TK

I am still using my small 8 inch light box,as i still have not finished the larger one base around your dimensions,as the paint takes a long time to dry,when the temperatures are around 10*C here ATM. As far as the nine LEDs go,i am not sure how they are connected,as the small PCB they are mounted on is all glued into the ali housing of the torch body,and i cant remove it.

Ah.. OK, they are probably all in parallel then. You could test that by looking at the voltage required to turn the lights on, should be the same as for a single separate LED, rather than twice or three times that much. Sometimes these setups even include a small current-limiting resistor tucked away on the PCB.

I have a 24 LED array that came from a cheap LED flashlight that used 3 AAs,  that was originally all LEDs in parallel, with a tiny resistor, but I rewired it to be 2 series arrays of 12 parallel each and ditched the resistor. I use this for some JT demonstrations (but not of course for our testing here.)

The light beam is focused toward the light meter sensor,and so that would be the reason why the readings are so high-i would think.Once i get the light box all done,and switch to an LED such as yours,i would think it will all drop down to close to your values.

Yes, I'm sure that explains it, the focussed beam and the close distance. That's the problem with using Lux, it is a measure of the illumination falling on a surface rather than the output of the light itself, but that's all we have to work with.

As far as the graph go's,it was just a quick throw together job on windows paint,as i do not get a lot of time after work to get all fancy,but it dose give you a general ideal,and by following a horizontal axis from the lux number,and a virtical axis from the Mw value,each dot represents the results obtained from the tests. I used a 1 0hm CVR to measure current,and the voltage was set at 1.5 volts from my PSU,and also measured with a DMM across a 10 000uF smoothing cap.


Brad

The 1.5 volts must be for the JT measurements, not the single DC equivalent point, right? As the LEDs in your 9-led array will not shine with only 1.5 volts DC, right?

Anyhow, I figured that the numbers on your graph were in the right place and the grid lines were misplaced, so it didn't really cause me much grief. But really, for great ease in recording and organizing data and turning it into graphs, you could try using a spreadsheet program like Excel or (free and fully functional) LibreOffice Calc:
https://www.libreoffice.org/download/libreoffice-fresh/
(scroll down for the Windows or OSX version installers)
Once you've gotten the hang of graphing using Calc (or Excel), you'll never go back.
The really nice thing about using the spreadsheet program is that if you change your data once you have the graph set up, like to correct an error or just to see how it affects the graph, the graphs change instantly and automatically to show the change. And once you have a graph set up like you like it, to graph a different data set you just need to copy-paste the original graph and then point the copy to your new data set (and edit the titles and other names of course). That's how I did the three graphs I posted up above. It's a snap.