Zoltan's Simulation

[poynt99]

MileHigh, Hoppy...

Any comments?

Why the big (and strange) discrepancy in voltage and wave form?

.99

[poynt99]

Core problem?

The coil core I used was a 3E6 rather than the 3E5 Zoltan specifies.

This may explain the discrepancies?

The two cores are identical in dimension, but the 3E5 has a permeability of about 8,000, while the 3E6 is about 10,000.

My understanding is he operates the 3E5 core just before saturation, which would mean I might have to decrease the operating current in my version.

.99

[poynt99]

After having examined Zoltan's calculations on the one graph, it appears there are likely other problems causing the amplitude discrepancies.


.99

[MileHigh]

.99: I looked at your simulation and it is correct. This is a clear demonstration of two different exponential waveforms for two different time constants like I explained on the other thread. The time constant for an RL circuit is L/R. In this circuit the "R" changes back and forth between roughly 48 ohms and 830 (?) ohms.

In your simulation, when the MOSFET is on, the inductor is charging through a roughly 48-ohm resistor (whatever 830 (?) is in parallel with 50), and therefore the time constant is quite long. What looks like a straight line with a low slope at the bottom of the graph is actually the beginning of an exponential waveform with a very long time constant, so it looks like a straight line in this little snippet. When the coil switces off, all of a sudden it finds itself "charging" through an 830 (?) ohm resistor in your simulation. This means the coil voltage instantaneously jumps up to whatever the higher voltage is in your simulation because now the same current flow is going through a larger resistor. The coil finds itellf at a higer voltage than the battery voltage and therefore starts to discharge. Now the time constant for the exponential waveform has changed drastically and gotten much smaller because the load resistance is now much larger, 830 ohms. Thereore you see the much faster exponential drop on the second part. When the MOSFET switches back on, the load resistance changes to a much lower value so the voltage steps down instantly from half-way up to the much lower level. The voltage ratio between before and after the jump will be roughly 830/48. The cycle then starts all over.

I looked at the PDF and the number crunching is not credible. I assume that his non-standard models explain his unusual waveforms.

The bottom line is that your simulation is correct and this can be easily verified by anybody on the bench if they want to.

It has nothing to do with the core permeability.

[poynt99]

Thanks for the reply MileHigh.

I find it very strange though that the wave form seems swapped from left to right. It does not make sense to have such a difference such as this.

There are some strange things going on in his simulation indeed. I can't imagine that he would have made some custom models, but if he did that may explain a few things.

And yes, I agree that his calculation method is unconventional indeed. I hope Zoltan appears so he can address these issues first hand (moderator permitting).

I'll post the correct (as output from PSpice) input and output powers and COP next.

.99

[MileHigh]

.99: Look at your waveform and read through my text and hopefully it will make sense. When the MOSFET is on, the inductor is charging through both resistors. When the MOSFET switches off, the inductor discharges through the 830 ohm resistor and the voltage source.

Try changing the value of the 830 ohm resistor up and down and rerun the simulations and see what you get. Go back to my text again and see if it clicks.

Then try changing the value of the 50 ohm resistor and see what you get.

[poynt99]

.99: Look at your waveform and read through my text and hopefully it will make sense. When the MOSFET is on, the inductor is charging through both resistors. When the MOSFET switches off, the inductor discharges through the 830 ohm resistor and the voltage source.

Try changing the value of the 830 ohm resistor up and down and rerun the simulations and see what you get. Go back to my text again and see if it clicks.

Then try changing the value of the 50 ohm resistor and see what you get.

Of course your explanation makes perfect sense, I had no problem with that. Thanks.

I am having difficulty explaining Zoltan's wave forms.

.99

[poynt99]

Zoltan's Done it!

Zoltan's a genius!

Power dissipated in R1 (the load): 7.904mW RMS

Power supplied from V1 Source: 5.678mW RMS

COP = 1.39



.99

[Joit]

Now that is looking gooood

[MileHigh]

.99: I suggest that you double-check your data. During the first half of the cycle the energy in the coil is increasing. If you calculate the end-current energy and subtract it from the start current energy (after the waveform has stabilized) then you will get the energy increase. Then for the second half or the waveform the calculate start-current energy minus the end-current energy for the energy decrease. The difference between the two will give you the net energy change in the coil per cycle, which should be zero after the waveform has stabilized.

