Flyback core gap is too big?

[eem2am]

Hello,

My company is doing an offline, isolated Quasi-Resonent flyback power supply.

However, they have chosen to use an airgap of length 2040um (2.04mm).
The airgap is in the central core of the EE24 transformer.

I am sure that this airgap is rather too big and will mean overly high leakage inductance.

Do you agree?

Here is the spec of this QR Flyback:-
Vin = 370V (DC)
Vout = 16V
Pout = 54W
Switching frequency at max load = 67KHz.
AL (nH/N^2) of gapped core = 70
Np = 61
Ns = 19

Do you agree that the 2.04 mm gap is excessively large and a smaller gap should be used with less turns?

 

[mtwieg]

Can't tell with that info alone. Air gap prevents saturation, so to know for sure you'd need to give the core area, core saturation flux density, max on time (is it constant on time?). Off the top of my head, 2mm sounds kind of big, but it could be completely justified given the high input voltage and low switching frequency. And does leakage matter that much in a quasiresonant supply? I thought the leakage was used as part of the resonance and the leakage energy is recovered?

 

[FvM]

For a transformer, the number of turns for given core size will be choosen according to acceptable flux (either saturation or core loss limited). When the number of turns has been fixed, the air gap can be adjusted for the intended inductance respectively peak current of a flyback transformer.

Although a large air gap involves some disadvantages, you can't arbitrarily reduce the number of turns (unless it has been unreasonably high before).

By the way, the said windings ratio sounds unusual, resulting ín unsuitable low duty cycle. It may be a tribute to a primary switch with a low voltage rating?

 

[eem2am]

i apologise for my lack of information.

mtwieg:
it is difficult for me to get more info here as security is so tight at work.....i have only managed to remeber the above.
...the 2mm air-gap, and very low AL value sound very low to me...not what i am used to.
-it is indeed constant on time

i am afraid that leakage energy is definetely not recovered in a QR Flyback design.
-our design includes a dissipative clamp for the leakage (RCD)


FvM:
Absolutely right...the duty cycle is indeed very low, even at max load.........its around 0.121 as i remember it.
-the max peak switch current is around 2.8A. (if i remember correctly)

-As i mentioned, security is so tight and i don't have more info to hand now.

The primary switch is internal to the controller chip.........this controller chip is from Sanken Electrical Company, and it is STR2A153D.
-i think the RdsON was around 1.8R

...strangely , i can find absolutely no data for this chip on the internet.

Also, i believe that the low duty cycle was favoured because it means a longer secondary conduction time........which means lower rms secondary diode current....and less dissipation in the secondary diode.
Also, Ns/Np should be as close to 1 as possible, in order to reduce leakage current.
-Very low Ns/Np gives high leakage current, which equates to high losses in the primary clamp circuitry.

 

[Orson Cart]

Ns/Np is only 3:1 which is not soooo high, the design of this Tx is a bit unusual for an essentially offline flyback, however it is the coupling of the primary winding to the secondary that determines the leakage - the core has only a minor third order effect, a sandwhich winding will reduce the leakage, the fringing from the large gap may cause additional losses in the first (or all) layers of the windings. Regards Orson Cart.

 

[mtwieg]

Yeah, there's a lot that sounds weird about the design, especially 2.8A peak switch current. Your peak/average ratio is almost 20, which seems huge.

As for the low turns ratio, it might be to limit the Vds the switch sees during turn off. With the specs you gave, the Vds should be about 421V (neglecting the overshoot from leakage). This is conjecture, but it may be that they are using a 500V FET and don't want to push Vds any higher.

And also, I can't even imagine working a job where I'm not allowed to scrutinize design details. The just blows my mind.

 

[FvM]

i can find absolutely no data for this chip on the internet

There's a japanese datasheet, but I can only read a few numbers, e.g. VDSS of 650 V.

Apparently, the device isn't in the international product portfolio, but there's a similar device.

Also, i believe that the low duty cycle was favoured because it means a longer secondary conduction time........which means lower rms secondary diode current....and less dissipation in the secondary diode.

