Actually i personally do not need anything. I am definetly not the best power supply designer, but there are many years of some quite complex SMPS development tasks behind my shoulders. Are you one like me? Looking at the very proposal to use Zeners to dissipate tens of watts suggest you are not.
In Israel, the power supply design community is tiny, especially compared to DSP/Microcontroller/Comms hardware. If i am stumbled upon one that is interested in this area, i better help him.
I know that this sound rude for you, but here it is:
I am trying to help this guy. You may put you ego aside and learn yourself, or move aside.
In Israel, the power supply design community is tiny, especially compared to DSP/Microcontroller/Comms hardware. If i am stumbled upon one that is interested in this area, i better help him.
I know that this sound rude for you, but here it is:
I am trying to help this guy. You may put you ego aside and learn yourself, or move aside.
It is actually my typo it should be 1uH not 3uH.
P=0.5*39^2*1e-6*100e3=76W
Those will divide between the FETs as each fet is switching at half frequency.
I actually assume that the leakage will be 5% and the loss will be very heavy. Back in 2005 i designed a push-pull 35W 500KHz small converter. The input was 18-36V normal 80V in short pulses. I used 150V SO8 double fet, i think, with EI18 planar trafo embedded into the small PCB. And the planar trafo almost always have much reduced Lk ~0.25-0.4%.
Anywavy, i recall that the leakage kick at turn-off was considerable, and required some efforts to get rid off.
Also i tried to do a current-fed push-pull a year ago. I personally winded all of the magnetics. The Trafo used some toroide, i cant remember which with litz windings on both P&S.
Leakage was 5% its effect simply untolerable!
The expirience gathered lead me to adopt planat magnetics to almost everything i do, and use resonant reset current-mode forward up to 50W, then go to HB or even FB.
P=0.5*39^2*1e-6*100e3=76W
Those will divide between the FETs as each fet is switching at half frequency.
I actually assume that the leakage will be 5% and the loss will be very heavy. Back in 2005 i designed a push-pull 35W 500KHz small converter. The input was 18-36V normal 80V in short pulses. I used 150V SO8 double fet, i think, with EI18 planar trafo embedded into the small PCB. And the planar trafo almost always have much reduced Lk ~0.25-0.4%.
Anywavy, i recall that the leakage kick at turn-off was considerable, and required some efforts to get rid off.
Also i tried to do a current-fed push-pull a year ago. I personally winded all of the magnetics. The Trafo used some toroide, i cant remember which with litz windings on both P&S.
Leakage was 5% its effect simply untolerable!
The expirience gathered lead me to adopt planat magnetics to almost everything i do, and use resonant reset current-mode forward up to 50W, then go to HB or even FB.
Tolik
I think i have made another mistake by assuming 6A 60V output, while you want only 240W. Thats 4A 60V. OK
Scale down the currents and recalculate the Lout.
dIL that i am talking about is the current ramp in the output inductor, you may call it "Lo magnetizing current" if you like, its OK.
How is your Hebrew?
You can easily obtain little ferrite toroids from I.R.Metals, wind 2 gate drive transformers and go for a full bridge. The anti-parallel diodes in the bridge (you can even use schottkeys on 12V bridge) will safely recirculate the Lk power. Minimal snubbing will be needed, and if the transformer "rings" during dead time it is not so dangerous as in push pull. You can work with 20-30V mosfets, at a very high efficiency.
I think i have made another mistake by assuming 6A 60V output, while you want only 240W. Thats 4A 60V. OK
Scale down the currents and recalculate the Lout.
dIL that i am talking about is the current ramp in the output inductor, you may call it "Lo magnetizing current" if you like, its OK.
How is your Hebrew?
You can easily obtain little ferrite toroids from I.R.Metals, wind 2 gate drive transformers and go for a full bridge. The anti-parallel diodes in the bridge (you can even use schottkeys on 12V bridge) will safely recirculate the Lk power. Minimal snubbing will be needed, and if the transformer "rings" during dead time it is not so dangerous as in push pull. You can work with 20-30V mosfets, at a very high efficiency.
Alexch,
Why mistake, more is better for me.
But in you calculations I found one that can to contain mistake :
For toroid with Ac = 1.28cm^2 -
Np=4+4 Bmax (from zero to maximal positive or negative value) = 823 Gauss, So dB = 1646 gauss. Lprim = 64...65 uH, & I mag = 2.61A
The above calculated as:
Np = ((2/FswHz)*Dmax%*(UinV - UdsV)) /4* Ac cm^2*Bmax G* 10^6
Where:
Dmax % - is for SINGLE pulse, always <50%.
