Voltage mode CCM flyback cannot be done with type 2 compensator?

The TOPswitch flyback controller is a voltage mode flyback controller by Power integrations. It is the single most popular offline flyback controller in the world in terms of number of different SMPS’s that use it. It is a superb part, so why does app note 57 by Power integrations (attached) claim that it can be compensated by a “type 2” compensator? (A voltage mode flyback needs a type 3 compensator –surely?)

AN57 by power integrations:
https://ac-dc.power.com/sites/default/files/product-docs/an57.pdf
(top left of page 15 speaks of “type 2”, plus elsewhere in article)

it is current mode internally - so "type II" may be ok

Page 3 of the datasheet for topswitch shows the controller block diagram, and its extremely difficult to tell from that whether its voltage mode or current mode control.

However, Page 5 of the topswitch datasheet states that there is an internal ramp produced, so that looks like voltage mode control to me..

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

Also, top of first page of AN57 states that topswitch uses voltage mode control.....
https://ac-dc.power.com/sites/default/files/product-docs/an57.pdf


Also, I have definitely seen power integrations literature explaining that because topswitch is voltage mode controlled, that it lacks the advantage of current mode control in counteracting changes in the input voltage….but then goes on to say that they get round this by having a “feedforward” feature. –So topswitch must be voltage mode controlled, otherwise they wouldn’t need the feedforward feature.

I had the following discussion with power integrations several years ago asking why voltage mode control was chosen for topswitch’s, and they didn’t refute it…so I presume from that that it is voltage mode control, or else they would have said so I guess…
https://www.power.com/forum/low-power-design/topswitch-flybacks-are-voltage-mode/

I was just wondering, the topswitch controllers have "feedforward", and I wondered if this means that a true , full "type 3" controller is not therefore needed for them?

Nope, feed-forward of input volts is an old technique to reduce the 120Hz ripple on the o/p, this is automatic in a peak current mode controller, type 3 is best for volt mode as you need to counteract the inherent pole formed by the transformer and the o/p capacitors in volt mode...

type 3 is best for volt mode as you need to counteract the inherent pole formed by the transformer and the o/p capacitors in volt mode...

Click to expand...

I agree, but I would go further than you and say that not only is type 3 better, but it is essential for a voltage mode flyback in ccm, -type 2 just isn’t good enough here…so I am wondering why there app note “AN57” (in top post) says that a type 2 network is ok for their voltage mode flyback?…..they reckon they can get 1khz crossover frequency with the type 2 network...this surely isn't possible.?

it is current mode internally - so "type II" may be ok

Click to expand...

Are you speaking of the topswitch-jx's internal primary current limt reduction when you say that, as per the following?...

The Power integrations topswitch operate with a kind of “adjusted voltage mode”, do you agree?
If you look at fig 6 , page 4 of the topswitch-jx datasheet (below), then you see that as the control pin current (the error current), increases, the peak current level decreases, aswell as the duty cycle decreasing.
Such a decrease in peak primary current does not occur in “pure voltage mode” converters. Therefore, this is why the topswitch-jx is an “adjusted voltage mode” converter. Do you agree with this?

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

….the clever use of this “adjusted voltage mode” control by the topswitch-jx means that a type 2 compensator can be used to stabilise a flyback in ccm that uses topswitch-jx.
AN-47 by power integrations discusses how a type 2 compensator can be used with a power integrations topswitch-x.

AN-47 App Note on topswitch:
https://ac-dc.power.com/sites/default/files/product-docs/an47.pdf

We know that topswitch-jx is of superb quality, because Dr Christophe Basso, probably amongst the world’s best SMPS writers, has set aside a whole chapter of his latest book “Designing control loops for linear and switching power supplies” “ to the topswitch-jx feedback loop.

Power integration always does a poor job of explaining the operation of their controllers. I think they assume engineers are just going to start with a reference design and iterate from there, rather than derive an optimal design from scratch.

The part has peak current detection, but at fixed thresholds, whereas CMC uses a variable peak current setpoint. If it were to use peak current limiting with a variable frequency in DCM, then that would be similar to CMC. Otherwise it's not.

The datasheet states:

Although different modes operate differently to make transitions between modes smooth, the simple relationship between duty cycle and excess CONTROL pin current shown in Figure 6 is maintained through all three PWM modes

Click to expand...

This directly implies that the chip only operates in voltage mode control (where the duty cycle, not the peak current, is directly determined by control signal). Though throughout its load range, the actual method of modulating D changes from constant frequency to variable frequency and back to fixed frequency, and it's anybody's guess what the small signal transfer function is really doing for each mode.

When operating in the variable frequency mode the controller sets the peak drain current, so it could be argued that it is effectively operating at a CMC, if it's DCM. But if it's in DCM, then the distinction between current mode and voltage mode control becomes moot, since they both have a single-pole response.

