Rosemary Ainslie circuit demonstration on Saturday March 12th 2011

[cHeeseburger]

Let's see what you get if you put in a shunt right at the negative battery terminal.  Hook the scope ground to the battery and the probe tip to the shunt so as not to reverse the polarity.

Then using probe #2 also grounded at battery negative terminal, probe the true battery voltage excluding all wiring inductances and resistors.

Then we can see what the real power is coming out of the battery.

Humbugger

[poynt99]

Now let's look at the average power in the critical devices. The "W" probe in PSPice allows for a direct probing of instantaneous power in any given device. By applying the "AVG" function to these traces, we obtain a trace which converges on the real power dissipated in that device.

Measured is the REAL power in the following:

1) Q2-Q5 combined: ~14.6W
2) Q1: ~0.41W 
3) Battery: ~-33.3W*
4) Load Resistor: ~16.86W

* Normal battery power is measured as a "negative" because this represents power being supplied to a circuit, i.e. a loss of energy.

If we look at the power balance, we have:

33.3W = 14.6W + 0.41W + 16.86W
33.3W = 31.87W

The remaining ~1.43W can be accounted for by the power dissipated in the 2 Ohm Gate resistor, and the 0.25 Ohm CSR. Oddly, the function generator contributes about 3W to the circuit, and this is precisely the amount lost in the 2 Ohm battery lead resistor. These losses are not shown in the graphs.

It becomes quite apparent in the simulations, that if the gate impedance is too high the continuous oscillations stop. I am convinced that the FG somehow provides for a low AC impedance path through it's output terminal, and this is why the circuit still oscillates. This would also explain why the FG does not heat up even though it has a 50 Ohm resistor in series.

I hope that with the above details and analysis, the "problems" in all this mess are now starting to become apparent. Surely it is obvious that Q1 is providing almost no power to the load resistor. One might also observe that the power measured using the CSR and Vbat voltage points is largely reactive, and not indicating anything close to the truth in terms of what the battery is supplying or receiving.

.99

[WilbyInebriated]

Seems odd

the first time i read this post just a bit ago, it read

"  'nul-points = Lame indeed!  "


But no indication that it was edited

Selective edit notifications?

Mags

if you edit quick enough, say in the first five or ten minutes after posting, it won't show an edit. not sure if it's a forum feature or a bug...

[poynt99]

In reply #780 above, in the scope shot "Q1_scope03.png", the battery power was calculated using the average of Vbat x Vcsr.

It should be pointed out that this normally would provide the correct answer. In this case MEAN (Vbat) is the correct value, i.e. ~71V, but the value for MEAN (Vcsr) is incorrect. The reason for this as was pointed out above, is because the majority of current flowing through the load resistor and battery is doing so via Q2-Q5, and not via Q1 where the CSR is in fact connected.

As a result, the CSR will not indicate the true current through the load and battery. As a side note, Hum suggested that the CSR should be placed outside of the Vgs loop of the MOSFET Q1, and I fully agree. The results of this simulation also offer support for this. The CSR would indicate (and Rose would jump in glee) that the currents in the load are upwards of 6A in this case (1.5Vp/0.25), but in fact the average current from the battery is less than 0.5A.

.99

[Magluvin]

Thanks Wilby
Things like that can just get the imagination going for sure.

Mags