UdsV - looses on impedance of open MOSFET`s
As I understand, the diL of output inductor will be recalculated to primary winding current, that will be used for to drive Isens input of PWM chip properly ? Its important on low loads...
For high Fsw windings geometry & placement is critical. Otherwrise Lk will produse a lot of rigning & heat on MOSFET`s.
Luka & Perry Babin suggested to wind primary first on core, & secondary over it. The others - at www.vlaveaudio.tripod.com , Rod Eliott, etc are suggesting to wind secondary first, & primary over it. Now I have trafo winded as Luka & Perry Babin suggested. At 25 KHz it working fine. What is you suggestions here ?
Why mistake, more is better for me.
But in you calculations I found one that can to contain mistake :Due to my calculations:
For toroid with Ac = 1.28cm^2 -
Np=4+4 Bmax (from zero to maximal positive or negative value) = 823 Gauss, So dB = 1646 gauss. Lprim = 64...65 uH, & I mag = 2.61A
The above calculated as:
Np = ((2/FswHz)*Dmax%*(UinV - UdsV)) /4* Ac cm^2*Bmax G* 10^6
Where:
Dmax % - is for SINGLE pulse, always <50%.
UdsV - looses on impedance of open MOSFET`s
As I understand, the diL of output inductor will be recalculated to primary winding current, that will be used for to drive Isens input of PWM chip properly ? Its important on low loads...
For high Fsw windings geometry & placement is critical. Otherwrise Lk will produse a lot of rigning & heat on MOSFET`s.
Luka & Perry Babin suggested to wind primary first on core, & secondary over it. The others - at www.vlaveaudio.tripod.com , Rod Eliott, etc are suggesting to wind secondary first, & primary over it. Now I have trafo winded as Luka & Perry Babin suggested. At 25 KHz it working fine. What is you suggestions here ?
Generally it`s ok, but about electronics I able to speak English or Russian only. I dont know any terminology on Hebrew
Hey, I found these MPP cores-
http://www.cwsbytemark.com/index.php?main_page=index&cPath=206_218
Are you talked abous such Alex ?
http://www.cwsbytemark.com/index.php?main_page=index&cPath=206_218
Are you talked abous such Alex ?
Tolik.
Diff-FB schematic:
Remove RV2,RV3,D3, no need to use high-quality opamps if FB loop. The loop bw must be 5-10% of Fsx at the most. In case of 50KHz Fsx, this means 2.5-5KHz max. LM358 will work fine cheaper and easier to get on the market.
The secondary side of the OC must go to the primary control side.
You may also tie R4 to the cathode of D2 (the zener diode). It is simpler.
CS Schematic:
You may use 1A Schottky for CS trafo.
Control section GND is not separated from power GND. Watch for oscillative behaviour.
Control schematic:
In control section sch' the FB network connected at the secondary side of the opto is kinda wierd... it better to use diff-feedback schematic type of feedback.
The opto-NPN is totally miswired.
Diff-FB schematic:
Remove RV2,RV3,D3, no need to use high-quality opamps if FB loop. The loop bw must be 5-10% of Fsx at the most. In case of 50KHz Fsx, this means 2.5-5KHz max. LM358 will work fine cheaper and easier to get on the market.
The secondary side of the OC must go to the primary control side.
You may also tie R4 to the cathode of D2 (the zener diode). It is simpler.
CS Schematic:
You may use 1A Schottky for CS trafo.
Control section GND is not separated from power GND. Watch for oscillative behaviour.
Control schematic:
In control section sch' the FB network connected at the secondary side of the opto is kinda wierd... it better to use diff-feedback schematic type of feedback.
The opto-NPN is totally miswired.
Hello Alex,
2 CM270125 - OD=26.9mm, ID=17.7mm, H=11.17mm
1 CM330125 - OD=33.02mm, ID=19.9mm H=10.6mm
All above is MPP 125 material.
Else I ordered T130-26 , & T106-26 iron powder cores to compare these to MPP.
For the current sensing transformer I ordered E-375-77, E core set. The size is 34.9x28.5x9.5mm. Al=2760. Permeability u= 2000.
There is no E cores with u > 2000, so if this core will preform bad, I`ll try to use either high permeability ferrite toroid for the current sensing transformer.
Thank you for you help & advise Alex.
