|
Post by lamazoid on Apr 13, 2009 11:17:05 GMT -8
But they are same! I can see no difference
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Apr 13, 2009 12:08:12 GMT -8
They may well be, but... LTspice will not let you run a linear along with an non linear inductor. You can't have a K spice directive along with non-linear attributes like Hc, Lm etc. (at least not for the same set of inductors) When I ran Gary's schematic with an ideal transformer (not what we have) it oscillates beautifully. Puts out a pulse of about 130V at very high frequency (many gigahertz). When I tried to run Gary's format as non linear inductors with the attributes you list, it won't oscillate--gives a steady state output voltage. When I get a chance I will look at your net listing and see how you got it to run. I have not tried running your original circuit with 6 inductors (3 linear and 3 non-linear). richard Richard
|
|
|
Post by Gary Lecomte on Apr 13, 2009 13:46:52 GMT -8
They may well be, but... LTspice will not let you run a linear along with an non linear inductor. You can't have a K spice directive along with non-linear attributes like Hc, Lm etc. (at least not for the same set of inductors) When I ran Gary's schematic with an ideal transformer (not what we have) it oscillates beautifully. Puts out a pulse of about 130V at very high frequency (many gigahertz). When I tried to run Gary's format as non linear inductors with the attributes you list, it won't oscillate--gives a steady state output voltage. When I get a chance I will look at your net listing and see how you got it to run. I have not tried running your original circuit with 6 inductors (3 linear and 3 non-linear). richard Richard Thats why I RARELY Simulate. Very Inconsistant results. GigaHertz, Thats Funny. And Most often when I build and test a particular circuit, My Real Results are Different from the Simulation. In Your Schematic, It appears your Paralleling some coils, just to make it simulate? I'm quite sure that will also change Actual Results! Sorry, But Presently I don't have much time to Play. Too many NECESSARY things to do, While I Can.
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Apr 13, 2009 17:57:19 GMT -8
Iamazoid, . Now that I had a chance to look at this I have some questions. I do not see why you put in 3 non linear conductors to achieve coupling. The K1 spice directive is sufficient if you are simulating an ideal transformer. I think the 3 non linear conductors you have added as parallel windings are the problem. I think Gary's schematic accurately describes what a blocking oscillator circuit should look like. I don't think your LTspice simulation graphic is showing what you want the circuit to do. There is ringing going on and the pulse is not very clean. When I simulated Gary's circuit using a hypothetical linear transformer I got very clean voltage pulses varying from about 0 to 130V. The frequency varied from 10 khertz to 10 megahertz depending upon the value of Rser for the coils. When I tried to simulate Gary's diagram using 3 non linear inductors with the magnetic attributes you showed , I could not get oscillation. Other Ltspice demo circuits work with non linear transformers. Ive come to the conclusion that some of the magnetic attributes are incorrect. I read on the web that it is very difficult to reverse engineer non linear coils and determine their magnetic properties. Richard
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Apr 13, 2009 18:46:55 GMT -8
Here is LtSpice using a hypothetic ideal linear transformer as per Gary's diagram Richard
|
|
|
Post by lamazoid on Apr 13, 2009 22:30:19 GMT -8
But it works for me! And i see no other way to simulate REAL , not IDEAL transformer. My textbook says to do so. There are even 3 more coils needed according this book, connected in series with other paraller pairs, to simulate magnetic flux dissipiation. I think for small ferrite transformer that is not critical thing, especially for POT core. But nonlinear parameters such as saturation flux density are required to describe real core, and they affect real circuit operation strictly ! So i see no point to simulate "ideal" circuits at all. This is practical for students, learning physical principles of operation, but not for people making real devices.
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Apr 14, 2009 4:24:54 GMT -8
Iamazoid, Please give me details on the book. I've found very little on web concerning simulation of non linear transformers, other than stating it is difficult to do. Richard
|
|
|
Post by lamazoid on Apr 14, 2009 5:49:17 GMT -8
That book is in russian, i don't know other sources of information. In short, it describes transformer simulation as a set of inductances. Here is a picture from this book, showing model of a typical 50Hz AC power source I'm not sure, should it be only one non-linear inductance per transformer, or a parrallel pairs with each winding, as i did. The book doesn't tell that. At least in my case, simulation results are the same.
