Hi SkyWatcher,
This post will be a long one...
You wrote
"If i remove the prebias and it still oscillates, do you think it will be more efficient."
Yes it will, how much more, this can be estimated only partially because the collector-emitter waveform is
not known: removing just the 23 kOhm, the dissipated power in it could fully be saved. But this surely involves
the readjustment of the 15 kOhm resistor too. Sometimes a transistor with such low h
FE is difficult to use in a
self starting oscillator, higher supply voltage helps on this.
First consider the base current via the 23 kOhm:
(80.2V-12.6V-0.7V) / 23 = 66.9/23 = 2.9 mA and the dissipation in this resistor is 66.9*2.9 = 194 mW this would not
be consumed from the power supply. Also, the transistor would not have an idle current during the oscillator OFF times,
so this consumption would also be missing. 2.9mA*h
FE would be the idle collector current, let's say the h
FE would be
8 with your tv transistor, then collector current, Ic would be 2.9*8 = 23.2 mA.
At OFF times the collector emitter DC voltage may be 80.2-12.6=67.6V and with the 23.2 mA collector current, the
dissipation would be 67.6*23.2=1568.3 mW BUT because we do not know the duration of the OFF time during the
periodic oscillations we cannot reduce the 1.5W by the actual duty cycle in this oscillator to get the average dissipation,
the 1.5W dissipation would be correct for a continuous operation, not for an oscillating operation.
So you would need a oscilloscope to see the collector-emitter waveform. Now if I assume the average dissipation were
say 1/10 of 1.5W, then this 150 mW + 194 mW= 344 mW would be saved when you could omit the pre-bias resistor
due to the use of a transistor that has a higher or high h
FE value.
To estimate how the saturation voltage for the transistor during the ON times may influence efficiency, we would need
to know the ON time of the oscillator when collector current would flow, this would cause a certain dissipation in the
function of the saturation voltage and of course in the function of the collector current. So again the voltage waveform
across the collector and emitter would be needed.
Ideally, the ON time for the oscillator transistor should be as short as possible. This is because of the value of the
L/R time constant for your coil has 'something' to do with the DC resistance loss of your coil and also with the recovery
loss of capturing the flyback pulses. IT is very good you use multiple wires in parallel for the main coil, this automatically
reduces wire loss.
You would need to know the L inductance of your coil that has the 272 milliOhm resistance. Suppose, for simplicity,
it has 27.2 mH inductance with the ferrite tube cores. This would give an L/R time constant of 0.0272/0.272 = 0.1 sec
i.e 100 ms. What this means and how the ON time could be chosen to this constant to get the smallest coil loss, please
read member 'verpies' post here:
http://overunity.com/7679/selfrunning-free-energy-devices-up-to-5-kw-from-tariel-kapanadze/msg352483/topicseen/#msg352483 If you do not understand something with his text, I try to help.
Notice that if you remove the ferrite core from the coil, the inductance goes down hence the L/R also changes to a lower
value too, so that one would have to reconsider the ON time BUT the question is how the ON time can be controlled?
Well, in this circuit, this can hardly be done (to end up with less coil loss) because of the 'rigid' (not readily controllable)
feedback in the oscillator. A good solution would be to use a very low power variable duty cycle oscillator (like a CMOS
555 timer) to drive the base of the transistor and seek for the best efficiency possible.
Needless to say, that even with such duty cycle optimization, one has to face a trade-off because the too short ON time
(that yields less and less loss and involves a decreasing amount of input power) goes together with recovering less and
less captured flyback energy from the coil (smaller input current involves less stored energy in the coil).
In this situation the only means to reduce this trade-off is to increase input supply voltage to increase input power hence
increase the amount of the captured energy too.
What you observed as an increasing efficiency when you increased the supply voltage was very probably due to a near
ideal ON time - coil time constant relation to have a minimal coil loss. This you achieved with increasing the time constant
by reducing the R resistance of the coil i.e. you reduced the denominator in the L/R quotient and you inserted a ferrite core
into the air cored coil, this latter also increased coil time constant.
So if you remove the core, you go towards the higher coil loss possibility, unless you are able to adjust the ON time too to
partially compensate for that. (In this case you reduce L hence time constant also reduces.)
I agree with your opinion on what the best configuration would be and the thoughts above may help achieve or approach
that. Probably you would still need the core for the coil but you need to reduce a little the inductance wrt the 80.2V supply
voltage position by using less number of cores and / or inserting them partially and thus reducing supply voltage too.
Or you may find the 170 mA input current for the coil is quasy very good for the charging, then try to stay near to it but
remove some number of turns till you approach similar input current range with the ferrite cores inserted.
(This latter helps increase time constant hence reduce loss provided the ON time remains nearly similar for the cases.)
And try to find a switching transistor having a higher h
FE than the present tv transistor (the ZTX type would be ideal in
this respect, together with its very low saturation voltage).
Gyula
