Limits
of use of the Numerical Integrated circuits |
The
Shift of the clock |
Parasites |
|
Assembly
of the Digital components |
||
|
Return
to the synopsis |
To
contact the author |
Low
of page |
Created it, 06/09/09
Update it, 06/09/29
N° Visitors
2. - CONCEPTS ON THE MICROPROCESSORS
Currently, the complex numerical systems are carried out with microprocessors. With technological progress and the lowering of the cost of the integrated circuits, the use of the microprocessors extended to sectors where formerly traditional logical circuits were employed (logical doors, rockers, registers, meters…).
The microprocessor is an integrated circuit on a broad scale (L.S.I.) which includes a calculating unit dealing with the information provided by the outside of the system.
These various data processings are determined by a program.
This program is consisted a succession of operational phases having to be carried out in a given chronological order.
The working method used is appreciably different from that employed for the design of the traditional electronic circuits.
Indeed, for a traditional assembly, it is simply necessary to determine the logical components (doors, rockers, meters, registers…) necessary to fulfill the desired functions and to establish connections between these various elements.
In a circuit containing microprocessors, it becomes necessary to introduce an additional dimension which is the programming, namely that the circuit functions in close relationship to the course of a program.
This type of realization introduces a new stage into the flexibility and the flexibility of operation of the digital circuits.
Indeed, it is enough to change the program contained in the memory so that the microprocessor-based system carries out new functions.
One can announce the increasing use of the microprocessors in the world of today (machine tools with numerical controls, data processing, telematics, electric household appliances, road traffic, railway…).
Moreover, the microprocessor became a standard product since it can fill of the very diverse tasks. It is resulted from it a considerable lowering from the production costs.
In this short introduction to the microprocessor, we present figure 12 the synoptic one of a microprocessor met in the electric household appliances (washing machines, cookers…)
This microprocessor especially designed for the ordering of the english bonds is often called programmer (to control).
The ROM memory has a capacity of one or two kilooctets and it contain the program of work.
Memory RAM has a capacity of a few hundreds of bits. They are information relating to the state of the system and the control units which forward by this memory.
The program is a whole of instructions (orders) achievable in a given order.
The meter of program allows chronological unfolding in the order given of the program.
The calculating unit treats the data which reach him and generates information necessary to the achievement of the program.
The control unit ensures coordination between these various elements.
These programmers can also have other elements (analog-to-digital converter, clock real time (timer)…).
In theory, a microprocessor can treat all the functions traditionally carried out by a digital circuit. However, the microprocessor-based system can prove too slow to carry out certain operations. It will be, in this case, necessary to resort to cabled circuits (traditional circuits).
In addition, the use of a microprocessor often requires to have a system of associated development to conceive the program.
It thus results a certain cost from it from design. This is why in a certain number of cases, it is even more interesting to resort to traditional diagrams (logical doors, rockers…).
3. - LIMITS
OF USE OF THE NUMERICAL INTEGRATED CIRCUITS
3. 1. - THE MAXIMUM FREQUENCY OF CLOCK
It is a parameter of which it is often necessary to hold account in the numerical systems design.
During the examination of the meters, this concept was already presented. There had appeared that if one made work certain integrated circuits into too high frequencies, it were risks of operation, even complete breakdowns in certain cases.
In fact, this concept speed limit of operation returns to two fundamental data : travel time of a signal through a given circuit (logical doors, rockers…) and the time of transition from a logical state at the complementary logical state (transition or switching time).
It is for these reasons that the manufacturer always specifies the maximum speed of operation of an integrated circuit.
This speed (or frequency) maximum is a few MHz for circuits CMOS and a few tens of MHz for circuits TTL.
However, it is wise to leave an interval of safety and not to make function a system (or a component) at its authorized maximum frequency. In addition, the manufacturer generally provides two values of maximum frequency.
One is the typical value (or average value) which is the maximum frequency to which many the circuits of the same type can function. The other is the minimal value : it is the maximum frequency to which one is absolutely certain that all the circuits of this type can function. The latter is thus a little less low than the typical value.
In general, it is necessary to take account of this minimal value and not of the typical value bus if one works with the latter, the risk exists that the component one cannot function correctly.
