rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Jun 14, 2007 19:25:16 GMT -8
Imazoid, I'm getting a lot closer to accurately modeling your circuit. The problem I am having is with the transformer. Not any transformer will do! Correct me if I'm wrong , but you are dealing with a nonlinear core model. In order to select the correct core model in pspice I need a reasonable figure for (1), AREA (the cross-sectional flux area in cm2 (2), PATH (the magnetic path length in cm (3) GAP (air gap length) (4) type of core (ferrite pot?) (5) number of turns for each winding. I don't need the inductance measurement for this type of transformer. Could also use MS /magnetic saturation in gauss (500 guass is typical) With a reasonable estimate of the above parameters, your transformer can be modelled. It might have to be tweaked a little to match your measurements, but it would be valuable in simulating the circuitry that follows, such as the MOSFET, capacitors, voltages, etc. Richard
|
|
|
Post by lamazoid on Jun 14, 2007 21:44:47 GMT -8
I'll try to explain that.
The cross-sectional area is 13mm2, path length - 23mm, and there's no gap. Primary turns i'm currently using: 4 for primary itself and 2 for feedback. 0.5mm wire. I don't know actual number of secondary turns. Propably 250-300. Core is POT type. It's harder to explain about ferrite material. We use different marking system here, example MS measured in Tesla, and for that ferrite its 0.38 TL. Initial magnetic permeability is around 2000. That's all i can tell you. Hope this will help.
|
|
|
Post by lamazoid on Jun 14, 2007 21:59:08 GMT -8
Photo of transformer:
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Jun 15, 2007 4:13:22 GMT -8
Iamazoid, thanks for the additional info! That's great. I am going out for the weekend, but I will work on it next week. The really difficult part with Pspice is matching the transformer. They list 100's of such transformers, but you have to go through all of them untill a close match is found to the parameters you listed. Once a reasonable match is found, you can tweak the parameters for a even better fit. I enjoy the challenge. Richard
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Jun 28, 2007 6:35:41 GMT -8
Iamazoid, I wanted to let you know I've been working on simulating your circuit. It's been a challenge with all the family activities going on. I have been able to get your circuit to oscillate just as you posted it, but I'm having a least 2 problems. I can't get the voltages you show on the drain or primary winding. I'm getting voltages in the micro-volt or millivolt range! Also, I can't seem to be able to control the frequency or duty cycle of the mosfet. The model doesn't respond. I have an accurate model of the IRF7456, but I'm wondering if there is enough internal capacitance to make it oscillate as it should. I have been researching DC-DC converters and every technical note I have found says there should be an IC switch regulator used to control the frequency and duty cycle of the mosfet switch. I am continuing to work on this problem. If anyone out there has suggestions, lets hear them. Regards, Richard
|
|
|
Post by lamazoid on Jun 28, 2007 8:06:06 GMT -8
Yes this circuit is STRICTLY different. It does not work as traditional blocking oscillator. I think, switching occurs when voltage drops due to current rising. So i think the problem is simulating REAL power source. I don't know how this can be done.
Yes you can apply square form external oscillator and see how it works, but its still far from reality. I've tried some ICs to control mosfet and their operation is VERY bad.
|
|
|
Post by Gary Lecomte on Jun 28, 2007 15:02:22 GMT -8
Why not use a UJT, such as a 2N4870.
It will oscillate at Low Voltages.
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Jul 17, 2007 17:53:51 GMT -8
Iamazoid, I tried to model your self oscillating IRF7476 as above, but was not successful. I know I have a good level 1 model of the device, but there must be a problem with my transformer core model. It's really hard to match a given non linear core type transformer with the models available. Instead, for fun I simulated a simple DC-DC convertor using Pspice Pulse source instead of the MOSFet. Because of transformer core problems I used a linear, ideal transformer. By experimenting I obtained a design which takes 3VDc and converts is to 30VDC. You can observe this here: The simulation results are shown here: I am going to work on the MOSFET oscillator problem by itself and the non linear ferrite core model at a later time. Where is Soeren when we need him? I have reason to believe the circuit would work as shown, but it is not practical because of the large transformer required. Regards, Richard
|
|
|
Post by lamazoid on Jul 19, 2007 11:26:53 GMT -8
Sorry, i can't understand the purpose of this model. You did just a transformer demonstration circuit form physics course, and what? It's better to try to simulate some defferent oscillator design to get it work. let it be even regular oscillator with BJT in place of mosfet, but the same "blocking" principle of operation. I think there's no REAL difference between physics processes involved.
