Finally i build a full-bridge smps, but i have some problems with them. Then i turn it ON, some seconds he works, but later the power mosfets are destroid. I think there is a problem with reaching saturation on ETD49 core who has 3F3 material. I use 100khz frequency. I need to know how to calculate a optimal core frequency.
I've a powdered iron C core about 190 mm^2 cross section from a large old TV flyback, and I wound on a couple of secondaries of about 2500 turns each (wax paper for insulation). I'm wondering of the best way to drive the primary. I don't want to do it in flyback mode due to voltages that develop on the primary side (already fried one switch). I'd like a full-bridge, so I'm wondering if the circuit from this post is appropriate, but replace the caps with switches. Should I use MOSFETs instead of IGBTs given the low power (200 W), and how do I calculate C12 and C13? What wattage is R13? Also, in a lot of schematics that have been posted in this thread, there are feedback circuits. Can these be omitted? I don't need any voltage regulation, as the supply is intended to feed doublers anyway.
monitor smps
Dp104c is easily avaialable at less than 1$. It is actually equivalent to fairchild's KA 5Q series switching ics. These device also features a thermal shut down. The advantage of using them is both PWM and switching MOSFET are packed inside them. So no additional components except a optocoupler is needed. I had designed one with them and now powering a 200w amplifier. It works quite fine. But I need to measure how much power it can actually deliver safely. Is there any way to measure them? Application notes suggest that they can deliver upto 300W. Pls do reply...
Dp104c is easily avaialable at less than 1$. It is actually equivalent to fairchild's KA 5Q series switching ics. These device also features a thermal shut down. The advantage of using them is both PWM and switching MOSFET are packed inside them. So no additional components except a optocoupler is needed. I had designed one with them and now powering a 200w amplifier. It works quite fine. But I need to measure how much power it can actually deliver safely. Is there any way to measure them? Application notes suggest that they can deliver upto 300W. Pls do reply...
It's not a TC at all; TCs don't even have a core, they are air coils -- nothing to do with my project, which is for a + and -15 kVDC source. I know the transformer is for the gates. That wasn't my question. Let me quote myself: "Should I use MOSFETs instead of IGBTs given the low power (200 W), and how do I calculate C12 and C13? What wattage is R13? Also, in a lot of schematics that have been posted in this thread, there are feedback circuits. Can these be omitted?" and "Are complementary MOSFETs ever used in a bridge, or always N-channel?"
IGBT"S vs MOSFET'S
The debate on mosfets vs igbt's is about power and frequency but I would also include cost for your situation. Mosfet's are very good choice for all frequencies especially over 100kHz while igbt's are starting to replace them under 100kHz.
I use igbt's because the capacitance is much lower than a mosfet and therefore good gate drive is easier to obtain.
Have you considered a two switch forward converter for your Hi-Voltage supply?
I think the output filter made up of capacitor's should work fine, just make sure your transformer turns ratio supports your load.
chas1
The debate on mosfets vs igbt's is about power and frequency but I would also include cost for your situation. Mosfet's are very good choice for all frequencies especially over 100kHz while igbt's are starting to replace them under 100kHz.
I use igbt's because the capacitance is much lower than a mosfet and therefore good gate drive is easier to obtain.
Have you considered a two switch forward converter for your Hi-Voltage supply?
I think the output filter made up of capacitor's should work fine, just make sure your transformer turns ratio supports your load.
chas1
Nixie said:
N-CH MOSFETs are used in almost all SMPS Apps because of their lower "On" resistance. For example, an IRFP140 has about 52mW Rds(on), while it's P-CH compliment, the IRFP9140 has about 117mW Rds(on).
This would result in unequal driving of the transformer, and without appropriate protection, the transformer core would walk toward saturation very quickly, resulting in a catastrophic failure of the MOSFETs.
On the rare occasion that complimentary MOSFETs are used, their Rds(on)s are so closely matched; the transformer primary is capacitively coupled; and current-mode control are all employed to mitigate this potential condition from ever occuring.
Steve
Update on project
I got it to work as expected.Had to rewind output inductor and added a total of four 3300uF caps.Since I have only 4 ohme and 2+2 ohme loads all in parallel to make 1.14 ohme load.Output voltage is constant at 13v, so I got 11.4 amps with that load. Don't have better load to test supply with but i'm happy with it.
