Topswitch flyback controllers: Why not brag about short circuit protection?

The Topswitch-jx range of flyback controllers are the best in the world. Why do they not boast about their unique superb output short circuit protection feature?

In a topswitch-jx, the actual operating current to supply the topswitch actually comes from the feedback optocoupler.
This unique feature means that if the output gets short circuited, then the topswitch-jx gets no operating current, which is exactly what is wanted when the output is shorted.

With alternative (competitor) devices, the chip is supplied from the bais winding, and the bias winding may well not collapse during output short circuit (-due to the leakage inductance). Therefore, a competitor device may keep supplying power into an output short circuit for a longer time.

This is a superb advantage of the topswitch-jx, so why does the topswitch-jx datasheet not boast about this feature?

Another point is that the control current of the topswitch-jx, which comes from the opto, ranges from 2.8mA to 4.8mA. –And that corresponds to duty cycle variation from 78% to zero percent. That is a good magnitude of control current, which means that the topswitch-jx will be less susceptible to noise issues.
In comparison, the alternative Fairchild part, FSGM0565R, has a maximum control current of 210uA. –So obviously the FSGM0565R will be more susceptble to noise issues.
Again, why does the topswitch datasheet not boast about this advantage of having such a large control current, which acts to reduce noise issues.

A few days ago i was looking at the HX series, i see the JX replaces it. 90 watts less output power and new output good for 725 vdc from 700 vdc in HX. From the modded specs i bet these chips were having problems.

actually, the short cct protection is not as good as you think, given that a flyback is a constant power device, you can get 3 x I out for 1/3 Vout, however if the TOPswitch looses regulation for a while it will switch off which is usually an eventual save...

Thanks,
I guess what I am saying is that topswitch is the greatest, due to if nothing else its (unique) settable peak FET current..as well as lots of other great features..

So why does the topswitch run so much control current?, and why make the device operating current flow in the opto?…..this all means that you end up with up to 14mA flowing in the optocoupler secondary (with old opto’s at high temperature & with degraded CTR). –If the flyback output was say 48V, then you’d have 48V*0.014A = 672mW of power dissipation due to the feedback optocoupler current.
Why do it like this? Why have so much dissipation caused by the high optocoupler current?
Why not have a separate Vcc pin on the topswitch, and connect it to the bias winding? And why not change the topswitch so that you have less control current flowing?
There are obviously good reasons for these things, as we know the topswitch is a superb controller and the world’s best.

Topswitch-jx datasheet:
https://www.power.com/sites/default/files/product-docs/topjx_family_datasheet.pdf

FSGM0565R datasheet:
https://www.fairchildsemi.com/datasheets/FS/FSGM0565R.pdf

the control current magnitude is for noise reasons, for 48V out, then a pri side bias winding is needed to power this current thru the opto xtor, the main weakness of the topswitch is the high Rds on of the devices, requiring very effective heatsinking

the control current magnitude is for noise reasons, for 48V out, then a pri side bias winding is needed to power this current thru the opto xtor

Click to expand...

yes I agree a pri side bias winding is needed, -the thing is, you then end up with a relatively high current in the opto diode, and if your vout is 48v, then you will end up with (48V * [control current]) Watts being dissipated due to the control current in the secondary side.
-In fact, its worse than that, because added to the control current is the topswitch device operating current, which actually comes from the opto, since topswitch control pin and Vcc pin are one and the same pin.

the control current magnitude is for noise reasons

Click to expand...

…Thanks, that’s what I kind of suspected, the thing is, the FSGM0565R flyback controller has a control current of just 210uA maximum. –This is far less than the topswitch control current of 4.8mA maximum. If the topswitch needs such high control current to avoid noise issues, then how does the FSGM0565R controller get away with just 210uA of control current?

FSGM0565R flyback controller datasheet:
https://www.fairchildsemi.com/datasheets/FS/FSGM0565R.pdf

the main weakness of the topswitch is the high Rds on of the devices, requiring very effective heatsinking

Click to expand...

