# Define Linear Devices & Switch Mode Power Supplies?

Status
Not open for further replies.

#### Suraj143

##### Active Member
*What’s the meaning of LINEAR devices? Why they call like that?

*What is the meaning of switch mode power supply? What’s the difference between normal power supply and a switch mode power supply?

Thanks

#### Papabravo

##### Well-Known Member
Suraj143 said:
*What’s the meaning of LINEAR devices? Why they call like that?

*What is the meaning of switch mode power supply? What’s the difference between normal power supply and a switch mode power supply?

Thanks
In general, the term linear, applies to any device or system where the relationship between the input and the output can be described by the equation of a straight line.
Code:
y = mx + b

where m is the slope of the line
and b is the y-intercept
That's why they call it like that.

#### TheVictim

##### Member
Switching power supplies work in a similar manner to the Phased Locked Loop oscillator in that the actual output is compared with what the output SHOULD be, which generates a reference value known as the Error Voltage. This is feed back into part of the input to correct the problem, until there is no error voltage coming from the comparison circuit. This is why switching power supplies don't vary their voltage and PLL synthesized tuners don't frequency drift.

#### Nigel Goodwin

##### Super Moderator
TheVictim said:
Switching power supplies work in a similar manner to the Phased Locked Loop oscillator in that the actual output is compared with what the output SHOULD be, which generates a reference value known as the Error Voltage. This is feed back into part of the input to correct the problem, until there is no error voltage coming from the comparison circuit. This is why switching power supplies don't vary their voltage and PLL synthesized tuners don't frequency drift.
Sorry, but that's completely untrue! - linear power supplies work in that exact same way, using feedback to correct for any variations in the output.

#### Hero999

##### Banned
The difference is that switching power supplies use a switching stage on the output rather than a transistor that never saturates.

Basically linear regulators have a class A output stage and switching regulators have a class D output stage.

#### Speakerguy

##### Active Member
Actually, the true definition of 'linear' in linear systems is this:

f(x) + f(y) = f(x+y)

Which is quite interesting, because contrary to the completely natural seeming assumption that a line equation would obviously be a linear function, y = mx + b is surprisingly not a linear equation, unless b = 0. Technically, a linear system cannot have a DC offset.

This is, of course, if I remember my linear systems and signals course from college oh so many years ago. I only remember this because of a question essentially based on this on a test that killed 90+ % off the class.

#### Optikon

##### New Member
speakerguy79 said:
Actually, the true definition of 'linear' in linear systems is this:

f(x) + f(y) = f(x+y)

Which is quite interesting, because contrary to the completely natural seeming assumption that a line equation would obviously be a linear function, y = mx + b is surprisingly not a linear equation, unless b = 0. Technically, a linear system cannot have a DC offset.

This is, of course, if I remember my linear systems and signals course from college oh so many years ago. I only remember this because of a question essentially based on this on a test that killed 90+ % off the class.
Have a look:

http://en.wikipedia.org/wiki/Linear_system

The definition does not preclude "offsets." which, makes sense. We all know y=mx + b is linear even for non-zero b.

#### Papabravo

##### Well-Known Member
speakerguy79 said:
Actually, the true definition of 'linear' in linear systems is this:

f(x) + f(y) = f(x+y)

Which is quite interesting, because contrary to the completely natural seeming assumption that a line equation would obviously be a linear function, y = mx + b is surprisingly not a linear equation, unless b = 0. Technically, a linear system cannot have a DC offset.

This is, of course, if I remember my linear systems and signals course from college oh so many years ago. I only remember this because of a question essentially based on this on a test that killed 90+ % off the class.
Must have been a third rate school with third world TA that slipped that wowser by you. Are you unfamiliar with the concept of translation of the coordiante axes as a linear operator?

Last edited:

#### Speakerguy

##### Active Member
Sorry, top ten engineering school in the country. Maybe you guys should go through the math before telling someone they're wrong. Example:

f(x) = 2x + 5

f(2) + f(4) = 9 + 13 = 21

f(2+4) = 17

f(2) + f(4) != f(2+4)

Since f(x) + f(y) must equal f(x+y) for a system to be a linear system, f(x) is not linear (as far as linear system theory goes; it's still a straight line on graph paper though).

This is straight out of the B.P. Lathi textbook that most colleges use.

Last edited:

#### Papabravo

##### Well-Known Member

Your basic definition apllies to linear maps from one vector space to another. In this case adding the constant makes it an affine map. This is way beyond the original poster's question and our answers were given in the context of elementary algebra. I'm pretty sure the OP did not have this higher treatment in mind when he asked his original question.

Last edited:

#### Hero999

##### Banned
Smart arse.

As far as I'm concerned linear = analogue = an infinite number of quantities and switching = digital = two possible states (on and off).