- - - Updated - - -

treez said:

I was just wondering, the topswitch controllers have "feedforward", and I wondered if this means that a true , full "type 3" controller is not therefore needed for them?

Click to expand...

Feedforward to me refers to using the sensed input voltage to correct the duty cycle of a PWM modulator such that its average output is independent of the input voltage. See the UCC35705 datasheet for an example. This is very useful for voltage mode control loops, because otherwise the open loop gain would be proportional to input voltage. Also useful for making the loop response more independent of the source impedance.

When operating in the variable frequency mode the controller sets the peak drain current, so it could be argued that it is effectively operating at a CMC, if it's DCM. But if it's in DCM, then the distinction between current mode and voltage mode control becomes moot, since they both have a single-pole response.

Click to expand...

..sorry but I don’t agree it’s a moot point whether its in CMC or VMC for DCM flyback….

Posts #4 and #5 of the following…

https://www.edaboard.com/threads/351408/

..confirm that there is a big difference between DCM current mode flyback and DCM voltage mode flyback.

When operating in the variable frequency mode the controller sets the peak drain current, so it could be argued that it is effectively operating at a CMC, if it's DCM

Click to expand...

…again I see that like myself , you are trying to understand whether this controller is in CMC or VMC. I am becoming convinced that the topswitch-jx NEVER operates in CMC because for CMC, you MUST have a ramp signal going into your PWM comparator which comes from the primary current sense resistor. That never happens with topswitch-jx so from that viewpoint, topswitch-jx is NEVER in CMC.

However, having made that bold statement, I am then stumped by the fact that AN-57 (see top post for AN-57) by Power integrations states that a CMC Flyback with topswitch-jx, can be feedback compensated with a type 2 compensator to give crossover frequency around 1khz…….this just isn’t possible with a voltage mode controller, so I am wondering in what mode the topswitch-jx operates in?

I think Power integrations have hit on some superb new control method in topswitch-jx but are keeping it to themselves, as they are milking the huge sales of this excellent control chip, which indeed, is the most commonly used one in the world.

What is the method of control in topswitch-jx?

- - - Updated - - -

If you look at fig 6 , page 4 of the topswitch-jx datasheet (below), then you see that as the control pin current (the error current), increases, the peak current level decreases, aswell as the duty cycle decreasing.
What possibly could be the reason to vary the peak current limit?, what advantage does it give?…it is something special, as the topswitch-jx is the world’s best seller.

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

- - - Updated - - -

The topswitch-jx App Note called AN-57, (link below) on page 12, near the top of page, states that for a topswitch-jx flyback, it is useful to have a flyback transformer secondary winding resistance, which is significant enough to provide sufficient damping of the LC resonance between the flyback secondary and the output capacitor.

(As you know, such damping reduces the Q factor of this power stage LC resonance.
The App Note AN-57 states that a Q factor of 0.1 to 0.3 is good for topswitch-jx flybacks. As you know, if this LC resonance is lightly damped, then the phase falls away very steeply)

Why does Power Integrations want to specify this damping of the flyback power stage LC resonance?, -are they saying that if too lightly damped, then there will be problems? Surely, with a type 3 compensator, the gain peaking of the power stage LC resonance is taken care of?

After all, one would not wish to have much resistance in the secondary of a flyback transformer, as it would only lead to greater winding losses.

Another point, is that the App Note AN-57 of power integrations , on page 4 (bottom left of page), states that the flyback power stage LC circuit should have a resonant frequency of greater than 500Hz. Why is this? There is something special going on as topswitch-jx is the single most successful pwm controller ever made in the world, so what is the reason for all these things?, and is it VMC?, or CMC?

App Note AN-57 of power integrations.
https://ac-dc.power.com/sites/default/files/product-docs/an57.pdf

treez said:

https://www.edaboard.com/threads/351408/

..confirm that there is a big difference between DCM current mode flyback and DCM voltage mode flyback.

Click to expand...

No it doesn't. Both VMC and CMC can have the exact same transfer function at a given operating point in DCM. The devil is in the details, which are not given in those examples. There's no way to make sweeping generalizations based on the waveforms in those documents.

However, having made that bold statement, I am then stumped by the fact that AN-57 (see top post for AN-57) by Power integrations states that a CMC Flyback with topswitch-jx, can be feedback compensated with a type 2 compensator to give crossover frequency around 1khz…….this just isn’t possible with a voltage mode controller, so I am wondering in what mode the topswitch-jx operates in?

Click to expand...

I don't see where it categorically states that a CCM flyback can be controlled with a type II compensator. Perhaps it assumes that most designs will be DCM over their entire operating range, which is typical for low power flybacks.