..well, sorry to be such a datasheet spotter, but topswitch actually has less rds(on) than any of the monolithic flyback competitors, as page 28 of topswitch-jx datasheet shows..(I know what you mean though, you are comparing it with flyback controllers that use an external FET, and the external FET will likely have lower rds(on) than the topswitch-jx internal FET, if one spends enough money on it)

Topswitch datasheet:
https://ac-dc.power.com/sites/default/files/product-docs/topswitch-jx_family_datasheet.pdf

- - - Updated - - -

In truth though, as you know, the main FET loss factor for an offline flyback (of power level up to 60w say) is the switching loss, so I reckon the rds(on) loss is pretty insignificant for a flyback FET. If you want an external FET with significantly less rds(on) than a topswitch, then you really do have to spend decent money, which i'd say just isn't worth it.

the top-switch often has to work nearby other switchers, how do you know the FSGM0565 isn't susceptible to other noise sources? (other than its own), even the top switch can be susceptible to nearby switchers as we have discovered...

also there is no need to draw control current off the 48V, usually a sec side bias wdg provides power for control and more importantly keeps control power under short cct conditions

how do you know the FSGM0565 isn't susceptible to other noise sources?

Click to expand...

Surely Fairchild wouldn’t have made such a mistake as making a flyback controller that was liable to be susceptible to noise issues?

Also, the onsemi range of flyback controllers (NCP12XX) have a maximum optocoupler current of 3.2V/20k = 160uA. This again is far below the 4.8mA control current of the topswitch-jx. Are we saying that the onsemi NCP12XX range of flyback controllers have also got it wrong and are susceptible to noise?

Onsemi NCP12XX range of pwm controllers:
https://www.onsemi.com/pub_link/Collateral/NCP1217-D.PDF

also there is no need to draw control current off the 48V, usually a sec side bias wdg provides power for control and more importantly keeps control power under short cct conditions

Click to expand...

..So you are suggesting using an extra secondary winding in order to provide for the control current?…that means a more complicated and expensive flyback transformer.
For the case of short circuit on the main secondary output winding, there isn’t always a need for an extra secondary winding to handle that situation, -at least not with the topswitch-jx, this is because with the topswitch-jx, an output short circuit will be handled by the topswitch-jx output short circuit protection feature, and this can operate without needing an extra secondary side winding. (as it an with most other offline SMPS controllers, and I am sure you know this)

(I think I know what you mean, you want some bias rail on the secondary for whatever circuitry may be there…ok, but with a cheap flyback supply, as you know, very often you don’t need such a feature, you just need for the controller to go into auto-restart mode or latch-off mode when the main secondary is shorted)

Surely Fairchild wouldn’t have made such a mistake as making a flyback controller that was liable to be susceptible to noise issues?

Click to expand...

I am quite entertained by the above, until they have built it, how could they possibly know how susceptible it is? Take the FAN9612, it is extremely susceptible to external sw noise from other converters, a bad outcome for a boost IC, also there are plenty of top&tiny switch variants with lower control currents - just don't put them near other switchers...! as you can imagine if you are running 160uA control current the possibility of interference is higher, esp if the layout is bad...

- - - Updated - - -

p.s. a darlington connected opto removes the need for high opto-led currents n the o/p side (48V)

as you can imagine if you are running 160uA control current the possibility of interference is higher, esp if the layout is bad...

Click to expand...

Thanks..i am inclined to agree with you on that, but with that being the case, what I can’t understand is why is the topswitch-jx datasheet is not boasting from the rooftops that they have higher control current and thus a better chance of noise-free operation? (even if only for the irregular power supply designer who might not do ‘perfect’ layout)

p.s. a darlington connected opto removes the need for high opto-led currents n the o/p side (48V)

Click to expand...

yes indeed, and you’ve hit the nail on the head there…as this thread shows..

Darlington conncted optocoupler:
https://www.edaboard.com/threads/351617/

The problem with the Darlington connected optocoupler is that the BJT has a gain up to 300. This then really does mean that one is running ridiculously low current in the secondary side opto circuit, and may get noise issues because of that (?)

Also, another problem is that the BJT of the Darlington-connected-opto has a very wide tolerance on its gain (hfe). This is a bit of a problem because the topswitch-jx is voltage mode, and whereas a reduction in gain in the feedback loop (due to a low tolerance hfe BJT) is usually fine for current mode flybacks, it can often bring instability in voltage mode flybacks….all remedy-able by proper feedback loop calculation, but we are trying to evaluate topswitch from the point of view of a ‘novice’ who might fall short in that area. So the wide tolerance on gain in the feedback loop caused by using the Darlington could be a problem.

