Offline SMPS controllers with FET and controller on same chip

[treez]

The Range of Power Supply controllers by Power Integrations is truly ground-breaking. Some of their controllers offer features which no other semi company makes, such as eg the Innoswitch which offers offline flyback with non opto coupler regulation.
Do you know of any controllers which compete with any of the Power Integrations controllers? I don’t think that they even have any obvious competition?

 

[Velkarn]

FSFA2100 from Fairchild, Viper from ST, UCC28910 from TI, etc. there is more ics
i did not see something like Innoswitch, but i think this ic is too expensive for its field of application

- - - Updated - - -

https://www.digikey.com/catalog/en/partgroup/ac-dc-converters-offline-switchers/7474

- - - Updated - - -

for example FSL137H with current mode control

 

[treez]

The problem with the Fairchild monolithic offline controllers is that most of them have an internal fet with a VDS of 650V, which is less than the Power Integrations ones which are 725V rated.
Also, the Power integrations monolithic offline flyback controllers come with adjustable peak current setting. –Whereas the Fairchild ones have a fixed peak current, and this means that the transformer tends to be bigger for the Fairchild ones, because the transformer has to be able to manage the peak current.

Also, a few of the Fairchild ones have 1000V or 800v fets in them , but again these have fixed peak current limits in them which is very inconvenient. This is because as a power supply is being designed, typically the required maximum load is being repeatedly changed or adjusted. With Power Integrations flyback chips, that’s no problem, -because they have a whole family of chips that can do the job, and so you can just change to a different chip without having to vary the PCB much, whereas with Fairchild ones, they are more made for one specific power level, and certainly not the variable range of power that the families of Power Integrations chips facilitate.

Also, the power integrations chips are voltage mode chips, which means they tend to give off less audible noise in lighter loading, whereas the Fairchild ones are current mode, and one can get pulse bunching in lighter loads which can lead to audible noise.

The VIPER family is ok, but has a greater operating current draw than equivalent power integrations chips. Viper also goes into burst mode at light load so its not that good with a bias winding, as bias windings prefer regular pulses, rather than bursts interspersed with nothing.

UCC28910 again has a fixed peak current level, which of course may well not be what the design needs, and especially f if the design spec changes, as so often it does.
Also, UCC28910 has only a 700v rated fet, whereas topswitch is 725v rated.

The topswitch-jx device has a totally variable maximum fet current level, which is settable with a resistor

- - - Updated - - -

Another point about the topswitch is that it uses the opto-transistor connected in common collector configuration, and there is a large capacitor connected to its emitter..therefore, the internal pole due to the high capacitance of the base-collector region of the opto-transistor is irrelevant in the design, so if you get a different batch of opto’s with much different base-collector capacitance, then it wont matter and wont bring about instability. This is a great thing about the way the topswitch has its opto connected.
The fiarchild ones have the opto connected in common-emitter mode, so this advantage is not seen, as for them, the base-collector capacitance is very significant.

 

[schmitt trigger]

Power Integrations is AFAIK, the innovator on monolithic integrated (Switch + Controller) devices. They have been around since the late 80s. They are way ahead of their competitors, since this is essentially their core business, whereas to TI, Fairchild or STMicro they are only a niche product nested within one of their many other business units.

Prior to the TOPswitch, there were hybrid devices, meaning two separate die for the switch and controller. Sometimes other discrete components were incorporated into the package.
Sanken (Japan) was a big player on this technique, rendered obsolete (at least for powers below 100 watts).

 

[treez]

The following explains how the superb topswitch has the great feature that its overload allowance time can be altered by external circuitry. Do you agree with the following?…(“overload allowance time” is the amount of time that an overload can be present before protective measures are taken by the chip).
This feature is not available with the FSL128 competitor device for example.

In a Topswitch design, overload protection is achieved because in overload, the TL431 pulls little current through the opto, and in fact the TL431 output transistor goes totally high impedance such that no current at all is drawn through the opto, and therefore, the topswitch receives no bias current, because in the topswitch, the bias current is in fact supplied via the optocoupler. The overload allowance time is governed by the size of the capacitor on the topswitch VC pin. The overload allowance time cannot though be made arbitrarily large by using a massive Vc pin capacitor, because then that would create a very low frequency pole in the compensator, which would mean a very slow transient response.

The good point about the topswitch is that the overload protection can be bypassed by keeping the vc pin capacitor charged up above 4.8V. In any other offline pwm controller, such as the Fairchild ones, the internal overload protection cannot be bypassed, because the overload allowance time is determined by a timer inside the chip to which one has no access.

This is a great advantage of the topswitch…do you agree?

 

[treez]

Another advantage of the topswitch flyback control chips is precisely that the operating current for the topswitch comes via the optocoupler….this means that the opto-diode current has to be relatively large, and gets you out of the low current region suffered by the opto’s when used with other controllers…..as you know, when you have sub milliamp currents in optocoupler diodes, the CTR becomes ridiculously widely toleranced. –This problem will not afflict you with the topswitch.

 

[treez]

i did not see something like Innoswitch, but i think this ic is too expensive for its field of application

Innoswitch is one of the cheapest controllers on the market, just 65 cents for 1000 pieces on digikey...
https://www.digikey.com/product-detail/en/power-integrations/INN2003K-TL/INN2003K-TL-ND/5048460

...this is truly amazing when you consider what the innoswitch does

 

[Easy peasy]

There are other i.c.'s out there that do not need opto's for output volt sensing, they look at the flyback volts (via the sense or Vcc wdg) after the leakage transient has subsided.

Did you know power integrations started out making 800V half bridge driver i.c.'s ? I think this was their first chip

 

[treez]

Thanks yes, but as you know, due to the volt drop across the resistance of the main secondary windings, and the temp dependent volt drop across the main secondary diode, those bias winding regulators are not as accurate as eg the innoswitch, even if they do wait for the leakage ring to subside.
Also, as you know, the transient response of the bias winding regulators is not as good as the innoswitch. The innoswitch regulates the main secondary almost as if it were a non isolated dcdc design…excellent transient response.

Interesting to hear that they did half bridge driver IC’s….they now seem a bit unfashionable for an innovator like power integrations….though I appreciate that in the older days when someone wanted a half bridge and was using the split caps of a 120VAC bridge rectifier with a voltage doubler link that the half bridge was well popular then.

It must be obselete now, I also know that they used to do a combo chip with a pfc and LLC controller on the same semi….but I suppose it was all a bit too much to layout so they must have scrapped it.

 

[Easy peasy]

The innno-switch is on-off control, so not the smooth sort of o/p you would get from continuously variable PWM, OK for charging and miscellaneous power supplies, their flux link isolation is to be commended, but much easier to do for simple on-off control of the power stage...

 

[treez]

Yes, though I think the Innoswitch would be good for Class D amplifier supplies up to 20W. Eg radios, small guitar amplifiers etc.
This is because as you know, the Class D amp draws power in irregular pulses , and is never a smooth load, so it is pointless having an SMPS with a smooth PWM as the load is not smooth.
The innoswitch would do this without being too “noisy” because of the peak current reduction feature at lighter loading.