-
Categories
-
Platforms
-
Content
I need help with my first flyback converter
- Thread starter bonedoc
- Start date
I am pretty new at some of this stuff. From the beginners standpoint, it appears as it is just another type of transistor designed for high voltages.
Is the only way to control this by cutting power to the circuit? I can do that, but want to make sure there is not an easier way than adding another transistor that controls the inout voltage to the primary of the transformer.
Thanks again for any advice! Its a big learning curve.
Is the only way to control this by cutting power to the circuit? I can do that, but want to make sure there is not an easier way than adding another transistor that controls the inout voltage to the primary of the transformer.
Thanks again for any advice! Its a big learning curve.
At the very least, the switch on pulse must be shorter than the power pulse otherwise the SCR is kept conducting. I doubt if it is anything to do with rate of rise of voltage.
The transformer can reverse voltage if it rings, this will turn off the SCR when the current drops to a low enough value. I have used this circuit to repair an electric fence energiser without problems.
The transformer can reverse voltage if it rings, this will turn off the SCR when the current drops to a low enough value. I have used this circuit to repair an electric fence energiser without problems.
Last edited:
It can't ring because there's a diode in series with the transformer.
As long as the frequency is high enough (and the capacitor large enough), the SCR will not stop conducting because the power supply and capacitor will continue to provide enough current to keep it latched.
As long as the frequency is high enough (and the capacitor large enough), the SCR will not stop conducting because the power supply and capacitor will continue to provide enough current to keep it latched.
The common thing to do when firing these setups is to stop pulsing the charging transistor for as long as the discharge takes.
The correct procedure is; charge, stop when full, apply trigger, wait a millisecond, resume charging.
You haven't said what your load consists of. Is it a flashlamp, an inductor, a resistor, or something else?
If it is an inductor you can make sure it rings at a slow enough rate by adding a capacitor in parallel with it.
It seems pretty risky to me to apply 6V directly to the SCR gate w/o any series resistance. It may not kill the SCR if the pulsewidth is short enough though.
I would consider using an optocoupler (SCR type) on the secondary side to trigger the SCR - to keep the controller safe.
The correct procedure is; charge, stop when full, apply trigger, wait a millisecond, resume charging.
You haven't said what your load consists of. Is it a flashlamp, an inductor, a resistor, or something else?
If it is an inductor you can make sure it rings at a slow enough rate by adding a capacitor in parallel with it.
It seems pretty risky to me to apply 6V directly to the SCR gate w/o any series resistance. It may not kill the SCR if the pulsewidth is short enough though.
I would consider using an optocoupler (SCR type) on the secondary side to trigger the SCR - to keep the controller safe.
Ok, I am following your recommendations that you previously gave about overkill on the flyback transistor. I went to the local place and got the NTE2395 to test:
http://www.nteinc.com/specs/2300to2399/pdf/nte2395.pdf
It is rated for 60V, 50A, and has a Rds of .028 Ohms. This was a close to the 100V, 34A, .019 Ohm recommendation that I could get. I figured the lower Gate Rds and gate charge values would make for a faster capacitor charge. But, the charge times are almost identical.
Can you explain this? Did I overlook something on my new transistor selection?
Also, I am also testing at the same 50kHz frequency. I assume that I should test a higher frequency so that I may send energy faster, and that the lower gate resistance and capacitance should allow for this? Correct?
I am understanding it a little more every time I do this, so thanks for the tutorial. There arent any good classes in my area ;-(
http://www.nteinc.com/specs/2300to2399/pdf/nte2395.pdf
It is rated for 60V, 50A, and has a Rds of .028 Ohms. This was a close to the 100V, 34A, .019 Ohm recommendation that I could get. I figured the lower Gate Rds and gate charge values would make for a faster capacitor charge. But, the charge times are almost identical.
Can you explain this? Did I overlook something on my new transistor selection?
Also, I am also testing at the same 50kHz frequency. I assume that I should test a higher frequency so that I may send energy faster, and that the lower gate resistance and capacitance should allow for this? Correct?
I am understanding it a little more every time I do this, so thanks for the tutorial. There arent any good classes in my area ;-(
It could be due to a number of reasons but we're a little in the dark here. For one thing we don't know what impedance & voltage your gate drive has.
This transistor has 2.7 times the gate capacitance of your 400V transistor, and if the drive impedance is high (&/or the voltage is low) this could negate any Rds benefit.
Please remember that the switching frequency is not the most important thing. It's the on time & the off time that matters.
A slow turn-on & -off due to high gate drive impedance will seriously affect the on & off times the transformer sees.
