If your guide is forward-specific then its better not to try use it for full-bridge design.Pungie said:
Lately I'm unsure of using a simple Linear transformer or SMPS power supply. I have a full guide with the design guide of forward topology and what I'm trying to figure out if I can use the same guide for full bridge? The circuit images to follow...
Help will be appreciated.
Why do you want full-bridge? Want more power? 2-transistor forward is well-capable to over 3kW. And way more troublefree.
If you are switching at...lets say over 80khz performance difference between full-bridge and 2-transistor forward is minimal, only output inductor needs to be bigger for 2-tranny forward, other parts work with similar ratings.
In comparison with a two-switch forward converter rated at the same output power:
- A full bridge uses the same amount of primary winding turns and half the amount of secondary winding turns.
- A full bridge produces the same output power with half the primary current.
- A full bridge produces the same output ripple voltage and current with half the output inductance.
- A full bridge requires double the number of switching devices and gate drive cells.
Figure out which one is better suited to your needs. A full bridge yields somewhat higher efficiency at the expense of circuit complexity.
- A full bridge uses the same amount of primary winding turns and half the amount of secondary winding turns.
- A full bridge produces the same output power with half the primary current.
- A full bridge produces the same output ripple voltage and current with half the output inductance.
- A full bridge requires double the number of switching devices and gate drive cells.
Figure out which one is better suited to your needs. A full bridge yields somewhat higher efficiency at the expense of circuit complexity.
Eva said:
-Same amount of primary turns IF core saturation is limiting factor. At around 100khz and above core _loss_ is the limiting factor and you can use less turns in forward's primary.
-see above, less turns on primary, double current, Dmax=0.5, -->samesame primary winding performance. Same Rdson-losses if equal number of mosfets used in both designs, two in parallei in case of 2-forward.
-ripple, agreed.
-same number of switching devices after certain power level(above few kW), have to parallei switches anyways.
btw:
4kW 2-transistor forward smps in less than 10x20x25cm box....
sw freq 160khz
IRG4PC50W 6pcs total
mzzj:
You are wrong regarding the core losses. Given the same core, the same clock frequency, the same amount of primary turns and working at maximum duty cycle, the full bridge converter will yield the same net magnetic flux swing as the forward converter, but the forward converter only produces flux in one direction while the full bridge produces symmetrical flux in both directions, thus the peak flux values reached in the full bridge converter are half than in the forward converter. This means that the full bridge will yield less than half the core losses and may be happily operated at a lower frequency without excessive core losses or saturation, while taking advantage of lower switching losses.
Also, I forgot to mention that the forward converter requires input filter capacitors with double the RMS ripple current rating than the full bridge converter, because it draws double the current during half the time. In some applications this may imply doubling the number of capacitors.
The major pitfall of the forward converter is that, even at maximum duty cycle, the transformer rests idle and unused for half the time, so double the output energy has to be transferred during the other half of the time.
The second major pitfall of forward converters is single-ended magnetic flux in the core, that causes higher losses than the same flux excusrsion achieved in a symmetrical way, because the peak flux density reached is doubled and the core is stressed as if it were operated at half the frequency.
If you achieved 4kW in less than 10x20x25cm with a forward converter running at 160Khz, with a full bridge you could achieve the same power output in the same or slightly bigger size (due to more complex gate drive), but using a lower swiching frequency and reducing power losses uo to the half.
The case of the two-transistor forward converter is similar to the case of the current doubler output. Something is simplified and made cheaper at the expense of worsening the rest.
However, with the fast and cheap IGBTs now available, whose current capability exceeds by far any MOSFETs with the same die size, and whose conduction losses are no longer proportional to I^2 thus allowing for happily doubling the current without getting unacceptably high conduction losses, I expect to see a lot of two-transistor forward stuff from now-on. Current SMPS trends are towards smaller size and cost at the expense of poor 80% efficiencies (while my mind still thinks in 95% efficiency terms at the expense of some extra size and weight... you know, I did a 1800W converter using four $1 bipolar transistors as switches mounted on a pair of 3.75ºK/w heatsinks, and the thing is certainly bulky but without a fan it can work at maximum power for a few minutes without overheating).
You are wrong regarding the core losses. Given the same core, the same clock frequency, the same amount of primary turns and working at maximum duty cycle, the full bridge converter will yield the same net magnetic flux swing as the forward converter, but the forward converter only produces flux in one direction while the full bridge produces symmetrical flux in both directions, thus the peak flux values reached in the full bridge converter are half than in the forward converter. This means that the full bridge will yield less than half the core losses and may be happily operated at a lower frequency without excessive core losses or saturation, while taking advantage of lower switching losses.