I think Power integrations have hit on some superb new control method in topswitch-jx but are keeping it to themselves, as they are milking the huge sales of this excellent control chip, which indeed, is the most commonly used one in the world.

Click to expand...

I highly doubt the IC does anything to manipulate the dynamical response of the circuit (that is, the number or positions of the various poles and zeros). What it does do is vary the method of duty cycle control (which, in a small signal sense, should not have any effects on the dynamical behavior).

If you look at fig 6 , page 4 of the topswitch-jx datasheet (below), then you see that as the control pin current (the error current), increases, the peak current level decreases, aswell as the duty cycle decreasing.
What possibly could be the reason to vary the peak current limit?, what advantage does it give?…it is something special, as the topswitch-jx is the world’s best seller.

Click to expand...

Yes, their charts are confusing, because the documentation is poorly written.

The text in the left column of page 6 of the datasheet explains things reasonably. The middle plot of fig 6 does cause confusion, if you assume that it means that the controller is actually setting variable peak current limits in the sloped regions (which is a reasonable interpretation). But I believe this isn't what's actually happening. What happens is that at the end of every switching period, just before the FET turns off, the controller samples the outputs of the two peak current comparators Kps(upper) and Kps(lower). Depending on that, it will determine what control method to use next (variable frequency or variable duty cycle). It can only control peak current to one of those levels. Otherwise it's controlling duty cycle. When in a fixed frequency operating mode, the peak current will vary as D is changed, which gives the sloped regions of that middle plot. But the controller is directly controlling D, not peak current.

It can only control peak current to one of those levels.

Click to expand...

thanks, so do you mean in the variable frequency region, the peak current is set to the level of either kps(upper) or kps(lower)?

Sorry I know you are not a Support Engineer for Power Integrations, but please may I ask...

Also, looking at fig 6 page 4, the top graph, it looks as if you only get the maximum peak current setting if you have 78% duty cycle........but as you know, most offline flybacks often have a max duty cycle (at max load) of just around 0.2......so does that mean that they never get out of variable frequency mode?

What do the fig 6 graphs look like if you have set it up (NS/NP) for a max duty cycle of 0.2, say?

- - - Updated - - -

Both VMC and CMC can have the exact same transfer function at a given operating point in DCM.

Click to expand...

Yes that’s true but its not my point, my point is , that a DCM current mode flyback is always intrinsically stable, whereas a DCM voltage mode flyback can be unstable.

The following literature exerpt shows that a flyback that is stable in CCM current mode, will always be stable when it subsequently goes into DCM at lighter loading…..this is not the case for voltage mode flybacks.
Pg 10, (halfway down on Right hand side) confirms that for a current mode flyback...if you get it stable in max. load, min vin, then it will be stable at all other conditions of line and load, including when it goes into DCM...
https://www.fairchildsemi.com/applic...AN/AN-4137.pdf

Yes a DCM volt mode can be unstable, due to the inherent LC filter, formed by the Tx and the o/p caps (large), current mode operation removes the effect of the L, making closing the loop much easier, esp at light / no load.

Yes a DCM volt mode can be unstable, due to the inherent LC filter, formed by the Tx and the o/p caps (large), current mode operation removes the effect of the L, making closing the loop much easier, esp at light / no load.

Click to expand...

Thanks, and of course, the next question is, why is the topswitch-jx flyback controller using voltage mode control?, .....when current mode is easier to stabilise for a flyback in DCM

- - - Updated - - -

I highly doubt the IC does anything to manipulate the dynamical response of the circuit (that is, the number or positions of the various poles and zeros).

Click to expand...

I am not sure I agree with this, clearly, from the middle graph of fig6, page4 of the topswitch-jx datasheet, there is a variable frequency mode which occurs at medium loading, say from 20% to 55% loading….this is definitely not voltage mode control or current mode control, as for this interval, the peak primary current level is fixed and its just the frequency that varies. –So this surely must affect the controller dynamics?.
TOPswitch-jx datasheet:
https://ac-dc.power.com/sites/default/files/product-docs/topswitch-jx_family_datasheet.pdf

treez said:

thanks, so do you mean in the variable frequency region, the peak current is set to the level of either kps(upper) or kps(lower)?

Click to expand...

From the diagram, in the "middle" variable frequency mode peak current is set to 55% of set maximum. In the multi-cycle modulation mode it will be 25%.

What do the fig 6 graphs look like if you have set it up (NS/NP) for a max duty cycle of 0.2, say?

Click to expand...

The plot inherent to the converter itself, I don't think it can be changed. Depending on your design, you will make use of differing parts of the curves.

Yes that’s true but its not my point, my point is , that a DCM current mode flyback is always intrinsically stable, whereas a DCM voltage mode flyback can be unstable.

Click to expand...