This transistor has 2.7 times the gate capacitance of your 400V transistor, and if the drive impedance is high (&/or the voltage is low) this could negate any Rds benefit.
Please remember that the switching frequency is not the most important thing. It's the on time & the off time that matters.
A slow turn-on & -off due to high gate drive impedance will seriously affect the on & off times the transformer sees.
Well, you will have pretty slow on and off times. You probably want to be able to supply at least 100mA of gate current, more would be better. The 16F886 will be struggling to provide 25mA, and it certainly won't supply that at 6V (is that device rated for 6V operation anyway? -- NO it's 2V to 5.5V)
I see what you are saying. Sorry, I didnt have my actual pic data in front of me. So, do I then need to control a smaller transistor from my pic pin that will have access to my 6V rail so that the larger flyback transistor has more current (and a little more voltage) available?...rather than being limited by the output of the chip? Kind of like a darlington pair?
I have some 2sd965 transistors. The gain is 150, and a collector current of 5A:
http://www.datasheetcatalog.org/datasheet/WINGS/2SD965.pdf
I also have the typical radioshack stuff....2n222
I have some 2sd965 transistors. The gain is 150, and a collector current of 5A:
http://www.datasheetcatalog.org/datasheet/WINGS/2SD965.pdf
I also have the typical radioshack stuff....2n222
Yes, you'll need at least an NPN/PNP pair between the PIC and the MOSFET to increase the gate drive current.
But then there is the issue of the gate voltage needed to reach the specified RDS-on. I'm not sure that 5V is enough. 10V is usually specified for ordinary MOSFETS.
Instead of making a complicated voltage doubling (bootstrap) driver circuit you could try to obtain a logic-level MOSFET (they manage with only 5V gate voltage).
Here is an explanation of the bootstrap concept, and here is a discussion of how to improve gate driver switching times.
But then there is the issue of the gate voltage needed to reach the specified RDS-on. I'm not sure that 5V is enough. 10V is usually specified for ordinary MOSFETS.
Instead of making a complicated voltage doubling (bootstrap) driver circuit you could try to obtain a logic-level MOSFET (they manage with only 5V gate voltage).
Here is an explanation of the bootstrap concept, and here is a discussion of how to improve gate driver switching times.
Ok, I am going to go get a 12V battery pack to try. I also called the local store and all they have in the logic level mosfet is the NTE2987.
http://www.nteinc.com/Web_pgs/LL_MOSFET.html
The rds is 0.12 Ohm and the max current is 20A, and it is rated to 100V. I will give it a try until I can order the proper stuff.
http://www.nteinc.com/Web_pgs/LL_MOSFET.html
The rds is 0.12 Ohm and the max current is 20A, and it is rated to 100V. I will give it a try until I can order the proper stuff.
That one seems well suited to the task. At the design target of 3A peak current it'll drop only about 0.3V - which is perfectly acceptable in a 6V system (~5% drop).
Ok, I think I am good to go! Thanks for everyones help here!
By accident, I was programming my ccp module to do the 50k wave. Just so you know, my signal generator does not go over 200k. Well, I accidentally dropped a 1 out of a byte in the ccp module set up....and guess what. It was charging my capacitor like a mad man. I found that it was 250kHz.... and the 40% duty cycle is a little more efficient than 30% at this condition, and have not noticed any heat yet .I can charge to 300v and 120uf in 4 seconds off a 12V rail. I am using the logic level mosfet I picked up a couple days ago.
One last question. I have a fixed "load" in this experiment. Nothing will change. Just charging a capacitor. The company has 2 other models, that are almost exact...but the Ipk and inter winding capacitance are higher. Is there any benefit to those then my "load" does not pull that? I have some samples to try....but curious.
All three models:
http://www.coilcraft.com/pdf_viewer/showpdf.cfm?f=pdf_store:da2032.pdf
By accident, I was programming my ccp module to do the 50k wave. Just so you know, my signal generator does not go over 200k. Well, I accidentally dropped a 1 out of a byte in the ccp module set up....and guess what. It was charging my capacitor like a mad man. I found that it was 250kHz.... and the 40% duty cycle is a little more efficient than 30% at this condition, and have not noticed any heat yet .I can charge to 300v and 120uf in 4 seconds off a 12V rail. I am using the logic level mosfet I picked up a couple days ago.
One last question. I have a fixed "load" in this experiment. Nothing will change. Just charging a capacitor. The company has 2 other models, that are almost exact...but the Ipk and inter winding capacitance are higher. Is there any benefit to those then my "load" does not pull that? I have some samples to try....but curious.