#### Optikon

##### New Member
speakerguy79 said:
Sorry, top ten engineering school in the country. Maybe you guys should go through the math before telling someone they're wrong. Example:

f(x) = 2x + 5

f(2) + f(4) = 9 + 13 = 21

f(2+4) = 17

f(2) + f(4) != f(2+4)

Since f(x) + f(y) must equal f(x+y) for a system to be a linear system, f(x) is not linear (as far as linear system theory goes; it's still a straight line on graph paper though).

This is straight out of the B.P. Lathi textbook that most colleges use.
I don't think you are applying the definition right. Look at the Wikipedia definition. That's what I see in all signals books.

You need x1(t) & x2(t) inputs and compute y1(t) & y2(t) outputs from a system under examination. The system under examination is linear if for scalars a & b,

a*y1(t) + b*y2(t) = H{a*x1(t) + b*x2(t)}

if we use x1(t) = 2*t+5 and x2(t) = 3*t+1 and choose the system to be say differentiation d/dt then try it.

I worked it out and it shows that the *SYSTEM* is linear. And we all know the operator d/dt (differentiation) is linear. If the system is constant gain of 1 just maps input to output, that too meets definition of linearity. And the inputs have offsets to them.

Your example is something else. but to test a system for linearity, you must try two functions on input.

#### Speakerguy

##### Active Member
Yeah, I can be a smart ass for sure

But to answer the OP's post:

Linear regulators take a noisy voltage in, step it down usually by 2V or more to your desired voltage, and filter out a lot of the noise. They are inefficient because they dissipate power in direct relationship to how much voltage drop you have across them, and the current through them.

Ex. A regulator with 35V in and 5V out will dissipate a whopping 45W when supplying 1.5Amps.That's 45W dissipated as heat to 7.5W delivered as power This is of course an extreme example and no one would ever use a regulator like this.

The big plusses with linear regs are very very low ripple and noise, and good output accuract (1% easily acheivable). Also really really cheap. Good to use in analog cktry.

Switchers - don't know much about how they are designed, but they are much more efficient than linear regulators with the drawback of high frequency noise introduced onto the power supply lines. Switchers actually increase in efficiency as output power goes up, maxing out at about 80% of their full output power (at anywhere from 75 to 95 percent efficiency). Switchers can be quite a bit more expensive.

I only use switchers where efficiency is a concern or I need to generate multiple and/or positive and negative voltages from a single DC voltage source. Other than that I always use linears.

Last edited:

#### Optikon

##### New Member
speakerguy79 said:
Yeah, I can be a smart ass for sure

But to answer the OP's post:

Linear regulators take a noisy voltage in, step it down usually by 2V or more to your desired voltage, and filter out a lot of the noise. They are inefficient because they dissipate power in direct relationship to how much voltage drop you have across them, and the current through them.

Ex. A regulator with 35V in and 5V out will dissipate a whopping 45W when supplying 1.5Amps.That's 45W dissipated as heat to 7.5W delivered as power This is of course an extreme example and no one would ever use a regulator like this.

The big plusses with linear regs are very very low ripple and noise, and good output accuract (1% easily acheivable). Also really really cheap. Good to use in analog cktry.

Switchers - don't know much about how they are designed, but they are much more efficient than linear regulators with the drawback of high frequency noise introduced onto the power supply lines. Switchers actually increase in efficiency as output power goes up, maxing out at about 80% of their full output power (at anywhere from 75 to 95 percent efficiency). Switchers can be quite a bit more expensive.

I only use switchers where efficiency is a concern or I need to generate multiple and/or positive and negative voltages from a single DC voltage source. Other than that I always use linears.
This is a good general conception of linear versus switch-mode. I would only add that the more efficient switchmode power supplies can be made smaller (due to smaller magnetics due to higher frequency of operation) and that can be critical in some size demanding applications.

Any analog work I do, I avoid switchers like the plague because of the noise issues.

Also note, it takes a pretty experienced designer to squeeze every last bit of efficiency out of a switchmode design. 80% is acheivable with your eyes closed in most cases. In my experience to get into the 90%'s requires much more work. The companies who sell switchmode controllers are making it easier on designers these days to acheive the higher efficiencies.

#### TheVictim

##### Member
I'm waiting for a small package I can plug into a circuit like a 78xx regulator but is actually a tiny switching mode supply. That way if you want 5v and all you have for supply is 12v, you aren't wasting a buttload of power as heat.

#### mvs sarma

##### Well-Known Member
TheVictim said:
I'm waiting for a small package I can plug into a circuit like a 78xx regulator but is actually a tiny switching mode supply. That way if you want 5v and all you have for supply is 12v, you aren't wasting a buttload of power as heat.
HI,
nationl semi's LM257x series does it. you may have to use min components.

also tl431 has a switch mode application for small currents

#### Suraj143

##### Active Member
Wow thanks a lot guys now I have understood very well Whats the different between SMPS & Normal PS,& also what is linear devices.

Thanks again.

Status
Not open for further replies.