Hi All,
I have just finished the first draft of the schematics of a 500W offline SMPS. The chosen topology is the double switch forward.
Please take a look at that and let me know what do you think.
This is my first design of an offline SMPS (some experience with low voltage but never tried offline), so any comments or suggestions are highly appreciated.
Thank you
ciao
-marco
I have just finished the first draft of the schematics of a 500W offline SMPS. The chosen topology is the double switch forward.
Please take a look at that and let me know what do you think.
This is my first design of an offline SMPS (some experience with low voltage but never tried offline), so any comments or suggestions are highly appreciated.
Thank you
ciao
-marco
By now I have only the schematic, I have not yet started the PCB layout.
Current limit is included in the SMPS chip. The IS input is used for current mode feedback but also as an ever current protection.
When the voltage on is is >2V the OCP trips. I have set it to 10A, maybe a bit too high.
@start882
Why are you sayin this? It is because the two ouptut inductors are separated?
the +/-12V is used to power op amps and other low power stuff, the maximum current on that output will be under 100mA.
Current limit is included in the SMPS chip. The IS input is used for current mode feedback but also as an ever current protection.
When the voltage on is is >2V the OCP trips. I have set it to 10A, maybe a bit too high.
@start882
Why are you sayin this? It is because the two ouptut inductors are separated?
the +/-12V is used to power op amps and other low power stuff, the maximum current on that output will be under 100mA.
Thank you darkfenriz,
of course if I short the +12V there will be no current suppling the
optocoupler. This will open the feddback loop and make the PWM go at maximum duty cycle (and thus at maximum output voltage).
I will change the schematic deriving the optocoupler power supply directly from the +85V rail trough a resistive divider.
@ifrythings
In a double switch forward topology the energy goes from the bus capacitor to the load only when both switches are closed.
When the switches are open the two diodes will reset the transformer core.
The current sense resistor is therefore places in series with the two mosfet because it the real measure of the current supplied to the load
ciao
of course if I short the +12V there will be no current suppling the
optocoupler. This will open the feddback loop and make the PWM go at maximum duty cycle (and thus at maximum output voltage).
I will change the schematic deriving the optocoupler power supply directly from the +85V rail trough a resistive divider.
@ifrythings
In a double switch forward topology the energy goes from the bus capacitor to the load only when both switches are closed.
When the switches are open the two diodes will reset the transformer core.
The current sense resistor is therefore places in series with the two mosfet because it the real measure of the current supplied to the load
ciao
Hi All,
I have finished my SMPS schematics taking into account your suggeestions.
Now I have connected the opto power supply before the 7812 so that is the 12V output is shorted the opto will always have a power supply.
@ darkfenriz
I don't know if D11 and D12 are really needed, in any case I have had experience with low voltage SMPS where these diodes can help a lot. I think it worth to keep the place on the PCB and if they are not needed it is easy to remove them.
In attached you will find my hopefully final schematics, the PCB layout and the preliminary schematics of the audio amplifier that this SMPS will supply.
Thank you
ciao
-marco
I have finished my SMPS schematics taking into account your suggeestions.
Now I have connected the opto power supply before the 7812 so that is the 12V output is shorted the opto will always have a power supply.
@ darkfenriz
I don't know if D11 and D12 are really needed, in any case I have had experience with low voltage SMPS where these diodes can help a lot. I think it worth to keep the place on the PCB and if they are not needed it is easy to remove them.
In attached you will find my hopefully final schematics, the PCB layout and the preliminary schematics of the audio amplifier that this SMPS will supply.
Thank you
ciao
-marco
Marco,
I agree with Luka, Perry, ifrythings, & the others: Use a half-bridge. Since you already have two MOSFETs and the driver chip, going to a half-bridge effectively doubles core usage, and power throughput. You can now use D14-D17 in full-bridge fashion for lower ripple on the output. Also, because the core automatically resets by nature of half-bridge operation, no gap is needed. Not saying you have to do this, but, as m-tech said, you can leverage the vast amount of discussion on half-bridge threads in this forum.
Also, I would change the output sensing scheme from +85V & output ground, to (+) and (-) 85V out , as this will give true symmetric sensing about the 0-V line. This has been discussed (at length) on other threads in the Power SUpply forum and can be a source of 'agreeing to disagree", but it's worth a look.
Just my two cents' worth.