All three models:
http://www.coilcraft.com/pdf_viewer/showpdf.cfm?f=pdf_store:da2032.pdf
Last edited:
Wow....I have been testing more duty cycles. I said in my last post that I was at a 40% duty and charging in 4 seconds. I bumped up the duty cycle to 45% and it charged in 2.5 seconds. At 50% duty, I it is 2 seconds. This is great....but I do notice some heat after several cycles.
Should I be testing for continuous use and be concerned about overheating?
Should I be testing for continuous use and be concerned about overheating?
You should be concerned that you are still talking about frequency and duty cycle when it's ON and OFF times that are important.
As you change frequency the same duty cycle will correspond to different on and off times and thus different behaviour.
Do you understand what happens to the current through the transformer during the ON period?
As the current rises, the resistive power losses (I^2R) increase more rapidly, but the amount of energy stored also increases. Some heating is expected and acceptable. Have you ensured that the peak current is not exceeded?
"Some heat" is not useful. Can you measure the temperature rise?
If your on time is too long, you will saturate the inductor, the current will rise very rapidly, and at the very least, the inductor and the mosfet will get very hot.
Is the mosfet getting at all warm? Are you driving it directly from the uC?
As you change frequency the same duty cycle will correspond to different on and off times and thus different behaviour.
Do you understand what happens to the current through the transformer during the ON period?
As the current rises, the resistive power losses (I^2R) increase more rapidly, but the amount of energy stored also increases. Some heating is expected and acceptable. Have you ensured that the peak current is not exceeded?
"Some heat" is not useful. Can you measure the temperature rise?
If your on time is too long, you will saturate the inductor, the current will rise very rapidly, and at the very least, the inductor and the mosfet will get very hot.
Is the mosfet getting at all warm? Are you driving it directly from the uC?
Ok, I see what you are saying.
If I were at a 80% duty cycle at 250k, the "on time" would be .0000032 sec. In a slower 50k pulse, that same .0000032 sec would equate to a 16% of the total cycle.
That pulse transfers the same amount of energy, but the higher frequency does it more often. A person just needs to be certain that the "on time" is the optimal amount of time that the energy can be transferred, and then the rest of the period ("off time") allows it to recover.
Right now, it is warm to the touch without a heat sink. I will go read through the previous equations again and attempt to get values with my 12V rail and a logic level mosfet.
When they refer to peak current....is that out of the secondary side?
I am driving the logic level mosfet from the pic, directly. Will add a optocoupler for safety.
If I were at a 80% duty cycle at 250k, the "on time" would be .0000032 sec. In a slower 50k pulse, that same .0000032 sec would equate to a 16% of the total cycle.
That pulse transfers the same amount of energy, but the higher frequency does it more often. A person just needs to be certain that the "on time" is the optimal amount of time that the energy can be transferred, and then the rest of the period ("off time") allows it to recover.
Right now, it is warm to the touch without a heat sink. I will go read through the previous equations again and attempt to get values with my 12V rail and a logic level mosfet.
When they refer to peak current....is that out of the secondary side?
I am driving the logic level mosfet from the pic, directly. Will add a optocoupler for safety.
No, the peak current refers to the primary.
One way to measure it is to fit a small current sense resistor (say 0.1 ohm) in series with the inductor and measure the voltage across it using an oscilloscope.
Ideally, you want the peak current to be about 3A (so 0.3V across the resistor), however due to a number of factors you may want to set it a little lower.
Next you want to do the same thing with the secondary current. However this time you're looking for the time it takes for the current to fall close to zero.
In the secondary you can use a larger resistor because the current will be much lower.
This time will vary with the load, and you should choose the longest time (probably when the capacitor is almost fully charged).
This will represent your optimum off time. You may wish to extend it slightly.
Armed now with an optimum on and off time you can calculate the frequency and duty cycle if that's what you need.
Remember though, that if you then reduce the frequency without reducing the duty cycle that you will cause excessive current to flow through the inductor.
One way to measure it is to fit a small current sense resistor (say 0.1 ohm) in series with the inductor and measure the voltage across it using an oscilloscope.
Ideally, you want the peak current to be about 3A (so 0.3V across the resistor), however due to a number of factors you may want to set it a little lower.
Next you want to do the same thing with the secondary current. However this time you're looking for the time it takes for the current to fall close to zero.
In the secondary you can use a larger resistor because the current will be much lower.
This time will vary with the load, and you should choose the longest time (probably when the capacitor is almost fully charged).
This will represent your optimum off time. You may wish to extend it slightly.
Armed now with an optimum on and off time you can calculate the frequency and duty cycle if that's what you need.