- - - Updated - - -

as you can imagine if you are running 160uA control current the possibility of interference is higher, esp if the layout is bad...

Click to expand...

As you know, the reason these semi co’s have so little control current is because they want to meet the 0.5W (now worldwide) standby regulation for offline power supplies. This proves my point that sub 0.5W rule is bad news, and will just result in lots of dodgy switchers hitting the market.

I am actually thinking of putting a 5V SMPS on the output of the 48Vout flyback. The 5V smps will supply the feedback opto so that there is less dissipation due to the control current than if the 48V were supplying it direct.
Eg a LM5019 monolithic buck could be used…
https://www.ti.com/lit/ds/symlink/lm5019.pdf

This way, the dissipation of the high topswitch control current can be reduced as well as keeping the control current high for noise mitigation purposes..you agree this is a good idea.?

p.s. a darlington connected opto removes the need for high opto-led currents n the o/p side (48V)

Click to expand...

..yes, you mean like in fig31, page 25 of AN-47 (link below).
There are problems surrounding such use of a darlington BJT, and I believe that the Darlington connection should not be used…for these reasons…

Figure 5, page 4 of the attached (“Transient response and loop gains of power supplies”) shows that voltage mode control loops need a lot more care than current mode control loops. Figure 5 shows the open loop response of a voltage mode control loop, and shows that due to the severe dip in phase which can occur in a voltage mode converter, it is possible that either an increase OR a decrease in gain can make the power supply go unstable. –This can also happen in current mode supplies, but is not so likely because the current mode supplies lack the severe phase drop due to the double flter pole of the power stage in voltage mode control. Therefore, current mode power supplies can very often be made stable simply by reducing their feedback loop gain.

Of course, either current mode or voltage mode can be made stable with proper compensation. However, the problem comes when you are using elements in the feedback loops such as BJT’s, which have ridiculously widely toleranced gain values, and thus make compensating a feedback loop containing them more hit and miss.
Fig31 of page 25 of the following (AN-47) shows an SMPS feedback loop with a BJT in it……

AN-47 document:
https://ac-dc.power.com/sites/default/files/product-docs/an47.pdf

The point about topswitch is that it is good for beginners, but I would not like to subject a beginner to compensating a voltage mode feedback loop with a widely variant gain element in it.
Therefore, I would suggest if you are using topswitch and have a high vout in your flyback smps, then simply use a buck smps on the flyback secondary to produce a 5V bia rail which you can use to drive your feedback optocoupler from. Then you will not need to use the BJT in the feedback loop. Do you agree that this is good advice?
(LM5019 is a good example of a monolithic buck converter for this purpose).

- - - Updated - - -

the control current magnitude is for noise reasons

Click to expand...

Interesting that the UCC28600 controller has just 250uA maximum of control current, virtually all offline flyback controllers have such low control current, I would be surprised if there were significant noise issues with this low current when so many are using it that low.
UCC28600 controller datasheet:
https://www.ti.com/lit/ds/symlink/ucc28600.pdf

The noise issues often only arise when ANOTHER switcher is nearby...

- - - Updated - - -

using a darlington with degeneration resistors usually overcomes the gain spread issues, also you can use an pnp-npn darlington, where the npn is the opto xtor, and then for only a slight gain increase you can invert the C-E on the pnp to get a low gain device ~ 10, as long as the applied volts to the pnp are below ~6V

using a darlington with degeneration resistors usually overcomes the gain spread issues

Click to expand...

thanks but with a voltage mode feedback loop, even with degeneration resistors, the spread of hfe is too high to merit it.
And we cant say, even with degeneration resistors, what the gain will be.

Certainly degeneration resistors determine the maximum gain, regardless of the h-fe of the xtor, this is usually sufficient for a commercial design, as to the short circuit performance of the Top-switch there is a slight delay between applied short and the time taken for the control current to fall to the level required for the device to stop, a few mS, usually the o/p diodes can handle this, and the pk I limit on the top switch internals usually protect the Top-switch itself, so, by and large the Top-switch does manage to protect itself - a good thing..!