Grundlagen Microcontroller Analog I/O. Günther Gridling Bettina Weiss

Size: px

Start display at page:

Download "Grundlagen Microcontroller Analog I/O. Günther Gridling Bettina Weiss"

Gary Webster
6 years ago
Views:

1 Grundlagen Microcontroller Analog I/O Günther Gridling Bettina Weiss 1

2 Analog I/O Lecture Overview A/D Conversion Design Issues Representation Conversion Techniques ADCs in Microcontrollers Analog Comparators 2

3 Analog Digital Conversion (ADC) In this part, we will give a very basic introduction to analog to digital converters (assuming no prior knowledge) we will briefly describe three ADC variants hardware details will be ignored (aperture delay and jitter, filters, etc.) for more detail, refer to the lecture script 3

4 I/O: Analog Digital Conversion Problem: The point of a microcontroller is to control processes, but microcontrollers are digital logic devices, they use discrete (digital) values to represent data however, real world processes are continuous (analog) How to connect a (digital) microcontroller to the (analog) world? 4

5 ADC: Basic Design Issues Sensor Controller Actuator?? 0 / 1 5

6 ADC: Basic Design Issues Using a position weighed system (binary numbers), a MC can already perform computations on numbers beyond zero and one > solution: assign plain and simple numbers to represent the current value of a continuous variable, like 'low' analog value < > small number 'high' analog value < > big number 6

7 ADC: Basic Design Issues Sensor Controller Actuator 0 / 1 65 = = 12 7

8 ADC: Basic Design Issues Note carefully, however: the transformation from analog values to numbers is application specific the range of expected analog values must be carefully examined the number range must be chosen accordingly the particular value is still meaningless to the MC; it is just a number 8

9 Example value range: how to represent, e.g., a temperature? Depends on application: human body temperature > C might suffice ambient temperature > more like C foundry (iron) > up to ~ 1500 C and more... ADC Value Representation 9

10 ADC Value Representation Let's assume our application is an ordinary clinical thermometer for home use. below 35 C hypothermia: user is probably long past bothering with home thermometers above 42 C: user is probably dead we want some safety margin. (always remember: a safety margin is good for you and the user) > C might be a reasonable range 10

11 ADC Value Representation What range of numbers to use for our C? floating point numbers are A Bad Thing, since MCs usually do not come with big fat FPUs ignoring the actual value range is very inefficient contrary to most desktop applications, efficiency is still pivotal in embedded systems 11

12 ADC Value Representation Our range of values is = 12 C: try a common desktop CPU's 32 bit: 12 / 2 32 = ~ 2,8 x C > probably overkill being cheap, we try 4 bit: 12 / 2 4 = ~ 0,75 C > might be a tad too coarse 12

13 ADC Value Representation Let's just try it properly, why don't we: do we need extra high accuracy? > probably not for a 1 EUR home thermometer (might be different for hospital appliances) what type of display do we need? > could be, e.g., 2.1 digits, meaning 2 digits to the left and one to the right of the decimal point > we use a resolution of 0,1 C or slightly better 13

14 ADC Value Representation And now, finally, for the point of the whole thing we compute our number range: 12 C / 0,1 C = 120 > we need 121 different numbers MC uses binary values, so a power of 2 might be prudent nearest power of two is 128 = 2 7 > >= 7 bit can be used to represent our temp. 14

15 ADC Value Representation This is oversimplified, of course: economic facts (availability, cost) override engineering arguments usually, the software guy/gal has to work with whatever the hardware engineers fancied to build But: embedded software developers must have a basic grasp of hardware design issues if only to properly argue with the hardware developers. 15

16 ADC Value Representation There are some common pitfalls: mind the datasheets: don't waste time and/or resources enforcing 16 bit resolution ('the more, the better') when the actual temperature sensor may only provide a +/ 5 % accuracy but: don't set unnecessary limits: if 16 bit is economically just as viable as 7 bit, implement it you never know where this particular code might be used in the future 16

17 ADC Technology Analog digital conversion principles: flash (Parallelverfahren) tracking (Zählverfahren) successive approximation (Wägeverfahren)... and more (including various hybrids) 17

18 ADC Technology Example: hardware of a flash a/d converter What's it to me, you ask? Well, you need some idea of hardware to decide: how many bits can you trust? how fast can you expect the conversion to be? are there any special considerations involved? 18

19 ADC Flash Converter Simple idea: just compare the input voltage to an appropriate number of equidistant reference voltages. We need a so called 'comparator': U 1 > U 2 => U out = V+ U 1 < U 2 => U out = 0 V 19

20 ADC Flash Converter Example: Assume our input voltage maximum is U, and we want to convert with 2 bit resolution (four different values). We could use the circuit to the right: (Quiz: Why three converters, when we need four states = two bits)? 20

21 ADC Flash Converter Answer: because 'no active comparator output' is a valid state, too. Thus, with three comparator outputs, we get four states for our two bits worth. 21

22 In this example (U = 4.0 V), the converter claims the input voltage to be about 2.0 V, but it actually is much closer to 3.0 V. What to do? ADC Flash Converter Choice of reference voltages is critical: 22