Remember though, that if you then reduce the frequency without reducing the duty cycle that you will cause excessive current to flow through the inductor.
Ok, I will try this when I get home for lunch. I did the theoretical stuff on paper. These are my cliff notes on this topic. If I have 2 inductors of EXACT specs, and they only vary by Ipk....
It appears that the one with the lower Ipk can charge smaller current, but fast. The one with a higher Ipk will charge slower, but with a larger current.
Therefore, it is a trade off, and I should not need a larger Ipk unless the circuit demands it. Is there another trade off I am not aware of? Here are the cliffnotes:
***************************************************************************
Max Pulse Width = (Ipk x Primary Inductance) / Voltage Applied
Max period = 2 x Max Pulse Width
Frequency = 1/peroid
-As Ipk increases, so does the pulse width, so the frequency decreases.
-As inductance increases, so does the pulse width, so the frequency decreases
-As voltage applied increases, pulse width decreases, so the frequency increases
-One should decide the current drawn on the primary side with a 0.1 Ohm in series, when the load is the highest to determine what min Ipk can be used.
-If there are several inductor choices with the proper Ipk rating, it may be beneficial to select the lowest primary inductance to allow for a shorter pulse width, and faster conversion.
-Increasing the voltage within the voltage rating of the inductor will decrease the pulse width, so a high power rail rating will benefit.
--------------
Example -> 2 inductors, with 12v applied, and the same primary inductace of .00001 H. One inductor has an Ipk of 3A, and the second has an Ipk of 10A:
Inductor 1
(3A x 0.00001H) / 12V = 2.5us max PULSE width
Theoretical period @ 50% duty = 2.5us x 2 = 5.0us
Frequency = 1/0.000005s = 200kHz
Inductor 2
(10A x 0.00001H) / 12V = 8.3us max PULSE width
Theoretical period @ 50% duty = 8.3us x 2 = 16.7us
Frequency = 1/0.0000167 = 60kHz
---------------
It appears that the one with the lower Ipk can charge smaller current, but fast. The one with a higher Ipk will charge slower, but with a larger current.
Therefore, it is a trade off, and I should not need a larger Ipk unless the circuit demands it. Is there another trade off I am not aware of? Here are the cliffnotes:
***************************************************************************
Max Pulse Width = (Ipk x Primary Inductance) / Voltage Applied
Max period = 2 x Max Pulse Width
Frequency = 1/peroid
-As Ipk increases, so does the pulse width, so the frequency decreases.
-As inductance increases, so does the pulse width, so the frequency decreases
-As voltage applied increases, pulse width decreases, so the frequency increases
-One should decide the current drawn on the primary side with a 0.1 Ohm in series, when the load is the highest to determine what min Ipk can be used.
-If there are several inductor choices with the proper Ipk rating, it may be beneficial to select the lowest primary inductance to allow for a shorter pulse width, and faster conversion.
-Increasing the voltage within the voltage rating of the inductor will decrease the pulse width, so a high power rail rating will benefit.
--------------
Example -> 2 inductors, with 12v applied, and the same primary inductace of .00001 H. One inductor has an Ipk of 3A, and the second has an Ipk of 10A:
Inductor 1
(3A x 0.00001H) / 12V = 2.5us max PULSE width
Theoretical period @ 50% duty = 2.5us x 2 = 5.0us
Frequency = 1/0.000005s = 200kHz
Inductor 2
(10A x 0.00001H) / 12V = 8.3us max PULSE width
Theoretical period @ 50% duty = 8.3us x 2 = 16.7us
Frequency = 1/0.0000167 = 60kHz
---------------
Last edited:
Ok, I have the converter hooked up as normal, and I placed a 1 Ohm resistor at the bottom leg of the inductor (not the upper leg where the 12v comes in, but the other leg that goes to the drain of the transistor). I hooked the meter red up to the other end of this resistor, and the meter black to ground.
The peak current pulled is 4.4A. This is at the start of the cycle, and it drops as the secondary charges.
I have a 5A and 10A Ipk inductor. Would 5A be suitable? I dont know what safety margin I should have.
Also, if I am pulling 4.4A....should I use this in my max pulse equation, ot do I still use the 5A or 10A Ipk of the inductor?
The peak current pulled is 4.4A. This is at the start of the cycle, and it drops as the secondary charges.
I have a 5A and 10A Ipk inductor. Would 5A be suitable? I dont know what safety margin I should have.
Also, if I am pulling 4.4A....should I use this in my max pulse equation, ot do I still use the 5A or 10A Ipk of the inductor?
Last edited:
