What exactly is sliding mode (variable structure) control? !

MrCU204

What exactly is sliding mode (variable structure) control? ! [Copy link]

 

 

The opening is "variable structure", the closing is "sliding surface", it doesn't make any sense no matter how you look at it, will it be a loss to write it in a more user-friendly way? !

Take this extremely common feedback series regulator as an example. The parameters of all components in this circuit are fixed. When Vin or R fluctuates, the only means of suppression is negative feedback.

If Vin or R is artificially changed, is it still countered by negative feedback? Or, is it possible to change a certain component to be adjustable and controllable, such as R2 or R4?

maychang

[If Vin or R is artificially changed, will it still be countered by negative feedback?]

Of course.

maychang

[Or, change a component to be adjustable and controllable, such as R2 or R4]

Changing R2 will not change the output voltage (it will have a very small effect), but changing R4 will change the output voltage.

qwqwqw2088

R2 or R4 can be made adjustable, both are possible, R4 adjustable is more common.
The principle is based on the voltage negative feedback mechanism

qwqwqw2088

For small circuits like this, you can use simulation software to compare the conditions in various situations.

MrCU204

This formula in the figure only reflects how Vout is established, without considering the influence of R and Q2.

Vout is actually the voltage division of R2/β and R. If R decreases, the voltage division of R2 will definitely increase (regardless of whether Q2 is present or not), thus causing Vout to decrease.

The key is that you cannot prevent the voltage divider of R2 from increasing, so the increase in the voltage divider of R2 caused by negative feedback can only be weakened, not completely eliminated.

MrCU204

qwqwqw2088 posted on 2024-10-10 07:49 R2 or R4 can be made adjustable. The most common principle of R4 adjustment is based on the voltage negative feedback mechanism

Assume that there is no power fluctuation and the load is not disturbing.

Then, the only thing that can affect the dynamic balance of this voltage stabilizer is human changes.

If there is no adjustable component, the balance point of negative feedback will shift due to various changes.

If R4 is adjustable, then R4 can be adjusted according to the change so that the feedback coefficient always adapts to the new working conditions.

In this way, negative feedback only needs to monitor normal power supply fluctuations and load disturbances, without having to deal with extremely strong interference or human changes.

qwqwqw2088

This series adjustable voltage regulator is a more traditional series feedback type voltage regulator circuit. The series type voltage regulator circuit essentially maintains the output voltage stable through voltage negative feedback.

qwqwqw2088

When the grid voltage fluctuates or the load current changes, causing the output voltage Uo to increase, the voltage-dividing effect of the
sampling resistor is used. The adjustable function is to adjust the collector-emitter voltage according to the comparison circuit output to achieve the purpose of automatically stabilizing the output voltage.

MrCU204

qwqwqw2088 Published on 2024-10-10 14:45 When the grid voltage fluctuates or the load current changes, causing the output voltage Uo to increase, the voltage divider of the sampling resistor is used to adjust the voltage according to the comparison voltage...

Yes, the picture below is an adjustable voltage regulator. By manually adjusting Rp, you can set Uo.

However, no matter whether you move Rp or not, negative feedback will never stop and will never be absent, even if it is changed to a switching circuit.

Assuming that the Uo I want is 5V (Io is 10A at most) and Ui is 12V, then I adjust Rp, of course, to make the Uo of your circuit 5V, but it is not possible to adjust it by just hanging any load, or must it be based on no-load, half-load, or full-load? !

Assuming that I take 5A load (that is, half load) as the standard, then if the load is lighter than 5A, Uo will be higher than 5V, and if the load is heavier than 5A, Uo will be lower than 5V. This problem can never be solved by negative feedback alone.

So, is it possible to adjust Rp when R is switched on and off? Or, if Ui is changed to 15V or 24V, is it necessary to adjust Rp as well? !

MrCU204

Sliding mode variable structure control seems to be an algorithm, but if you want to use it, you also need a "terminal" to receive it, right? !

Anyway, one thing is certain, sliding mode variable structure control is absolutely different from negative feedback. Negative feedback can be done directly with the circuit in this figure, but if you want to achieve sliding mode variable structure control, do you need to add something? !

Or you may think that sliding mode variable structure control is not applicable to this circuit. Well, why not change the series regulator from linear to switch, and then it will become a buck circuit.

Sliding mode variable structure control is applicable to switching power supplies, right? Then, which part of the "structure" has changed, and which component is "sliding"? Is this a commercial secret that cannot be made public? !

MrCU204

Assume that Vout in the title figure is 5V, and further assume that U4 is 4V and U3 is 1V.

The question is, is it that easy to establish this balance? Don't forget that R2 (short-circuiting the emitter junction of Q2) or R2/β2 is in series with R, and there is also Vin. Simply changing Vin is enough to invalidate the formula in the figure.