23 ADC Flash Converter => We could achieve a 'round to nearest' by shifting the reference voltages down one half step (adding a half step to the input voltage is more difficult). Now the converter produces the expected result.... but we introduced a new problem. 23

24 ADC Flash Converter Now, the voltage range assigned to the top comparator is too wide: 3U/8 versus U/4 for the middle comparators and U/8 for the bottom one. What now? 24

25 ADC Flash Converter We change the voltage assigned to the LSB from U/4 to U/3 like this: 25

26 ADC Flash Converter We still need to convert the four output lines to two data bits. This is accomplished by a so called 'priority encoder': this gadget gives the 1 based number of the highest active input in binary (or 0 if no input is active) 26

27 ADC Flash Converter Yet another problem: how to provide the reference voltages? => use voltage divider: R 1 : R 2 = U 1 : U 2 27

28 ADC Flash Converter Now, we can derive all necessary reference voltages from of a single one: 28

29 Still one problem: ADC Flash Converter the priority encoder does not sample all inputs at the same time input voltage can change at any time => MC might receive invalid states solution: edge triggered D flip flop, where Q is sampled at an edge on C 29

30 ADC Flash Converter Finally, our 'flash' analog to digital converter is complete. 30

31 Advantages: ADC Flash Converter extremely fast (up to GHz sample rates) digital sample & hold via D flipflop (analog sample & hold is much harder to implement) Disadvantages: very complex: a flash converter with a measly eight bits precision already needs 255 comparators 31

32 ADC Tracking Converter Analog digital conversion principles: flash (Parallelverfahren) tracking (Zählverfahren) successive approximation (Wägeverfahren)... and more (including various hybrids) 32

33 ADC Tracking Converter counter starts at 0 d/a conv. counter value is fed back to comp. as long as U in is > conv. value, counter goes up when U in becomes < conv. value, counter goes down 33

34 ADC Tracking Converter A tracking analog to digital converter in action: 34

35 ADC Tracking Converter A tracking analog to digital converter in panic: 35

36 ADC Tracking Converter Advantages: simple hardware high resolutions (20 bit and more) achievable Disadvantages: very slow 36

37 ADC Successive Approximation Analog digital conversion principles: flash (Parallelverfahren) tracking (Zählverfahren) successive approximation (Wägeverfahren)... and more (including various hybrids) 37

38 ADC Successive Approximation Quite similar to tracking converter: The SAR instead of a simple up/down counter makes all the difference... 38

39 ADC Successive Approximation Successive approximation register: 1) start with msb 2) set bit to one 3) if d=0, reset bit 4) repeat from 2) with next bit 39

40 ADC Successive Approximation A successive approximation register in action: 40

41 ADC Successive Approximation A successive approximation register going astray: 41

42 ADC Comparison A successive approximation and a tracking ADC in action SA ADC achieves higher sample rates (but much slower than flash ADC) for slow signals, tracking ADC may perform better 42

43 ADC Successive Approximation The SA ADC is a good compromise: quite fast moderately complex SA ADCs are common in MCs 43

44 Analog I/O not every controller has a/d converters controller typically has 8 16 analog channels ADC resolution is typically 8 12 bits controller normally also has a comparator analog I/O is controlled through registers (control register, data registers) 44

45 Using the ADC: Analog I/O set up ADC (select channel, select mode) start conversion > ADC converts (some cycles), sets ready flag when done either poll ready flag or set up interrupt when ready flag set > read result 45

46 ADC Features: Analog I/O conversion mode (single cycle, continuous) automatic channel increment user sets initial channel after each conversion, channel is incremented reading value triggers next conversion conversion trigger modes on timer event ( > periodic measurements) on external event,... 46

47 ADC Features: Analog I/O channels normally single ended possibly differential channels channels grouped pairwise possibly programmable gain (1x, 10x,...) 47

48 Analog Comparator: Analog I/O two analog inputs AIN0 and AIN1 comparator output is high if AIN0 > AIN1 interrupt sources on comparator output rising edge falling edge toggle may trigger input capture (see Timer) 48

49 Things to consider: Analog I/O conversion speed (succ. approx. takes several cycles) noise (variation of the input voltage can make huge difference!) high switching frequency of neighboring digital lines causes noise do not use digital I/O on an analog port use high quality (filtered) reference voltage use noise cancellation (turn off unused parts of MC) 49

50 Analog I/O Summary several ADC variants; tracking ADC is most simple, successive approximation ADC is common in MCs, flash ADC is fastest finding the right ADC is an involved task (resolution, precision, speed, cost) when programming ADCs, always keep hardware constraints in mind once again: a basic understanding of the particular target hardware is essential 50

51 Lecture Summary controller has 8 16 analog channels analog comparator some interrupt sources (conversion complete, comp.) different conversion modes, channel increment problem of noise, accuracy 51

Grundlagen Microcontroller Counter/Timer. Günther Gridling Bettina Weiss

Grundlagen Microcontroller Counter/Timer Günther Gridling Bettina Weiss 1 Counter/Timer Lecture Overview Counter Timer Prescaler Input Capture Output Compare PWM 2 important feature of microcontroller