Lecture 1 Flashcards

1
Q

What is quantised boxcar

A

quantised but not sampled continous

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is sampled data

A

sampled but not quantised

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is the 3rd step of A2D conversion after sampling and quantising?

A

Coding- putting into binary

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Formula for quantisation width

A

(xmax - xmin)/(no. levels -1)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

For Ensemble recording what is the best amount of recordings needed?

A

many

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is Ensemble recording

A

At each point work out the average from many trials

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Why is Median recording better than Ensemble recording?

A

It can deal with a smaller amount of samples as high noise samples wont throw it off

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

How many trials do you need for moving average filtering

A

only one

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is moving averaging filtering?

A

find average of points near eachother

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What do you need to be careful with with moving average filtering?

A

length will change so need to pad with zeros at end

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What kind of filter does moving average act like?

A

low pass

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

In a moving average filter what is the comprise between order and distortion?

A

As order increases passband gets smaller (removes more of the noise) but also increases distortion (as removes part of the signal also)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

opposite of periodic

A

aperiodic

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What does a finite total energy signal mean for power?

A

zero power

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What does finite average power mean for energy?

A

infinite

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Power=

A

energy/time

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Reasons for incorrect sampling [3]

A

hardware problems
Cant sample that first
noise might cause highest frequency

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

How to fix incorrect sampling [2]

A

Use a really high frequency
low pass in time domain before converting to frequency domain

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

Difference between DFT and DTFT

A

DTFT has an infinite sum, frequency is continuous - DFT everything is discrete

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

DFT symmetric?

A

magitude is symmetric around fs/2 and phase is anti symmetric

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

With DFT what needs to be thought about in terms of sample size

A

big enough to carry a full cycle to get proper results.

22
Q

What is the picket fence effect?

A

If you want to know the value of X(m) that is between discrete values

23
Q

How do you solve picket fence effect?

A

Zero padding to increase the display of DFT

24
Q

When picking a cycle to do DFT what causes spectral leaking

A

the truncation if the signal doesn’t start and end at the same place- (as we are assuming its periodic but it doesn’t need to be)

25
Q

How do you solve spectral leaking?

A

Windowing

26
Q

Why the z transform?

A

can verify stability

27
Q

What is the region of convergence?

A

area where z transform is defined and it also has an impact on BIBO stability

28
Q

What is delay in z transform

A

z^-n (where n is delay)

29
Q

What is spectral filtering?

A

Keeping the components of the frequency that you want and removing undesired frequencies

30
Q

an ideal filter can do what?

A

pass signal components without any distortion

31
Q

When you are far from a pole zero pair what will h(w) do?

A

stay close to 1

32
Q

To emphasise signal components what do you need to do?

A

What poles very close to unit circle in this range

33
Q

To attenuate signal components what do you need to do?

A

Place zeros close to unit circle (notch filter)

34
Q

What is an IRR filter?

A

has current inputs, past inputs and past outputs

35
Q

What is a FIR filter?

A

current and past inputs only

36
Q

Which of IRR and FIR structure diagram have feedback

A

IRR

37
Q

Which of FIR and IRR only have zeros and therefore is always stable

A

FIR

38
Q

Out of FIR and IRR which one has finite impulse response

A

FIR

39
Q

Why do you sometimes filter in the time domain not the frequency domain?

A

Less computational time needed

40
Q

How can you make a sum filter better?

A

Using more than one in a cascade

41
Q

Why would you transport from analogue rather then straight to digital?

A

techniques highly advances, extensive tables

42
Q

In Butterworth what is omega c

A

cut off frequency i.e. frequency at -3db

43
Q

What is the window method

A

Truncate by N+1 then shift N/2 to keep causality

44
Q

What is the Gibbs phenomenon?

A

oscillatory behaviour caused by noise from using a rectangular window in FIR filtering

45
Q

what can Increasing N of FIR not decrease

A

ripple peak

46
Q

What can increasing N of FIR do?

A

sharpen transistion

47
Q

Which window has best transition

A

Hamming

48
Q

Which window has best ripple decrease

A

Blackman

49
Q

What is wc normally

A

wp+ws /2

50
Q

with SNR is higher or lower better

A

higher is better

51
Q

What is the fast Fourier Transform?

A

Same method but tried to reduce complexity so its quicker