Also, I forgot to mention that the forward converter requires input filter capacitors with double the RMS ripple current rating than the full bridge converter, because it draws double the current during half the time. In some applications this may imply doubling the number of capacitors.
The major pitfall of the forward converter is that, even at maximum duty cycle, the transformer rests idle and unused for half the time, so double the output energy has to be transferred during the other half of the time.
The second major pitfall of forward converters is single-ended magnetic flux in the core, that causes higher losses than the same flux excusrsion achieved in a symmetrical way, because the peak flux density reached is doubled and the core is stressed as if it were operated at half the frequency.
If you achieved 4kW in less than 10x20x25cm with a forward converter running at 160Khz, with a full bridge you could achieve the same power output in the same or slightly bigger size (due to more complex gate drive), but using a lower swiching frequency and reducing power losses uo to the half.
The case of the two-transistor forward converter is similar to the case of the current doubler output. Something is simplified and made cheaper at the expense of worsening the rest.
However, with the fast and cheap IGBTs now available, whose current capability exceeds by far any MOSFETs with the same die size, and whose conduction losses are no longer proportional to I^2 thus allowing for happily doubling the current without getting unacceptably high conduction losses, I expect to see a lot of two-transistor forward stuff from now-on. Current SMPS trends are towards smaller size and cost at the expense of poor 80% efficiencies (while my mind still thinks in 95% efficiency terms at the expense of some extra size and weight... you know, I did a 1800W converter using four $1 bipolar transistors as switches mounted on a pair of 3.75ºK/w heatsinks, and the thing is certainly bulky but without a fan it can work at maximum power for a few minutes without overheating).
Bridge is best path "forward"
Pungie,
EVA is right (as always!) about what I will call double-ended forward converters -v- single-ended ones. You can use a two transistor forward converter and still not be using the transformer core to its full potential. Remember she said it sits idle and unused for at least half the time. And you still have to deal with the LC output filter maintaining the output voltage for at least half the cycle. Since you want to do a two transistor forward, you already have all that you need to do a two-transistor half-bridge: the two transistors, a driver transformer, a double diode (assuming you only want (+) output) and a (single output) control IC.
You will only have to replace the single-output control IC with a double-output one, like the venerable TL494, SG3525, UB1846, or MC33025, etc. All these PWM chips wil fit the bill, and the last two are current-mode ones, enabling much tighter control of the output voltage(s). (Though the TL494 would need a driver circuit to drive the driver transformer
)
Using a half-bridge topology will eliminate all of the setbacks of the forward converter that EVA points out. I'm not trying to disparage the single-ended forward converter, it just seems that, in your case, a half-bridge converter would be a much more efficient use of the materials at hand.
Hope this helps clarify things a bit.
Steve
Pungie,
EVA is right (as always!)
about what I will call double-ended forward converters -v- single-ended ones. You can use a two transistor forward converter and still not be using the transformer core to its full potential. Remember she said it sits idle and unused for at least half the time. And you still have to deal with the LC output filter maintaining the output voltage for at least half the cycle. Since you want to do a two transistor forward, you already have all that you need to do a two-transistor half-bridge: the two transistors, a driver transformer, a double diode (assuming you only want (+) output) and a (single output) control IC.You will only have to replace the single-output control IC with a double-output one, like the venerable TL494, SG3525, UB1846, or MC33025, etc. All these PWM chips wil fit the bill, and the last two are current-mode ones, enabling much tighter control of the output voltage(s). (Though the TL494 would need a driver circuit to drive the driver transformer
)Using a half-bridge topology will eliminate all of the setbacks of the forward converter that EVA points out. I'm not trying to disparage the single-ended forward converter, it just seems that, in your case, a half-bridge converter would be a much more efficient use of the materials at hand.
Hope this helps clarify things a bit.
Steve
The same transformer with the same turns ratio will suit both the half bridge and the two-transistor forward converter, but the output power for the latter will be twice for the same copper losses (assuming same ripple current). However, given the same clock frequency, the half bridge will allow to either operate the transformer at up to one fourth of the frequency or reduce the turn counts up to one fourth.
I believe that big companies are opting for things like forward converters and current doublers for mass production because assembling them is somewhat cheaper due to the lower part count, even at the expense of reduced performance, greater power losses and increased component stress.
I believe that big companies are opting for things like forward converters and current doublers for mass production because assembling them is somewhat cheaper due to the lower part count, even at the expense of reduced performance, greater power losses and increased component stress.
This may be the only real advantage of forward topologies, because you can put a half or full bridge in severe trouble if you draw current from the output at the switching frequency and the control circuit tries to compensate for it too much (that will unbalance the converter).