So I ordered :
2 CM270125 - OD=26.9mm, ID=17.7mm, H=11.17mm
1 CM330125 - OD=33.02mm, ID=19.9mm H=10.6mm
All above is MPP 125 material.
Else I ordered T130-26 , & T106-26 iron powder cores to compare these to MPP.
For the current sensing transformer I ordered E-375-77, E core set. The size is 34.9x28.5x9.5mm. Al=2760. Permeability u= 2000.
There is no E cores with u > 2000, so if this core will preform bad, I`ll try to use either high permeability ferrite toroid for the current sensing transformer.
I attached modified schematic, please check if I understand you suggestions correctly. The RV2,RV3,D3 are interactive components, for simulation purposes only. The software used here is PROTEUS 7, that allowing me to play with values of interactive components during simulation, & to see changes in real time.
I meaned that PWM GROUND is ground of control system ONLY. It`s not connected to power common ground. I redrawed the schematic & attached it else. The PWM ground is ground of control cirquit only. The power ground is -12V common, that connected to sources of MOSFET`s , input capacitors , etc as I understand from our phone dialogue.
This cirquit is voltage mode controller that used for tests only. After MPP cores will be here (next week), I`ll make the new output inductor,& if the power stage will work good at 50KHz, I`ll starting to build the current mode controlling cirquit with UC2825, as we talked above.
Thank you for you help & advise Alex.
In CS application E core is good, toroid is alway better, it lacks any parasitic airgaps common in combined cores.
Dont forget to use gate drive trafo between the driver and the FET's. What is this -12V? Is it battery return? Or a new rail that you need?
Why are you wiring the opto-transistor in diff-amp-feedback with the LED as you do? What are you trying to acheive?
Hello Alex,
. It helps during simulation. It will not be used on real PCB.
Ok, my mistake, next time I`ll look for toroidal core. This not a problem.
Look here : http://focus.ti.com/lit/ds/symlink/ucc27324.pdf , this IC has TTL/CMOS Compatible Inputs Independent of Supply Voltage. Can I to avoid gate drive trafo usage with help of this IC ?
It`s a battery return.
This LED is another interactive component of PROTEUS 7
. It helps during simulation. It will not be used on real PCB.
Alex ùìåíI found some intresting design tool at magnetics website -
http://www.mag-inc.com/software/ctd1_2.zip
It`s seems to be helpful for the current trafo design.
What you think about it ?
My order from CWSbytemark (MPP cores ) shipped today, so next week it will be here, & I`ll be able to try them out.
So this weekend I`ll be able to play around 26 iron powder only.
Did you tried this material at 50KHz ? How it preforming ?
Hello Alex,
First, I must to consider that you was right, as always
I tried to run my setup at 50KHz with redesigned snubbers, & control circuit AC decoupled by 2 small inductors.
It works just prefect - output voltage is flat, even without TRANSIL diodes.
But drain waveforms are contain some ringing... I didn`t tried to build & test current transformer, but will I be able to acquire normal current ramp from trafo that working such way ?
First, I must to consider that you was right, as always
I tried to run my setup at 50KHz with redesigned snubbers, & control circuit AC decoupled by 2 small inductors.
It works just prefect - output voltage is flat, even without TRANSIL diodes.
But drain waveforms are contain some ringing... I didn`t tried to build & test current transformer, but will I be able to acquire normal current ramp from trafo that working such way ?
1. How did you drived the MOSFETS if the control GND is AC-HF decoupled from the output stage? I see no gate transformers...
2. What about the current ramp? If the circuit is still being run by SG3525 (Voltage mode) consider yourself lucky, however good the drain vaweforms are.
3. The drain waveforms looks manageable, not very high, as i thought they will be. I do NOT have a physical explanation for this, except that a step-up trafos have less primary leakage than 5%.
4. Do 22R snubbers get hot? You know, by changing to RCD snubber, you may save some power?
2. What about the current ramp? If the circuit is still being run by SG3525 (Voltage mode) consider yourself lucky, however good the drain vaweforms are.
3. The drain waveforms looks manageable, not very high, as i thought they will be. I do NOT have a physical explanation for this, except that a step-up trafos have less primary leakage than 5%.
4. Do 22R snubbers get hot? You know, by changing to RCD snubber, you may save some power?
Hi Alex,
The control is decoupled by 2 inductors as you suggested - one from +12V to control, second from control to ground. I noticed that control stage being to preform more stable. Temporary I have no suitable core to make gate trafo. CWS order not received yet.
Yes, It`s still be SG3525, I didn`t received core for current trafo.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.