Note that the current in the coil never reaches zero after the waveform has stabilized. That implies that there is a continuous dissipation of power through the resistive loads.

For your calculations, are you sure that you are averaging your power over the whole time period of the waveform? For R1, the power calculaton is a little bit complicated. Very low power is going through it during the charging cycle. That has to be averaged over one full cycle. Then higher power is going through it during the discharging cycle, and that has to be averaged over one full cycle. Then you add the two averages together to get the power through R1.

[poynt99]

This is a very simple and straight-forward analysis.

No "1000 times" factors or unconventional and strange mix of "beginning" and "end" values were used such as what Zoltan did. Zoltan used "Average" values and that is also incorrect.

The proper analysis is simply the RMS power dissipated in the R1 load compared to the RMS power delivered by the supply.

Only the "end" values were used. A cursor is placed at some point in time near the end of the curve, then the two values for power are read off for each curve. After this it is a simple division to get the COP.

Note that the rising curve IS partly caused by the circuit settling, but it is MOSTLY due to the calculation process within the program to obtain the RMS value. The more cycles it uses, the more accurate the result. I could have ran the transient analysis for much longer, but that is unnecessary. It is quite obvious where the two curves are heading and one is quite clearly higher in amplitude than the other.

.99

[MileHigh]

.99:

> Only the "end" values were used.

Just a few questions for clarification. I don't undertstand that graph with the two curves that you posted because all of the text is scrunched down and can't be read. For the "end" values, and the RMS power calculations, are you implying that you have two vertical cursors, "start" and "end" that you place at the start and end of the full period of the waveform? Then you can ask PSpice to compute an RMS power over that period? Is PSpice the freebie or if not how much is it?

Thanks,

MileHigh

[poynt99]

.99:

> Only the "end" values were used.

Just a few questions for clarification. I don't undertstand that graph with the two curves that you posted because all of the text is scrunched down and can't be read. For the "end" values, and the RMS power calculations, are you implying that you have two vertical cursors, "start" and "end" that you place at the start and end of the full period of the waveform? Then you can ask PSpice to compute an RMS power over that period? Is PSpice the freebie or if not how much is it?

Thanks,

MileHigh

MH (hope you don't mind the abbrev.),

It's unfortunate that attached pictures are scrunched down in size by the forum. I assure you that these pics I posted were much better resolution.

I'll post these pics in a PDF so they can be downloaded.

There is a free version of Orcad with PSpice, and the only limitation is the number of components. Also, it will not have the non-linear cores included.

LT Spice (Linear Technology website, I already posted the link in the RA thread) is free and it will have core models. I would recommend that one, or download the full PSpice package somewhere.

In short, NO. I am only using one sample in time. It is not only not necessary to use a "range" of values, but it is incorrect. One single cursor indicating the RMS power at precisely the same point in time for the load and the source...after much settling time. This IS a periodic wave form so eventually sampling one instant gives you the readings you need.

This is synonymous to taking a temperature reading on the resistor after a period of time (or measuring power electronically), while electronically measuring the power delivered by the supply.

.99

[MileHigh]

.99:

The abbreviaton is fine.

Yikes, let me just review an RMS calaculation in a circuit simulation package. This will be pretty basic because there are other people reading this thread.

Supposing that you sample the current waveform over the full period as 10,000 separate sample points. Therefore the computer has a list of 10,000 current sample point numbers in memory that define the waveform. Each sample point represents a certain number of amperes of current going through the load resistor.

To calculate the RMS current, the software first calculates the square of each sample value. Then it adds up all of these numbers and divides by 10,000 to get the average "squared sample" value. Then it calculates the square root of this "squared sample" value to give you the final value for the RMS current.

So, by definition, you have to calculate the RMS current or voltage over a certain time span, which is typically the period of your waveform. I have to assume that your package can do this.

MileHigh

[poynt99]

MH,

Some of the pics.

Note the last one is the one used for the manual COP calc. Clearly you can see that the R1 Power is higher than the Supply V1 power.

.99