On other hand I've bought four mur1560's for the main project, so I will mount them tommorow and test it how it does.
I got it to work as expected.Had to rewind output inductor and added a total of four 3300uF caps.Since I have only 4 ohme and 2+2 ohme loads all in parallel to make 1.14 ohme load.Output voltage is constant at 13v, so I got 11.4 amps with that load. Don't have better load to test supply with but i'm happy with it.
On other hand I've bought four mur1560's for the main project, so I will mount them tommorow and test it how it does.
N-ch P-ch
Chas,
I don't know.
Seriously, this was near the end of the application Note that IR put out on their then-new IR211X series Floating-Gate MOSFET driver chips. It said something like how to calculate the power dissipation of the chip when driving the upper N-Channel MOSFET.
It said that if the upper MOSFET was a P-Channel, then the drive for that FET would be ground referenced (as opposed to floating for the N-CH), and then the driver chip's power dissipation would be zero.
As a matter of practicality, the only P-Channel I would use is Harris' RFG60P05 (50V 60A P-Ch), which has an Rds(on) of only 26mW. Come to think of it, this FET would be ideally suited for a classic Buck regulator!
As a high-side switch and as a P-channel, no special floating driver circuit would be needed, as its drive signal is ground-referenced. 
Steve
Chas,
I don't know.
Seriously, this was near the end of the application Note that IR put out on their then-new IR211X series Floating-Gate MOSFET driver chips. It said something like how to calculate the power dissipation of the chip when driving the upper N-Channel MOSFET.It said that if the upper MOSFET was a P-Channel, then the drive for that FET would be ground referenced (as opposed to floating for the N-CH), and then the driver chip's power dissipation would be zero.
As a matter of practicality, the only P-Channel I would use is Harris' RFG60P05 (50V 60A P-Ch), which has an Rds(on) of only 26mW. Come to think of it, this FET would be ideally suited for a classic Buck regulator!
As a high-side switch and as a P-channel, no special floating driver circuit would be needed, as its drive signal is ground-referenced. Steve
Mosfet's
Steve
I will file this away, thanks a lot for the info which is a nugget indeed. I have finished testing phase of voltage mode bridge and halfbridge and working out a Weinberg converter design (500 watts) just for evaluation and curiosity. I have set aside the current mode for now as I think voltage mode is the way to go for audio amp supply, what are your thoughts on this.
By the way thanks for all the help so far the amps seem very happy with the new supply.
After much thought I put the controller on the secondary side, the added cost of a transformer and small bridge is a small price to pay for piece of mind in this application and adds very little to pcb size.
The purpose for the Weinberg is to get rid of the output inductor, Pressman devotes a liitle time to this design and it can operate in both CM and DCM.
chas
Steve
I will file this away, thanks a lot for the info which is a nugget indeed. I have finished testing phase of voltage mode bridge and halfbridge and working out a Weinberg converter design (500 watts) just for evaluation and curiosity. I have set aside the current mode for now as I think voltage mode is the way to go for audio amp supply, what are your thoughts on this.
By the way thanks for all the help so far the amps seem very happy with the new supply.
After much thought I put the controller on the secondary side, the added cost of a transformer and small bridge is a small price to pay for piece of mind in this application and adds very little to pcb size.
The purpose for the Weinberg is to get rid of the output inductor, Pressman devotes a liitle time to this design and it can operate in both CM and DCM.
chas
VM -v- CM
Chas-
I would concur with Luka that VM is the best way to go for audio applications. Couple o' reasons: 1) the instability problems that can occur if the duty cycle for CM goes over 50% (easily solvable, but with more complex circuitry); 2) uh, I forgot ther second reason.
Also, since you will be employing a coupling capacitor in series with the primary winding, even if you have unequal on-times from the MOSFETs, flux imbalance will never occur, because of AC coupling on the primary. Yoi will just have a little more voltage across the coupling cap, and less across the primary. The PWM will compensate for this by adjusting the duty cycle accordingly.
Steve
Chas-
I would concur with Luka that VM is the best way to go for audio applications. Couple o' reasons: 1) the instability problems that can occur if the duty cycle for CM goes over 50% (easily solvable, but with more complex circuitry); 2) uh, I forgot ther second reason.