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Jul 19, 2007 12:56:08 GMT -8
But it's not just a transformer demonstration! Better to get the overall flyback convertor design to perform as theory predicts and then work at the individual stumbling blocks. Also, from this physics exercise I've learned how sensitive a convertor circuit is to changes in component values. Simulate this with PSpice and see how ripple changes with only small changes in filter capacitance. Try changing the turns ratio and inductance of the primary winding. Make the primary winding too low in inductance and not enough energy (emf) is created to make the transformer work properly. I can now work on one element of the design while knowing the overall concept performs as expected. I am going to take Gary's BJT idea and see what happens. And I am going to work until I get an accurate model of your ferrite pot.
Edit by Gary: You mean UJT I hope. Richard
|
|
|
Post by Gary Lecomte on Jul 19, 2007 19:21:59 GMT -8
Actually I have been thinking about this circuit. And considering there is no Gate Resistor, I think its resulting in an output because of NOISE generated by the fet and introduced into the transformer. Or Maybe Parasitic Oscillation of the Fet. ** Not actually an Oscillation of the circuit. (ie: Transformer and Fet) Ideally this circuit should have a Gate resistor (100 ohm) between the winding and the Gate. Also a 10K resistor between the Gate and Source to shut the fet off inbetween cycles. If I correct, I not Not sure these situations can even be simulated.
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Jul 20, 2007 5:15:12 GMT -8
Yes, parasitic oscillation is exactly what the simulator shows. No pattern to it or predictability. That's why I never bothered to post it. Let me work on your idea. By the way I simulated a relaxation oscillator using the 2N4870 and it works well, even with only a 3 volt supply. UJT 's are interesting devices with a sawtooth type output. I will post it when I can. Richard
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Jul 27, 2007 9:29:20 GMT -8
A question for Gary/ anybody! The 2N4870 makes a great relaxation oscillator and sawtooth wave generator. But is it practical to shape this circuit, amplify it, and rectify it as in a DC - DC converter? I've played around with the sawtooth wave form, but an am having difficulty amplifiying it with a step up transformer! I tried a push pull amplifier, but it does not work well because of the form of the sawtooth. Any ideas? Regards, Richard
|
|
|
Post by Gary Lecomte on Jul 27, 2007 18:49:21 GMT -8
A question for Gary/ anybody! The 2N4870 makes a great relaxation oscillator and sawtooth wave generator. But is it practical to shape this circuit, amplify it, and rectify it as in a DC - DC converter? I've played around with the sawtooth wave form, but an am having difficulty amplifiying it with a step up transformer! I tried a push pull amplifier, but it does not work well because of the form of the sawtooth. Any ideas? Regards, Richard POSSIBLY Send the sawtooth output Through a Capacitor, than across Two Diodes, back-to-back to ground. This should give you a Square wave. Unfortunately it will not be a Strong Signal.
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Oct 10, 2008 9:26:14 GMT -8
I thought it would be worthwhile to re-activate this thread. As you know, I like to use simulation software in my electronic endeavors. I have found one that really rocks!!! It is called LTspice and provided by the Linear Technology Company. You can down load it by Googling LTspice. Be sure to download the "Getting Started Guide" and the "Users Guide" Now, the upside of this software (1) it's free and it has no limitations on the number of nodes, complexity,etc. (2) it's easy to use. I got it to work without reading the help guides. (3) you can simulate modern integrated circuits in PWM switching supplies, DC-DC converters and the like. (4) many,many models are provided and they are tested beforehand. Now, the downside: (1) the only integrated circuits available are Linear Technology products. This is not a problem since you can readily find substitutes. (2) it is possible to import other devices, but this is not easily accomplished.
I encourage all you simulators to give it a try. (Including Gary). And no, I don't work for Linear Technologies. I must admit though, it is a pretty good marketing strategy on their part. LTspice is no replacement for the standard Pspice, but it gives us hobbiests a real, practical simulation tool rmeyn
|
|