one transistor forward converter...or flyback?

[eem2am]

With Reference to Offline SMPS's in the lower power range.

Why would you ever use a one-transistor forward converter instead of a flyback?

There are absolutely no advantages in favour of the one transistor forward converter.

The one transistor forward converter needs an extra de-magnetisation winding, which means more expense.

Also, the RMS FET current is just the same as in flyback, since (with a demag winding with the same turns ratio as the primary) the one transistor forward converter can only have a maximum duty cycle less than 0.5.

Also, the one transistor forward converter needs an output inductor.

So why would anyone ever use a one transistor forward converter?

 

[Orson Cart]

the single transistor forward converter does not need a de-mag winding - you can just have high turn off volts on the device and limit the max ON time.
forward converters are used for higher output currents where flybacks are not so efficient - also the current in the mosfet has a lower rms rating than for a flyback (discontinuous) - hope this answers your questions.
Regards, Orson Cart.

 

[u04f061]

For the same power ratings, transformer size of forward converter is smaller than flyback. Output current ripples in forward converter is also less than flyback. In CCM mode, right half plane zero in flyback poses a control challenge.

 

[eem2am]

hi

the thing is a CCM flyback has just as low RMS FET current as a forward converter.

so why use a forward converter?

i am certain a demag winding is needed for forward converter.

 

[Orson Cart]

CCM flyback has 10x greater noise - so harder to meet EMC, many forward converters have been built with no de-mag winding - just a light snubber on the main mosfet, having a demag winding may limit your ON time for the mosfet, Regards, Orson Cart.

 

[u04f061]

the thing is a CCM flyback has just as low RMS FET current as a forward converter.

What about absence of right half plane zero in forward converter?

 

[eem2am]

The EMC for 1 transsistor forward is just as bad as flyback.

This is because the forward has highly discontinuous FET current.

It is not realistic to dump the demag energy into a snubber.

I agree you cannot drive a 1 transistor forward above 45% duty if your demag winding has the same number of turns as the main primary.

You can have higher duty if you have more turns in the demag than the primary, but then your FET off-state voltage increases.

 

[eem2am]

If there are two 40W offline SMPS's operating as follows, both with V(in) = 330V.............

1....a CCM flyback with duty cycle 45%.
2....a 1 transistor forward converter with duty cycle = 45%

....then its obvious that the RMS FET current in each converter would be pretty much the same.

Therefore, i cannot understand why the 1 transistor forward converter is supposed to be more suitable for higher power than a flyback.

(CCM does have a RHPZ , but this is not such a problem...one simply reduces the Zero-Gain Crossover frequency to suit)

 

[Orson Cart]

Depending on the o/p current the forward will be more efficient - with quite a bit less RFI, and a lot faster control - Regards, Orson Cart.

 

[eem2am]

There is no reason why a 1 transistor forward converter should be any more efficient than a flyback.

....Regarding RFI, both flyback and 1 transistor forward converter are similar in RFI......

...this is because both have a discontinuous FET current.
....and if anything, the 1 transistor forward converter is worse for RFI because the secondary series diode is hard-switched OFF. (switched off whilst carrying full secondary current)

A 1 transistor forward converter is not necessarily with a faster control than a flyback....
....after all, a flyback can be done in current mode..............it is really stretching it to say that a 1 transistor forward converter can be done in current mode.

...in current mode, you need a decent amount of ramp of the current.

 

[Orson Cart]

usually forward converters are done using peak current mode to limit the max current in the switch - you can have any amount of current ramp in the choke current that you desire, you will find the output diodes are harder switched in a CCM flyback - this is the reason for the increased RFI, in a forward converter the peak currents need not be much higher than the average - but in a CCM flyback the peak currents on the main switch and o/p diodes tend to be up to twice as high (or more) than the average - when you have buiit a few you may see why...
Regards, Orson Cart.

 

[FvM]

I'm not involved with off-line SMPS design, thus I can't verify the statements from own eperience. I was however under the impression, that low power off-line SMPS are generally using flyback mode. Although some claims of forward topology advantages sound plausible, I'm not aware of industry standard products that are using it. In my understanding, the requirement of wide input voltage range, at least 100 - 240 VAC, already suggests flyback mode.

Reference designs for flyback off-line switchers are given for nearly any switch mode controller and integrated switcher IC. Can you refer to a forward switcher reference designs with similar application parameters?

