Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Positive Feedback in a Passive System

Jan Pompe

April 4, 2009

1 This is a work in progress and subject to change

2 Overview
2.1 Simple Active system
First of all we need some basics and we will do this by looking at a simple gain block with
feedback see. All signals in feed back control systems concern energy, energy potentials
or power so are subject to the laws of energy conservation and thermodynamics.

Figure 1: Simple closed loop gain block with feedback

First looking at the open loop case in Fig ?? the output EO = AEE = AEI where
A is gain and EI input signal since the feedback block is open EE is not modified. This

1
straight forward but adding a feedback loop and EE becomes EI − BEO so the output
now given by ansatz
EO = f (EI , A, B) (1)

EO = AVE = A (EI + BEO ) = AEI + BAEO (2)


Solving for EO yields
AEI
EO = (3)
1 − AB
Giving a gain for positive feedback and negative feedback where B will negative val-
ued or the feedback loop inverts so that EE = EI −BEO and the sign in the denominator
will be positive:
EO A
G= = (4)
EI 1 ± AB
For the case where the feedback function block is replaced with a short circuit B = 1
then gain is”
A
G= (5)
1±A
Even though the block diagram doesn’t show it to obtain a gain A > 1 requires an
internal power supply because we can’t get a larger signal out than we put in. In order
to get inversion of the feedback signal we also need an active system this also implies
the need for an internal power supply. A passive system is one with out such an internal
power supply. Then can we have positive feedback in a passive system?

2.2 Simple passive system


First we’ll define the passive system as having these factors:

• A Passive System is one without internal energy or power source.


• A lack of a power source implies the open loop gain A ≤ 1
• A feedback ratio B ≤ 1

The trivial case where A = 1 and B = 0 (i.e open loop) we have EO = EI unchanged.
In reality this never occurs as when there is energy transferred there are always losses
to entropy due friction etc though we won’t try to quantify this loss to entropy but gain
A < 1 always and the same goes for B see Fig ?? for schema.
The gain for the open loop where BEO = 0 is given by
EO
=A<1 (6)
EI
Thus we have a problem already with Eout < Ein we are trying to push a portion of
the output energy against a potential gradient. E might be seen as either temperature or
volts and the flow is heat or charge respectively and like water neither will spontaneously
run uphill.

2
Figure 2: Block Diagram of Positive Feed Back Schema

3 Potential Divider
A simple resistive potential divider see Fig ?? can be represented as a feedback control
system.
In order to this we need the output potential Vo as a function of Vo , Vi , RS and RL
we begin with ansatz:
Vo = f (Vi , Vo , RS , RL ) (7)
We can write the current as a function Vo , Vi and RS thus:
Vi − Vo
i= (8)
RS
multiplying i by RL gives the output voltage then we may write
RL
Vo = (Vi − Vo ) (9)
RS
Is it really a feedback system? The answer is ”no”. It is important to realise that it
cannot be represented as a positive feedback system because the current can only flow
if there is a potential difference across RS and of course RL . If we gather terms we can
see how Vo is a function of only Vi and the two resistors i.e. Vo = f (Vi , RS , RL ).

Vi RL
Vo = (10)
Rs + RL

3
Figure 3: Potential Divider

4 Matching the Model


Now we can look at simplifying even this we have Thevenin’s Theorem which states:

A combination of voltage sources and resistors may be represented as a


single voltage source Vth which is the open circuit voltage at the terminals
and a single series resistor Rth whose value is Vth divided by the current in
the circuit with the voltage sources replaced with short circuit.

Fig ??A shows the circuit we are going to Thevenise which is just a potential divider
where the output potential Vo having no external load will become Vth . If the voltage
source is then replaced with a short circuit (shorting the power supply is not recom-
mended) then the resistance RS ||RL is seen by the (virtual) Vth source and this becomes
Rth in Fig ??B.

Figure 4: Thevenising a Potential Divider

We know that Rth = RS ||RL = RRSS+R RL


L
(generally known as product over sum
rule)and we can draw that as in Fig ??C and now one can easily recognise the structure
in Fig ?? where RS ≡ g(T ) and RL ≡ F (T, pH2 O, pCO2 ). It should now be clearer
that all energy flow is in the same direction and nothing returns to provide feedback.

4
However if it is still not clear we can normalise and set RS = 1 and RL = z then:
RS RL z
R= = (11)
RS + RL 1+z
which as well as being in the familiar feedback form of a closed loop transfer function
with negative feedback (Eq ??), can never be greater than or even equal to 1.

5 Power Transfer Optimisation


Power delivered into RL can be worked out by ohms law.
2
V2 Vi2 RL

Vi RL 1 z
PL = o = × = 2 ≡ (12)
RL RS + RL RL (RS + RL ) (1 + z)2

which we can plot (See Fig ?? ).Expanding the equation gives:

Vi2
PL = Rs2
(13)
RL + 2RS + RL
Rather than trying to find the derivative of the whole thing to find maximum PL it
will suffice to find the minimum of the denominator so differentiating with respect to
RL gives:
d Rs2 R2
+ 2RS + RL = − S2 + 1 (14)
dRL RL RL
yielding a maximum or a minimum when

RS2
2 =1
RL
(15)

or when
RS = RL (16)
z
Figure ?? is a plot of (1+z)2 which has a source resistance of 1Ω we can see that a

maximum power to the load is delivered when the load resistor is equal to the source.

6 Where to Now
We have seen that feedback as it is generally understood is just not going to happen
in a passive system. Adaptive parameter control however is possible as a feedback it is
unlikely because we still have the problem of doing work against a gradient but a feed
forward where a bit of the power passing through is diverted to to change the control
law. For example the load can be replaced with a thermistor with either a positive or
negative temperature coefficient. A thermistor with a positive temperature coefficient
(PTC) increases it’s resistance with temperature and one with a negative coefficient

5
Figure 5: Plot of power delivery into varying load

(NTC) decreases its resistance with temperature. One of these will better resemble our
atmosphere and the claimed feedback and the claim is that more heat implies more
opposition so it must be the PTC.
The more energy the PTC will dissipate the hotter it becomes the higher resistance
the less current to the ground and Vo climbs. That should do it, now where are the data
sheets and soldering iron.

7 Atmospheric Model
Negligible

You might also like