Cheers,
Steve
I agree with Luka, Perry, ifrythings, & the others: Use a half-bridge. Since you already have two MOSFETs and the driver chip, going to a half-bridge effectively doubles core usage, and power throughput. You can now use D14-D17 in full-bridge fashion for lower ripple on the output. Also, because the core automatically resets by nature of half-bridge operation, no gap is needed. Not saying you have to do this, but, as m-tech said, you can leverage the vast amount of discussion on half-bridge threads in this forum.
Also, I would change the output sensing scheme from +85V & output ground, to (+) and (-) 85V out , as this will give true symmetric sensing about the 0-V line. This has been discussed (at length) on other threads in the Power SUpply forum and can be a source of 'agreeing to disagree", but it's worth a look.
Just my two cents' worth.
Cheers,
Steve
Thank you N-channel,
I don't think I will move to an half-brigde topology. This is mainly because this double switch forward is something I designed by myself from scratch.
If I move to an half brigde design the temptation of copying someone else design (that I can easily find on this forum) with minor modifications will be too strong for me.... This is mainly an experiment to see if I am able to design an off line SMPS and to see if it can meet what I need.
Another point that make me prefer the double switch forward is the availability of the controller chip. There are plenty of high perfomances current mode control chip as the FAN7554 that are readily availables. What about controllers that allows half-bridge? Except from the old 3525 and similar I can not find anything else
that implements all the features needed (mainly the dead time control). Going with 3525 means no current limit and no current mode control....
About the output rectifer: what is the benefit of connecting the diodes in full brigde way? Can I do it with my current design or I need an half-bridge.
Thank you for pointing me out the voltage to control trough the feedback loop. In any case I think that sensing only the positve +85V rail or sensing the difference between the positive and the negative rail will be quite the same on my application.
The positive and the negative output are are coupled trough the output buck inductance, moreover the amplifier this SMPS will supply has two channels wired out of phase. This means that the current on the positive rail will be more or less equal to the current on the negative rail; so what the benefit of sensing the difference between the two rails?
thank you
ciao
-marco
I don't think I will move to an half-brigde topology. This is mainly because this double switch forward is something I designed by myself from scratch.
If I move to an half brigde design the temptation of copying someone else design (that I can easily find on this forum) with minor modifications will be too strong for me.... This is mainly an experiment to see if I am able to design an off line SMPS and to see if it can meet what I need.
Another point that make me prefer the double switch forward is the availability of the controller chip. There are plenty of high perfomances current mode control chip as the FAN7554 that are readily availables. What about controllers that allows half-bridge? Except from the old 3525 and similar I can not find anything else
that implements all the features needed (mainly the dead time control). Going with 3525 means no current limit and no current mode control....
About the output rectifer: what is the benefit of connecting the diodes in full brigde way? Can I do it with my current design or I need an half-bridge.
Thank you for pointing me out the voltage to control trough the feedback loop. In any case I think that sensing only the positve +85V rail or sensing the difference between the positive and the negative rail will be quite the same on my application.
The positive and the negative output are are coupled trough the output buck inductance, moreover the amplifier this SMPS will supply has two channels wired out of phase. This means that the current on the positive rail will be more or less equal to the current on the negative rail; so what the benefit of sensing the difference between the two rails?
thank you
ciao
-marco
Marco,
Thanks for your thoughtful and detailed reply. Let me answer a couple of questions in your post that caught my eye:
1) Availability of 2-Ch PWM controllers: true, the 3525 is a voltage-mode chip, though a crude form of current-mode control can be realized by conditioning a current-sense signal via op-amp, and feeding that into the Shutdown Pin (10). However, there are the Unitrode (not part of TI) UC1846 current-mode chip, OnSemi's MC33025 hi-speed current-mode chip, UCC3806 (the BiCMOS version of the UC1846), and TI's UC1856 hi-speed current-mode chip, just to name a few.
2) Why full-bridge diodes? I meant this scheme for half-bridge, not single-ended forward.
Have to run, but I'll pick up on this tomorrow.
Bye,
Steve
Thanks for your thoughtful and detailed reply. Let me answer a couple of questions in your post that caught my eye:
1) Availability of 2-Ch PWM controllers: true, the 3525 is a voltage-mode chip, though a crude form of current-mode control can be realized by conditioning a current-sense signal via op-amp, and feeding that into the Shutdown Pin (10). However, there are the Unitrode (not part of TI) UC1846 current-mode chip, OnSemi's MC33025 hi-speed current-mode chip, UCC3806 (the BiCMOS version of the UC1846), and TI's UC1856 hi-speed current-mode chip, just to name a few.
2) Why full-bridge diodes? I meant this scheme for half-bridge, not single-ended forward.
Have to run, but I'll pick up on this tomorrow.
Bye,
Steve
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.