In return, the primary switch losses and required diode voltage rating will be very high.

 

[eem2am]

Thankyou.

all excellent comment.

You may also have noted my embarassing mistake.....i thought the STR2A153D was a Quasi-resonant type.....but in fact its just a plain current mode PWM chip.

.....the correct story is that this plain flyback is meant to replace the existing QR Flyback......(so presumably, QR Flybacks aren't all that cracked up to be....)

 

[rvionics]

The 2mm Gap will have more gap related losses, The reflected voltage on primary looks very small (54~55V) due to the selected ratio (3.21)
QR flyback will be more efficient compared to normal CCM or DCM current mode flyback as it gives an zero voltage operation for primary FET
& no ultrafast recovery diode required on the secondary as the operation is zero current turn off always. The disadvantage will be large secondary current & variable frequency but 54W at 16V looks managable unless it is difficult to manage primary currents at lower input voltages.

 

[eem2am]

hello,

QR flyback is , i believe not worth the while.

We are replacing QR with standard flyback because Ns/Np is greater with standard flyback......and this means less leakage inductance.....which means less leakage related losses.............also, diode RMS current much higher with QR solution.

We are replacing QR with straight flyback , and no longer need heetsink for fet.

QR is a waste of time.

...you got to have ridiculously low Ns/Np to get drain voltage to swing to anywhere near zero volts.

 

[mtwieg]

Yeah, I always suspect that QR converters often don't give much benefit. It's probably especially worthless in flyback, since the core losses and leakage inductance losses are far greater than turn on losses in the FET.

With flyback you'll always be stuck with efficiency around 85% or less. If you want much better you'll have to find another topology entirely.

 

[Orson Cart]

Where you have higher powers (e.g. 300W-1kW) and high drain voltages on the mosfet, QR can become quite attractive as the turn on losses are related to Cds x V^2, not to mention the EMC benefit, for smaller flybacks as mtwieg says there is not huge benefit. Fully discontinuous design is advised (to keep turn on losses down and RFI from the o/p diode). Transformer should be optimised to get a reasonably low flyback voltage while keepng the diode voltage within reasonable bounds also, sandwhich winding of the Tx will halve the leakage (1/4 for 5 layer sandwhich) which will lower the energy in the turn off spike. Regards, Orson Cart.

 

[eem2am]

Thankyou,

I have many conflicting articles on CCM vs DCM flyback, with regard to the switching losses and EMC....i am not sure who to believe.

I have heard of sandwich winding a secondary between two primary layers.

But are you suggesting that in a single output flyback...... that by splitting the secondary into two separate layers, and sandwiching these between three primary layers gives even less leakage inductance?

 

[Orson Cart]

Correct re the 5 layers and lower leakage, CCM flyback also has a RHP zero which makes closing the (voltage) feedback loop trickier, than for fully DCM (unless you use peak current mode control), Regards, Orson Cart.

p.s. We have redesigned a number of flyback TX for clients to make them fully DCM, and the improved results for lower EMC and lower uP intereference are marked.

 

[eem2am]

Thankyou Orson,

The thing is, powerint.com do not agree with yourself, powerint.com take the opposite stance to your own regarding CCM vs DCM flybacks...
(if you refer to Puzzler 3 on the powerint.com website....

**broken link removed**


But i would beg to refer yourself to powerint.com

..if you go there..then....

Click "Community" tab...then....
Click "Papers, Circuit ideas, Puzzlers"......then
Click "Puzzlers"...then.....
Click "Puzzler 3..optimizing multi-output designs"...then....
Scroll down to question 3....then....
Click "Show the Answer".......then you may read the following regarding flybacks...............