That is to say the following example : Meter CMOS
40193.
The table below gives the typical and minimal maximum frequencies according to the supply voltage used.
The maximum frequency is related to three factors.
The first, as the table indicates it above, is the supply voltage. The maximum frequency increases when the supply voltage increases.
A second factor is the load capacity CL for an exit MOS. The maximum frequency also increases when this capacity decreases.
The table of figure 13 indicates the effects of these two factors on the maximum frequency.
For that, one deferred the travel times tp circuit 4011 B according to these two parameters.
The increase in the maximum frequency (equivalent to a reduction in the travel time) is explained by the reduction in the switching time of transistors MOS.
The last factor is the temperature. The maximum frequency increases when the temperature decreases.
The example of figure 14 relates always to circuit HEF 4011 B.
Up to that point, we took into account one component. However, in a numerical system, there is generally a whole whole of integrated circuits connected. Other factors can then interfere on the maximum speed of operation.
Let us consider for example the circuit of figure 15 consisted by two rockers D synchronous and two doors NAND.
When an active face of clock arises, the two rockers commutate simultaneously.
The data present on the D1 entry is transferred at exit Q1 and two doors NAND commutate one following the other. The data present on the D2 entry changes.
Figure 16 shows the various delays which are added, due to the travel times through the first rocker and two doors NAND.
The time of presetting (set up time) is the time during which the new data must be present on the entry of the rocker before the active face of the clock.
We will calculate the maximum frequency to which can function this circuit.
For that, we calculate the minimal period necessary to its operation.
The rockers D are of the type 74C74, doors NAND of the type 74C00 and the voltage supply is + 5 volts (when the supply voltage increases, the travel time decrease).
The first delay (travel time) is worth 300 ns, the second and the third delay are worth 90 ns and the time of presetting is of 100 ns.
The minimal duration of the period of clock is thus equal to the three time lags added to the time of presetting i.e. :
300 + 90 + 90 + 100 = 580 ns
The maximum frequency is worth (1 / 580) x 109 = 1,7 MHz.
This frequency is lower than the maximum frequency of clock relating to the rocker 74C74 which is worth 2 MHz.
If the number of logical doors in series increases, the maximum frequency of operation of the circuit decreases.
With 4 doors NAND in series, the period would be worth :
300 + (90 x 4) + 100 = 760 ns
The maximum frequency would be : (1
/ 760) x 109 = 1,3 MHz.
It is thus necessary in general to limit the number of logical
doors put in series the ones after the others.
Whenever it is not possible, it is then necessary to insert additional
rockers as indicated figure 17.
If one preserves the same presetting and travel times that previously, we obtain the following results :
In the first case, when the additional rocker does not exist:
Minimal period of clock = 300 + 90 + 90 + 90 + 90 + 100 = 760 ns.
In the second case, the minimal period of clock must be sufficient so that the data presents on the entry of a rocker n can be on the entry of the rocker n + 1 in assigned times.
In this case, this period is equal to the sum of the travel times of a rocker added to those of the two logical doors, like with that of the time of presetting.
We find:
300 + 90 + 90 + 100 = 580 ns
The additional rocker is used as intermediate register of storage for the data which forward entry at the exit of the circuit.
This system makes it possible to increase the flow of the data through the digital circuit.
3. 2. -
THE SHIFT
OF THE CLOCK
The shift of clock (or CLOCK SKEW) is also a problem involved in differences in travel time through logical doors.
The circuit of figure 18 makes it possible to highlight this problem.
If the travel times of the two buffers B1 and B2 are very different, there can be a faulty operation of the circuit.
Initially D1 is in state 0 and D2 with state 1. If the travel times of the two buffers are identical, the active face of clock is applied simultaneously in CLK 1 and CLK 2, as shown in the figure 19.
From the moment t0, Q1 passes to state 0 after a time T which corresponds to the switching time of rocker 1. Q2 remains with state 1 since at the moment t0, D2 is to 1. Q2 changes state only with the face of clock according to (moment t1).
Let us suppose now that the travel time of the B2 buffer is definitely larger than that of the B1 buffer. It can occur disfonctionnement illustrated figure 20.
