# frequency response -- for linear regulator

In a linear regulator on input side digital voltage enters (say DC 60V ) & on the o/p side digital voltage(let say DC 15V) is produced.

What exactly do we mean by the frequency response of the Linear Regulator in this link. http://www.onmyphd.com/?p=voltage.regulators.linear.series&ckattempt=1

Device which take a DC voltage as input and a DC voltage as output, then why this factor Frequency is in frequency response. What i know is frequency is related to time varing AC signals, but here linear regulator is dealing with DC voltage ?

• What is this "digital voltage" you refer to ? Last time I looked all voltages were purely analog. Digital refers to circuits that are abstracted to have to possible input/output states: 0 or 1. Mar 16, 2016 at 10:50
• What is i/p as DC & o/p as DC value? Mar 16, 2016 at 11:00
• means 60V i/p & 15V o/p ... Mar 16, 2016 at 11:09
• Regarding this i/p o/p: you're the only one using it so that's why you get questions about it. Just write input / output like everyone else does and avoid confusion. Mar 16, 2016 at 11:12

## 3 Answers

No real linear regulator is perfect. In the ideal case, you could vary the regulator input voltage arbitrarily over its input voltage range without any change at all in the output voltage. Real parts you can actually obtain will all pass some of the changes in the input voltage to the output voltage.

For example, I just looked up a datasheet for a 7805 regulator. It has a line regulation spec of 10 mV at 500 mA out and 7.5 to 20 Volts in.

The line regulation spec is at DC, meaning the frequency content is not taken into account. For higher input voltage frequencies, more of the input voltage variation will be passed to the output. This is usually not very well specified, if at all. In the 7805 datasheet I was looking in, this isn't explicitly specified, only shown in a graph:

For the 5 V out part, it seems to match the DC case up to only about 300 Hz. In that range the ripple rejection is 80 dB. This means input variations will be attenuated by 80 dB to the output. For example, if the input contains a sine at 300 Hz with 1 Vpp amplitude, then you'd get 100 µVpp on the output.

After 300 Hz, the ripple rejection goes down, meaning more of the input variations are passed to the output. At 20 kHz you're already down 20 dB from the DC case. A 20 kHz 1 Vpp sine on the input would result in 1 mVpp on the output. And, it gets progressively worse at higher frequencies, to the point they don't even want to show you how bad past 100 kHz.

This is one reason for putting a good high frequency cap on the input. The impedance of this cap will go down with frequency, thereby attenuating those frequencies before they get to the regulator. Put another way, the active electronics deals with the low frequencies, and the cap deals with the high frequencies. Together, they give you good rejection across a meaningful frequency range.

LDO (low dropout) regulators usually have worse ripple rejection. I often put a ferrite chip inductor of about 1 µH and a few 100 mΩ in series with the LDO input, followed by a 20 µF ceramic cap to ground right at the regulator input.

A voltage regulator is not only dealing with a DC voltage.

There are always small disturbances superimposed on to (added on top of) that DC voltage. These disturbances can be ripple, noise, signals coupled in from clocks.

The behaviour of the regulator regarding these disturbances is the what is sometimes called "AC" behaviour.

This AC behaviour is especially important in relation to voltage regulator which use a feedback loop to control the output voltage (this includes almost any regulator you can find). If the AC behaviour is not well designed then the feedback loop can make the regulator oscillate so it starts to generate its own signal ! You want to avoid that like anything. That's where AC behaviour analysis comes in.

Also you might be interested in how much a regulator can suppress voltage variations at the input ending up at the output. That's also AC behaviour.

To give a specific example:

Lets say you are running a microcontroller and you are toggling an output or two.

Every time the output toggles, there will be a 'blip' of extra current (in one direction or another) pulled through the power supply pins of the microcontroller.

That current must ultimately come from the voltage regulator (yes, there is decoupling for fast transients, but this must still be recharged).

A blip is not steady state, and therefore must be analysed using transient methods; a transient is, by definition, an ac event.

As FakeMoustache notes, the regulator must be able to deal with the transient event and therefore you will see frequency dependent graphs in voltage regulator datasheets.