|
Feynman Joule Thief Experiment 001: Frequency Drift vs Circuit Placement
Okay guys, so I'm still setting up my lab, but this was a quick experiment I worked up to make use of my frequency counter.
Materials: I'm using the standard Joule Thief setup with a 2N3904 NPN transistor, a somewhat-new 1.5V AA battery, a trifilar wound 3cm ferrite toroid, with one of the windings left 'floating' and the other two windings in standard bifilar JT configuration. This matches the professor's circuit , minus the CSR resistor. This setup is also using a blue LED with a draw of approximately 30mA and it is very bright. No resistor or capacitor is connected in this JT. The frequency counter I used was a Vicor VC3165 , capable of 0.01hz - 2.6ghz measurements, measured with channel A and a 50% gate time (about 10-15 seconds).
Methods: I tested the circuit to see how time and position it's resonant frequency, as measured at the collector/emitter of the transistor. My previous rough scope calculations based on counting squares put it in the vicinity of 8khz. The more-accurate frequency counter measurements were closer to about 7khz. I used a 10-15 second sampling interval on the frequency counter, which then updated and displayed the mean frequency for this interval. In any case , I did this:
1) Connected up Joule thief and started taking notes. Noticed the freq counter number 'drifted' significantly over a few minutes, from about 7.013khz down to 7.009khz, then up to 7.043khz (<1% deviation). This placement of the circuit was next to my laptop. I decided to run two tests: one with the JT on the floor, and one next to the laptop.
2) I took the still-powered on Joule Thief and placed it on the floor , approx 1m from any potential localized EMI from the laptop / cell phone etc. I observed the frequency every 30 seconds and recorded it for the graph for 5 minutes, which is displayed as 'ground'.
3) I then picked up the still-powered on Joule Thief and placed it back next to my laptop on my desk, in the vicinity of both my laptop as well as the frequency counter and cell phone. Again, I recorded 5 minutes of measurements.
Results:
See attached graph.
-The 'floor' frequency was significantly lower than the 'table' frequency . I suspect this is experimental error from jostling the circuit when moving it to the floor, as you can see by the circuit beginning to creep back up to 7khz over the time interval. Further tests not included in this study show 'floor' frequency can be as high as 7.2khz, so the likely culprit was physical trauma to the circuit changing its resonance.
-Joule thief may be susceptible to localized EMI from laptops, but the effect is less than that of physical changes (movements) to the circuit configuration.
-The JT graph when it was placed near EMI is subjectively more volative, but more tests would be needed to confirm this effect.
-In both high-EMI and low-EMI conditions, Joule Thief experiences significant frequency drift.
Conclusions:
Joule Thief has a significant frequency drift, with potentially unexpectedly long period of oscillating drift (the drift period, if it exists, is measured in minutes, and may be aperiodic, or perhaps more complex, possibly even a mathematical attractor.) The drift is approximately +/- 1% of the fundamental frequency, and may be as high as 2%. Subjectively, physical movements of the JT circuit and apparatus appear to have more frequency-related effects than to localized EMI, but we do not have enough information to make definitive conclusions yet.
Future work may include the testing of Joule Thief within a Faraday Cage to minimize the effects of transverse EMI, as Faraday Cages are only permeable to Scalar EMI. This test would apply to any OU / Out-of-phase operation conditions as well.
Other work perhaps should also include a bifilar wound (rather than trifilar) toroid, as well as perhaps grounding the third winding to itself to minimize inductive effects of localized EMI.
Another important question that remains to be answered is does standard Joule Thief have a resonant frequency, or its resonant frequency a range depending on outside factors (temperature, EMI, internal effects, etc)?
The most significant finding is that Joule Thief has internal frequency drift which appears somewhat independent of local EMI.
-Feynman, 3/15/2011
P.S. I left the device running untouched on the floor for around an hour as I posted on overunity.com, and the frequency climbed up to 7.310 khz. So there's definitely at least a +/- 1% percent drift here. It began pouring rain, so perhaps humidity and/or RF attenuation is having an effect. . . Or perhaps the oscillator is simply a nonlinear dynamical system with chaotic features (sensitivity to initial conditions, density of periodic orbits, etc). Either way, it's important to know it's properties moving forward with this and other open-source, potential overunity projects.
« Last Edit: 2011-03-16, 05:10:17 by feynman »
|
feature is available which provides a place all members can chat, either publicly or privately.
Author
Topic: Possible breakthrough with the JouleThief (JT) circuit (Read 80658 times)