 

[Orson Cart]

FvM is correct - flyback is first choice for manufacturers - but almost exclusively DCM - as can be seen from the PowerInt website where they have almost zero designs featuring CCM (for EMC reasons) - at the 40W level flyback (DCM) is a good choice. The discussion below was centered on forward vs CCM flyback - forward is much easier to get a good working result than CCM flyback - especially at output currents above 5 amp. CCM flyback does work but the issues of RFI(EMC) and switching losses (snubbing the turn off) make it more of an effort to tame than forward - which is very straightforward...
Regards, Orson Cart.

 

[macubo]

Hi all, I'm in the very same situation when choosing between 1-mos forward and CCM flyback for wide-input 90W, 14.6Vout battery charger.

Forward has 3 diodes in power-plant, output inductor and demag winding, while flyback just has 1 diode, 1 xformer, and probably the same voltage rating for the MOS when operating a 1:1 pri:demag turn ratio for the forward.

slow response in flyback wouldn't be a problem, as per battery charger response.
The problem is the 6Amps o/p current

The overall cost is to be absolutely minimized, less than 15EU for a stated consumption of 1.5Kpcs/y. And EMC is really an issue and it increases EMI filter cost.

Do you have some suggestion to operate the best possible choice?
Thank you

 

[Orson Cart]

This could be done with a DCM flyback with two o/p windings and two diodes to share the 6amp out, PowerInt TOP249 or 250 to keep the temp down, We have designed these up to 32V 8A output (three diodes) 180-265VAC, at 90W you will be able to get 90-265VAC - all down to good transformer design with low and similar leakage and good coupling for the o/p windings.
Regards, Orson Cart.

 

[macubo]

This could be done with a DCM flyback with two o/p windings and two diodes to share the 6amp out, PowerInt TOP249 or 250 to keep the temp down, We have designed these up to 32V 8A output (three diodes) 180-265VAC, at 90W you will be able to get 90-265VAC - all down to good transformer design with low and similar leakage and good coupling for the o/p windings.
Regards, Orson Cart.

I guess you mean bifilar or trifilar winding on secondary, with diodes in parallel, do you? What efficiency were you able to reach? Our design should have a 88% min at 230Vac, full load.

 

[Orson Cart]

Bifilar is better (than side by side secondaries - although these do work acceptably well - (TIW - triple insulated wire) and one diode per winding with its own initial cap to keep the loop area down, then paralleling, efficiency is limited by the use of the TOP250 and the forward drop of the output diodes, e.g. for 14.6 vout, a 0.8 volt forward drop loses you 5.6% in efficiency straight away. This is why a flyback eff of 88-89% is considered good. Regards, Orson Cart.

 

[macubo]

for 14.6 vout, a 0.8 volt forward drop loses you 5.6% in efficiency straight away.

You got the point. The remaining 7W of power losses must be shared between transformer core&copper losses, shunt resistors on primary and secodary side, MOSFET conduction&switching losses, EMI filter&diode bridge drop, and biasing&startup circuitry.

At 230Vac, the worst seems to be MOSFET switching losses at turn-off, which for a flyback depends on peak primary current, forward transconductance of the device, gate resistor and driver, reflected voltage and rectified mains. To reduce the peak primary current, CCM would be suitable when compared to DCM, along with a low reflected voltage.

Do you agree?
Best regards, Manuele

 

[Orson Cart]

Turn off losses can be near zero for a gate turn off of <40nS (12v -> 0v) it is the turn on losses that usually dominate in a flyback, CCM has its share of problems, RHP zero in the control loop, very hard switching of the o/p diodes, even more turn on losses in the mosfet due to the Irr in the output diode - more RFI to deal with, which is why DCM or boundary mode is used up to 250W and beyond... Regards Orson Cart.

 

[macubo]

Well, I don't think you are right. I made some preliminary calculations, leading to almost 6W switching turn-off losses, with an expected 96nsec current-fall-time and a gfs of 8S (@265Vac, starting from a peak drain current of 3.5Amps)

How can turn-off losses be so small for you?
You must walk the Miller plateau all way down, during which Vds rises from Ipk,pri*Rds,on to the overshoot drain voltage (caused by the leakage inductance), and then bring the MOS' gate from Vgs,miller to Vth, where drain current diminishes.

Turn-on losses are capacitive, in DCM or boundary mode.
In CCM, the drain-source switching transition during the drain current rising to the non-zero initial value, increases the turn-on losses, plus those due to the recovery charge for o/p diode (schottky here?).

In boundary conduction mode, for a 1nF estimated total drain capacitance, turning-on on the valley of the resonant oscillation between Lpri and Cd,tot, even at 265Vac, losses seem to be less than 50mW.

ciao, manuele