user TURBO?s replication of Steven Mark?s TPU ?

[starcruiser]

Otto,

I was looking at the Marckovich design and I am thinking the TPU is based on this with a slight modification. Take alook at the link posted on page 52 near the bottom. Two things that may help you increase the current output ( I am going to try this as well, but you are probably in a better position to try this sooner) for the feedback coil increase the wire gauge to something like 16 or 14 gauge and the other is to try and reverse the winding direction in relation to the control coils.

When you look at the Marckovich design you may see what I am talking about. I am thinking the reversed direction will cause a flux cancellation that will increase current flow. The reasoning for the heavier gauge wire is 2 fold, the load will be pulling from this coil as well as the collector coils and I am thinking we need a 3:1 ratio from the control coil windings to the feedback or secondary windings.

Another thing I just thought of is if we resverse the windings of the outer coil the reversed polarity of the control coils may become un-necessary.

Also, the additional capcitance from the Fets may be providing the required capacitance in the Markovich design (look at the top of the drawing). We also may find the need for only one frequency (injected at the top where the capcitance is located). This would mean the simple transistor oscillator would be the only signal source required since it would contain the harmonics we are seeking anyway. The addition of additional frequencies may just improve the performance. More testing eh?

Take a look at the diagrams I mention and let me know your thoughts regarding my theory.

[Groundloop]

[EDIT] Deleted.

[Vortex1]

I highly recommend if you can afford $30 USD get a KILL-A-WATT line monitor(by P3International.com) for your power supply. I picked one up on impulse and was impressed with the device's 0.2% accuracy. It also computes volts, amps, watts, frequency, VA, and most importantly power factor and total kilowatts vs.time. You can put this on the line side of your power supply and get a baseline unloaded. Then hook up your TPU and drivers, you can determine your power drain in watts or kW/Hr by subtracting the baseline (unloaded power) .

For power measurements (if using a light bulb or other resistive load) on the output of your TPU, I recommend using two thermocouples inverse series connected, one to measure light bulb envelope temperature, and one to subtract and compensate ambient temperature. The output of the thermocouples goes to your millivoltmeter. Calibrate the readings using known input from your variac at several points. Since the load is mainly resistive you can use amps X volts for your calibration of output watts without needing to compute PF. Once the load (light bulb) is calibrated you can hook it up to your TPU and look up power on your mV table. Keep the compensating thermocouple shielded from light and heat from the bulb.

With these two devices you can make fairly accurate Pout vs. Pin measurements.

No need to guess.

[Groundloop]

[EDIT] Deleted.

[Vortex1]

Or you can use high speed diodes and a capacitor on the output of the TPU. Then run it to a load and use Ohms law.

Some devices/oscillators might get finicky and refuse to start into a capacitive load, although generally the diode/capacitor scheme will work.

Even a lightbulb has considerable inductance at high frequencies. Best to use carbon composition (slug) resistors array or non-inductively wound resistor.