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Continuous conduction mode has the following major advantages:

Lower RMS currents yield higher efficiency
Lower peak currents lead to better cross regulation in multiple output designs
Lower peak currents also lead to better differential mode conducted (and radiated) EMI performance (or smaller input capacitors).
Lower peak and RMS currents contribute to better output ripple (or smaller output capacitors)

Continuous mode operation gives higher loop gain and therefore has faster transient response than discontinuous mode
In it’s basic form current mode cannot go beyond 50% duty cycle without risk of instability. A maximum duty cycle limit of 50 % causes peak currents to be higher for a given power level than if the duty cycle were allowed to increase. It is possible to overcome the 50% duty cycle limit by using slope compensation but this requires careful choice of components.

Conversely Voltage mode in its basic form, does not have a limitation on maximum duty cycle. This allows deeply continuous voltage mode designs providing their inherent advantages.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

 

[Orson Cart]

Ah yes, the power-int site, oh how nice it is for the power-int app note writers to write about something that they have never really built and tested and made into a sale-able product.

Lower RMS currents only yeild higher efficiency if they more than offset the higher switching losses (they don't)
In practice cross regulation if just as good in fully DCM designs (it is really determined by coupling)
TRue - lower peak currents allow smaller inputs caps (but not o/p) but not less radiated EMI due to the increased switching noise of the now harder switching in CCM.
In practice o/p caps are un-affected, you need to go higher freq in CCM to get the same power thru-put as you would for a lower freq DCM design.

Higher loop gain but a right hand plane zero - in practice DCM is easier and faster re control loops, using simple peak current mode.

A careful study of CCM vs DCM will show you that for the same freq and primary inductance CCM has higher peak currents than DCM as the TX energy is not fully transferred every cycle for CCM. To overcome this you must go to a lower Tx primary L and a higher operating freq, the hard switching involved in commutating the o/p diode(s) creates 10-100 times more RFI than for DCM, the pri mosfet has much higher turn on losses due to this effect, and the common mode noise (VHF) on the o/p and i/p is also much higher. Regards, Orson Cart.

 

[mtwieg]

Lower RMS currents only yeild higher efficiency if they more than offset the higher switching losses (they don't)

Agreed.

In practice cross regulation if just as good in fully DCM designs (it is really determined by coupling)

Hmm, yeah I can't see how cross regulation would be affected by DCM vs CCM.

TRue - lower peak currents allow smaller inputs caps (but not o/p) but not less radiated EMI due to the increased switching noise of the now harder switching in CCM.

Right, in CCM you will have two discontinuous current transitions per switching cycle, vs just one for DCM (though the peak of the DCM may be much higher). I think overall the radiated power will be higher in CCM, but will be at a higher frequency (might be preferable in some situations).

Higher loop gain but a right hand plane zero - in practice DCM is easier and faster re control loops, using simple peak current mode.

Correct.

A careful study of CCM vs DCM will show you that for the same freq and primary inductance CCM has higher peak currents than DCM as the TX energy is not fully transferred every cycle for CCM.

This I'm not sure about. Yes, you don't transfer all your energy every cycle, but that doesn't mean you need higher peak currents at the same power level, I think.

 

[eem2am]

Thanks to mtwieg and Orson for clarification.

Yes I've always considered it that for the same power level and switching frequency, the CCM trafo would have higher primary inductance.

Though i am wondering what is the use of CCM for flybacks.....because from our damning of CCM, it is wondered why anyone would ever use a CCM flyback.

...But ......i can think of one use for CCM flyback..........

in Wide mains flybacks (85-265VAC), you could not realistically have DCM for the whole range........you would need it to be CCM at the lower line, and let it go into DCM at the higher line.

I can think of no other use for CCM flybacks, based on the criticism which we have heaped upon it here.

 

[Orson Cart]

Yes, giong to CCM makes a lot of headaches for EMC compliance, it is possible to make a fully DCM flyback for 85-275VAC mains input and a careful reading of the PI app notes etc will reveal that this is what they aim for and also what their Tx design software does, Regards, Orson Cart.

 

[mtwieg]

Seems like sticking to DCM would be easy if you used constant off time modulation (since frequency decreases with increasing load).

As far as I understand though, CCM has a lot of benefits for other topologies, especially at high power levels (especially buck and buck boost).