Also, since you will be employing a coupling capacitor in series with the primary winding, even if you have unequal on-times from the MOSFETs, flux imbalance will never occur, because of AC coupling on the primary. Yoi will just have a little more voltage across the coupling cap, and less across the primary. The PWM will compensate for this by adjusting the duty cycle accordingly.
Steve
VM -v- CM
Chas-
I would concur with Luka that VM is the best way to go for audio applications. Couple o' reasons: 1) the instability problems that can occur if the duty cycle for CM goes over 50% (easily solvable, but with more complex circuitry); 2) uh, I forgot the second reason.
Also, since you will be employing a coupling capacitor in series with the primary winding, even if you have unequal on-times from the MOSFETs, flux imbalance will never occur, because of AC coupling on the primary. You will just have a little more voltage across the coupling cap, and less across the primary. The PWM will compensate for this by adjusting the duty cycle accordingly.
Steve
Chas-
I would concur with Luka that VM is the best way to go for audio applications. Couple o' reasons: 1) the instability problems that can occur if the duty cycle for CM goes over 50% (easily solvable, but with more complex circuitry); 2) uh, I forgot the second reason.
Also, since you will be employing a coupling capacitor in series with the primary winding, even if you have unequal on-times from the MOSFETs, flux imbalance will never occur, because of AC coupling on the primary. You will just have a little more voltage across the coupling cap, and less across the primary. The PWM will compensate for this by adjusting the duty cycle accordingly.
Steve
CM or VM
Steve
Thanks for the feedback; I will stay with the VM. I hate that this project is in the last phase just a little cleanup and then I start the real test but I don't foresee any problem. I was hung up for a while with compensation. For a while I thought my computer had a virus unknown to man, while my calculations based on the examples from 2 books seemed ok and even some old notes I had from the 80's reinforced them but all my spice plots were off the chart and since I was in a rush I overlooked a couple of things like really read the printed page , research your subject and remember GIGO. I relaxed in the recliner and read the books , to my surprise both had errors at crucial points I was really miffed. I was lucky to have some old tips from TI that bailed me out. I don't understand why a publisher does not check a manuscript before printing. I tried to publish back in the 70's and the work I submitted was kicked back numerous times for error's and it was never finished.
chas1
Steve
Thanks for the feedback; I will stay with the VM. I hate that this project is in the last phase just a little cleanup and then I start the real test but I don't foresee any problem. I was hung up for a while with compensation. For a while I thought my computer had a virus unknown to man, while my calculations based on the examples from 2 books seemed ok and even some old notes I had from the 80's reinforced them but all my spice plots were off the chart and since I was in a rush I overlooked a couple of things like really read the printed page , research your subject and remember GIGO. I relaxed in the recliner and read the books , to my surprise both had errors at crucial points I was really miffed. I was lucky to have some old tips from TI that bailed me out. I don't understand why a publisher does not check a manuscript before printing. I tried to publish back in the 70's and the work I submitted was kicked back numerous times for error's and it was never finished.
chas1
Re: trafo assembly
You can use Polyurthane (found in hardware stores) to coat the windings. It will secure them to prevent them vibrating. Apply with light coats. If you apply too much it will bubble up. Poly urthane glue can also be used since its less runny. Its a good idea to use gloves since its not water soluable. Acetone and xylene work well for clean up.
luka said:
You can use Polyurthane (found in hardware stores) to coat the windings. It will secure them to prevent them vibrating. Apply with light coats. If you apply too much it will bubble up. Poly urthane glue can also be used since its less runny. Its a good idea to use gloves since its not water soluable. Acetone and xylene work well for clean up.
The windings of a high frequency SMPS transformer should not buzz in the fisrt place. If they do then there is probably some problem with the regulation loop. Note that audible noise in a SMPS transformer is a sign of large flux swings at audio frequencies (that should not happen) or saturation.
noise 2nd time
Forget the hardware store and listen to Eva’s comments if you have audible noise in your transformer it’s more than likely saturation or improper feedback compensation or no compensation. A few simple check's with an o'scope should find the problem. It's in the pole's and zero's and where there located in time.
chas1
Forget the hardware store and listen to Eva’s comments if you have audible noise in your transformer it’s more than likely saturation or improper feedback compensation or no compensation. A few simple check's with an o'scope should find the problem. It's in the pole's and zero's and where there located in time.
chas1