In fact, I have suffered these problems and I'm considering a special circuit to circumvent that (mostly based in an integrator that forces symmetrical pulses and a PWM comparator that can only control the lenght of the odd pulses).
In fact, I have suffered these problems and I'm considering a special circuit to circumvent that (mostly based in an integrator that forces symmetrical pulses and a PWM comparator that can only control the lenght of the odd pulses).
SMPS for disco/club amplifier
Thanks all for the advice. What else I would like to know or can someone correct me on the following.
The guide I have on the Forward-converter says that this 2-switch design's power level is from 100W to 500W. The power requirements for the device requiring the power are as follow: 55V 21.81A(Max) (1200W)
12V 1A(Max) (12W)
Total Max Power Output = 1212W
I have read the other replies in this thread. Some of the information I'm unsure of. I'm still new to the design process of smps. Eva and N-Channel. Say this was a power supply you had to design, that would be used for an Amplifier for a disco/club rated at this high power then? Would it be 2 switch forward, half or full bridge? Then the diagram I supplied in the start of the thread, how would that schematic change or what other schematic would be usefull for the other topologies for half and full bridge? Please guide me.
The second problem I'm experiencing with the design is the original circuit uses mosfet IRF840 which brougth me to look at it's data sheet. The maximum power disappation = 125W with TC = 25 degrees. I have been looking for other mosfets on the net with no luck. What else should I be using then?
Thanks all for the advice. What else I would like to know or can someone correct me on the following.
The guide I have on the Forward-converter says that this 2-switch design's power level is from 100W to 500W. The power requirements for the device requiring the power are as follow: 55V 21.81A(Max) (1200W)
12V 1A(Max) (12W)
Total Max Power Output = 1212W
I have read the other replies in this thread. Some of the information I'm unsure of. I'm still new to the design process of smps. Eva and N-Channel. Say this was a power supply you had to design, that would be used for an Amplifier for a disco/club rated at this high power then? Would it be 2 switch forward, half or full bridge? Then the diagram I supplied in the start of the thread, how would that schematic change or what other schematic would be usefull for the other topologies for half and full bridge? Please guide me.
The second problem I'm experiencing with the design is the original circuit uses mosfet IRF840 which brougth me to look at it's data sheet. The maximum power disappation = 125W with TC = 25 degrees. I have been looking for other mosfets on the net with no luck. What else should I be using then?
Attachments
I assume that your amplifier is going to have forced air cooling and that slightly increased power losses doesn't matter. On the other hand, you appear to have not much experience and the forward is just easier to deal with, so I would go for it if I was you.
Forger about MOSFETs as switches. MOSFETs have become outdated and inefficient in high-voltage medium and low speed switching applications at frequencies below 150Khz. You should look for fast IGBTs instead.
Forger about MOSFETs as switches. MOSFETs have become outdated and inefficient in high-voltage medium and low speed switching applications at frequencies below 150Khz. You should look for fast IGBTs instead.
Eva said:
Less than half core loss? based on what theory?
DC biasing effects on core loss are more kinky than that. difference can wary wildy depending on material, peak flux densities and flux swing.
See for example:
"Simple Approximations of the DC Flux Influence
on the Core Loss Power Electronic Ferrites
Their Use in Design of Magnetic Components
Wai Keung Mo, Member, IEEE, David K. W. Cheng, Member, IEEE, and Y. S. Lee"
I can't read that paper because it's pay-per-view.
If you look at the datasheet of any power ferrite material, the core losses versus frequency are specified as a function of PEAK flux density, not as a function of AC flux amplitude. The same graphs reveal that halving the PEAK flux while keeping the same frequency not only halves losses but may even reduce them by an order of magnitude for some materials.
If you look at the datasheet of any power ferrite material, the core losses versus frequency are specified as a function of PEAK flux density, not as a function of AC flux amplitude. The same graphs reveal that halving the PEAK flux while keeping the same frequency not only halves losses but may even reduce them by an order of magnitude for some materials.
Power Dissapation
Good morning every one. Busy with calculations on my design. One thing I like to know. (Perhaps a stupid question). I'm selecting IGBT's for the switching. In the data sheets it says 140W Power Dissapation. If the IGBT is rated 200V 30A then V X I isn't = to the power dissapation. What is power dissapation then?
Good morning every one. Busy with calculations on my design. One thing I like to know. (Perhaps a stupid question). I'm selecting IGBT's for the switching. In the data sheets it says 140W Power Dissapation. If the IGBT is rated 200V 30A then V X I isn't = to the power dissapation. What is power dissapation then?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.