Either can be stable or unstable. It all depends on the compensation. The point is that in DCM there is no inherent advantage to CCM over DCM, as far as I can tell.

The following literature exerpt shows that a flyback that is stable in CCM current mode, will always be stable when it subsequently goes into DCM at lighter loading…..

Click to expand...

Okay, that sounds believable.

this is not the case for voltage mode flybacks.

Click to expand...

I don't see this claim anywhere in the documentation.

I don't see this claim anywhere in the documentation.

Click to expand...

this claim is shown in the documentation..AN-57...

App Note AN57 on fig. 13, page 9 shows that a voltage mode flyback that is stable in CCM, can go unstable in DCM.
(the plot shows vout when load is stepped from high load in CCM, to lighter load where its in DCM)..

AN57:
**broken link removed**

I definitely have read in either a book or app note that the current mode dcm flyback is intrinsically stable. I will try and find the extract.

- - - Updated - - -

The plot inherent to the converter itself, I don't think it can be changed. Depending on your design, you will make use of differing parts of the curves.

Click to expand...

..thanks, but that's terrible news, as it means that the only way you can get full frequency PWM for the flyback at max load is to design the turns ratio so that the duty cycle is 0.74 at maximum load....and that is generally totally impractical

- - - Updated - - -

As Easy Peasy says in post#12 above, Current mode dcm flyback has one less pole than voltage mode flyback due to the way that current mode kind of circumvents the inductor. So DCM current mode is esier to compensate than dcm voltage mode....
-and having said that, why do the topswitch-jx family, the most popular pwm controller in the world, use voltage mode.?

The answer is, the TOP-switch is not strictly voltage mode... AND in most apps it is designed to be on the DCM/CCM border at full load, low line, at full load the damping is maximum, hence the effect of the RHP zero is reduced...

treez said:

this claim is shown in the documentation..AN-57...

App Note AN57 on fig. 13, page 9 shows that a voltage mode flyback that is stable in CCM, can go unstable in DCM.
(the plot shows vout when load is stepped from high load in CCM, to lighter load where its in DCM)..

Click to expand...

That's a terribly weak argument for your broad claims. A more thorough explanation would include bode plots at each operating point, not just a couple transient load responses. I'm assuming that the DCM oscillation is due to large signal nonlinearity, and is stable in the small signal sense.

- - - Updated - - -

treez said:

..thanks, but that's terrible news, as it means that the only way you can get full frequency PWM for the flyback at max load is to design the turns ratio so that the duty cycle is 0.74 at maximum load....and that is generally totally impractical

Click to expand...

The only downside to high duty cycle is that it means your peak switch voltage will have to be higher to get flux balance. I think this is one of the reason that the topswitch controllers are rated to relatively high switch voltages.

Ok thanks, also, I am still trying to work out the implications of fig6, page 4 of the topswitch-jx datasheet.

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


Fig28 and fig29 (page 41) of the attached DER243 doc show that the control current’s actual value doesn’t determine what mode of operation the topswitch-jx is in. The control current determines the duty cycle. But you can have full frequency pwm mode no matter what is your control current. This is proven by fig28 and fig29, which are both switching at the maximum switching frequency (and therefore must both be in ‘full frequency pwm mode’) even though they have vastly different duty cycles.

Do you agree that the control current for the fig28 and fig29 must be very different, even though they are both in “full frequency pwm mode”?

Fig6, page 4 of the topswitch-jx datasheet seems to suggest that the “full frequency pwm mode” only happens at the highest possible duty cycle of the topswitch-jx

- - - Updated - - -

Well, it all makes sense now, the topswitch-jx uses voltage mode control because for a wide mains flyback, you are likely to need a wide range of duty cycle, and so voltage mode means you don’t need slope compensation which would be awkward on a monolithic controller.
So with topswitch, you can go 60% plus duty cycle at low mains and no slope comp needed….also, the 50% plus duty means you can reduce your diode off-state voltage and use a schottky on the secondary.
Also, the increase in fet voltage that this gives is ok with topswitch, because of the unique 725v internal fet rating.
Topswitch-jx wins hands down, I did wonder why it was using voltage mode control.

Also, topswitch has frequency settable to either 66khz or 132khz, so better than the competitors.
Also, topswitch has a completely adjustable peak fet current limit, which is well tightly toleranced….far better than the competitors in monolithic ac-dc flyback.
Also, the last mystery of the topswitch-jx was wondering why they end up running the operating current, and a relatively high control current through the opto…..but this makes sense, because you don’t want to run sub one milliamp currents in your opto as CTR then gets ridiculously poorly toleranced.
However, if you do want to run lower opto current, you can just put a darlington to the opto transistor.
This all means that topswitch-jx is miles ahead of even the closest competitor, and that doesn’t even mention several other advantages.