G≈Go/(1+Go)F, which is the formula for negative feedback (G is the gain of the active device, NF is negative feedback, and OL is an open loop). What is this 1? In this figure, it is the ΔVout. That means that even if Go is infinite, ΔVout cannot be reduced to zero. In other words, when R or Vin changes (or fluctuates), negative feedback alone will never be enough to keep Vout unchanged.

MrCU204

The function of the error comparator amplifier is the same as Q1 in the title figure, which is a negative feedback module.

Sliding mode variable structure control, or any other scheme, if only the "transfer function" of the error comparison amplifier is changed, it is just like replacing Q1 in the title figure. The inherent defects of negative feedback cannot be cured.

Assuming that the Io of my title circuit is 0 to 10A, if negative feedback is used alone to hold Vout, the operating conditions of the error comparator amplifier will inevitably change significantly. A few years ago, I found a blog post from the Chinese Academy of Sciences, which mentioned that sliding mode variable structure control can reduce the labor intensity of negative feedback.

But now I know one thing very well. Changing the "transfer function" of the error comparator amplifier only changes the amount of stimulation required for the same labor intensity, which is equivalent to the relationship between ΔIc1 and ΔVout in the title figure. However, the problem that negative feedback still has to take the lead and go all out has not been solved. So I wonder if variable structure control requires R4 or R2 to be controllable? !

MrCU204

The formula in the title picture is actually inseparable from the rule of "A＝Ao/(1+Ao)F≈1/F".
One thing is certain, Vz is fixed, and F is also fixed in this figure. However, Vbe1 is not a fixed value and will be affected by R and Vin.
The difference between linear regulation and switching regulation is the use of Q2. Linear is variable resistance, and switching is chopping. It is clear and there is no ambiguity. Even people like me can understand that
this circuit is purely discrete. It is a complete negative feedback regulator with independent operation. However, whether it is a linear circuit or a buck circuit, the voltage regulation principle is only negative feedback. If I want to build a purely discrete sliding mode variable structure control regulator, how many transistors do I need? Is fifteen enough? !

MrCU204

qwqwqw2088 Published on 2024-10-10 14:45 When the grid voltage fluctuates or the load current changes, causing the output voltage Uo to increase, the voltage divider of the sampling resistor is used to adjust the voltage according to the comparison voltage...

The sampling resistor string (R1+Rp+R2) is the feedback chain of this circuit.

The direct effect of this feedback chain is to "bind" Uo to VDz through the DC feedback coefficient F, and the "virtual short" effect of negative feedback is the "rope" that ties Uo.

The sampling ratio is equivalent to the AC feedback coefficient f. Unless it is deliberately designed, f will never be less than F. If R1 is connected in parallel with a capacitor, or R2 is changed to a constant current element, then regardless of the size of F, f is always 1.

If f is less than 1, ΔUo will definitely be much larger than ΔVb2. If f is 1, ΔUo is equal to ΔVb2. This is the best result that negative feedback can achieve. If you want ΔUo to be smaller than ΔVb2 or even zero, you can only introduce additional bias directly from Ui to the base of VT2. This is "feedforward".

But I found that feedforward control requires variables, such as ΔUi, and Uo should not change, so it is impossible to provide variables. Therefore, feedforward will not have the function of correcting the change of R. If you want Uo to be completely unaffected by R, you may have to resort to sliding mode variable structure control.

qwqwqw2088

MrCU204 posted on 2024-10-14 01:35 The sampling resistor string (R1+Rp+R2) is the feedback chain of this circuit. The direct effect of this feedback chain is to make Uo follow...

It's actually very simple, but I feel like you're saying it's a bit profound

dwwzl

It’s a good name for a mechanical structure, isn’t it just a feedback voltage regulator circuit!

qwqwqw2088

For example, the commonly used LM317, LM7805 and other similar three-terminal voltage regulator ICs have internal circuits based on this series feedback voltage regulator circuit, and the internal voltage regulation of this type of linear voltage regulator chip has long been mature.

qwqwqw2088

Besides, the design of linear voltage regulator circuit is mainly based on three-terminal integrated voltage regulator.

There are two types of three-terminal regulators:
1. The output voltage is fixed, that is, a fixed output three-terminal regulator. Common three-terminal regulators include the 78 series (positive power supply) and the 79 series (negative power supply).
2. The output voltage is an adjustable linear regulator, generally called an adjustable output three-terminal regulator. This type is represented by the LM317 (positive output) and LM337 (negative output) series.

qwqwqw2088

The basic principle of these two types is the same, both use series voltage regulator circuits. In linear integrated voltage regulators, because the three-terminal voltage regulator has only three leads, it has fewer peripheral components, is convenient, stable, and low-priced, and its efficiency is slightly lower than that of current switch chips.