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Apr 14, 2009 6:08:19 GMT -8
Thanks Iamazoid, This gives me something to work with. You know there must be ways of accurately simulating non linear inductors. Modern switching circuits ( a specialty of Linear Devices) are designed with simulators. Simulation is the first technique used in designing new circuits, followed by bread boarding and testing , then production. I am going to do some research on books covering this topic. Plan on checking out amazon.com books. Richard ps. When you get time, please post a oscilloscope picture of the output of your design, so I have something to compare with
|
|
|
Post by Gary Lecomte on Apr 14, 2009 7:56:07 GMT -8
Here is what my Electronic Workbench Ver5.1 Manual Says about these Transformers. Maybe some Help?
|
|
rmeyn
Full Member
My primary interests are electronics, sailing, bowling, and camping
Posts: 163
|
Post by rmeyn on Apr 14, 2009 11:17:09 GMT -8
Thanks Gary, This will help both of us. Will keep everybody posted on our efforts, successes, failures,etc. Richrd
|
|
|
Post by lamazoid on Apr 15, 2009 4:33:19 GMT -8
Well , here are some practical results. Transformer wound on 18mm pot core. Primary 18 turns 0.5mm , secondary 200 turns 0.16mm , feedback 18 turns 0.2mm. Current draw is 0.25 A, transistor gets a little warm. It'not 2n3055 but similar russian transistor , and it has PNP structure. This is my test setup: And here is waweform at collector, showing pulses of about 18V amplitude and 20 microsecond width: Now, waweform at lamp (both scope terminals connected to lamp directly) Main question is the same: what is heating the transistor and how to avoid this. Also, how to optimize overall performance.
|
|
|
Post by Gary Lecomte on Apr 15, 2009 6:10:06 GMT -8
Well , here are some practical results. Transformer wound on 18mm pot core. Primary 18 turns 0.5mm , secondary 200 turns 0.16mm , feedback 18 turns 0.2mm. Current draw is 0.25 A, transistor gets a little warm. It'not 2n3055 but similar russian transistor , and it has PNP structure. This is my test setup: And here is waweform at collector, showing pulses of about 18V amplitude and 20 microsecond width: Now, waweform at lamp (both scope terminals connected to lamp directly) Main question is the same: what is heating the transistor and how to avoid this. Also, how to optimize overall performance. Your Results Here look Reasonable, Good Job! If the transistor just runs a LITTLE Warm, Whats the Problem? Your Circuit is Creating a Power Pulse, So the transistor MUST Dissipate the Current coming from the transformer, so It will NEVER Run COLD. Current Causes Heat! 1) And do you have a heat sink on it? Transistors and Mosfets are rated at there Correct Power, ONLY when Properly Heat Sinked. 2) Is this the circuit I posted? Guess not if your using a PNP Transistor. Where is This Circuit Diagram? 3) In the circuit I Posted, There should also be an Additional Resistor, between the Transformer Feedback wire and the Base of the Transistor. (to Limit Base Current.)
|
|
|
Post by lamazoid on Apr 15, 2009 9:26:53 GMT -8
Gary, thanks for reply. Here is exact circuit which i'm currently using: I started my experiments after disassembling one commercial lamp rated at 8w. It was built around 2SD882 transistor on a very small heatsink (transistor itself is in to-126 package) and while it was consuming 0.7 A it wasn't even warm. In my test setup current is only 0.25 A while CFL is 9w rated! It's not running at full power now, so it's not heating! But i want at least 8w drive with no excessive heating. I've built one complete lamp (maybe i'll post a photo later) with exactly the same circuit but transfomer had 20 turns primary and 10 turns feedback, and it worked well with small heatsink (but it heats to ~40C anyway) driving tube almost at full power (draws 0.72 A) I'm trying to understand, how the number of turns (primary and feedback) affects circuit operation and where is the optimum. Since i was unable to find any theory on this, my experiments are based just on guess
|
|
|
Post by lamazoid on Apr 16, 2009 2:17:35 GMT -8
Some news:
I rewound feedback with 30 turns, and added small air gap to decrease windings inductance. Now it draws 0.5 amp, light output increased strongly and transistor is still only a little warm! It seems i'm on the right way.
|
|