The Q1 exit of the first rocker passed to state 0 before the face of clock does not arrive on entry CLK 2 of the second rocker.
The D2 entry is thus with state 0, whereas it should be with state 1 at this time.
Term “SKEW” represents a “slip” of the clock signal.
A solution with this problem consists in inverting the two clock signals as indicated figure 21.
Thus, the second rocker will commutate a little before the first and one is sure that the initial data present in D2 is transferred in Q2.
3. 3. -
PARASITES
The parasites are defined as being disturbances affecting an electronic signal. The origins of these parasites are innumerable but one can classify them in two categories : parasites of natural origin like those generated by the storms for example and the parasites of artificial origin such those produced by an engine.
These parasites can be sufficiently important to disturb the operation of a logical unit. That can be translated in a concrete way by the taking into account at one time given of a logical level H to the place of a logical level L or vice versa.
In general, one will endeavor to minimize the effects of the parasites on a logical system.
It should be also noted that a logical unit can function very well during the phases of study and tests carried out by the manufacturer and inoperative once to be installed at the customer. Indeed, the electronic environment is not any more the same one and certain physical phenomena which were not taken into account at the time of the design can appear at this time.
These various phenomena can be of electromagnetic nature (radio transmission in high frequency, fluorescent tubes…) or of static nature (electric charges stored in various substrates such as fitteds carpet…).
Technology employed is not without consequence on the effect of these parasites.
The circuits carried out in technology MOS are much less sensitive to the parasites than those carried out in technology TTL. First of all, technology MOS is slower (times of transition are longer than for the TTL) and the margin of noise is much more important than in TTL. It can reach 45 % of the supply voltage in MOS, whereas it is only 0,4 volt in TTL.
It is for these reasons that in high-risk industrial circle of parasites, it is preferable to employ technology MOS.
There is also a problem involved in the adaptation between a line of transmission and the load located in end of line.
The line has a characteristic impedance. It is necessary that this characteristic impedance (Z0) is appreciably equivalent to the impedance of the load if one wants to limit the phenomena of reflection of the signal.
Figure 22 shows an example of reflections on a line which is not adapted to the load.
The reflections can create problems for fast circuits (TTL, ECL…), but also for slow circuits (CMOS).
Rebounds can occur in a synchronous system with clock, and they can be taken into account like clock signal. It is the case presented figure 23.
Inductive couplings can also occur between two or several lines.
It is necessary that the lines of transmission as well as the connections of the printed circuits are shortest possible.
In addition, one installs condensers of decoupling on the feeders near the cases of integrated circuits. These condensers have a value of 0,1 µF with 0,01 µF and absorb the parasites which forward on the feeder (Vcc) and on the line of mass.
3. 4. - ASSEMBLY OF THE DIGITAL
COMPONENTS
You know that the numerical components belong to different logical families : TTL, CMOS, ECL…, each one having its particular characteristics (food, time travel, consumption…).
Consequently, when it is a question of connecting between them components of different families, it is necessary to take account of their electric characteristics. In general, it is necessary to insert a circuit of interface (= circuit of adaptation) between these components.
A circuit of interface is also necessary to connect a numerical system to the external world.
First of all let us consider connections between components of the same family.
In technology CMOS, it is possible to connect a high number of logical doors to the exit of another. On the other hand, in TTL, this number is limited much more.
These various notions (fan-in and fan-out) were seen in the chapters relating to technology of the components.
In technology MOS, a logical entry of door absorbs or produced (according to the logical level) a current of 0,005 µA.
An exit MOS can provide (or absorb) at least 1,75 mA for a supply voltage of 5 volts (8 mA for 10 volts).
Therefore, a logical exit of door MOS can theoretically be connected to a number of entries MOS located between 350 000 and 1 600 000.
However, the fan-out of a door MOS does not exceed 50 because of the input capacitance of a door MOS which is worth 5 pF.
With 50 entries, one obtains 250 pF.
The travel time increases notably with the number of entries connected to an exit MOS.
When one passes from 50 pF to 100 pF, with 5 volts of supply, typical travel time voltage of a door AND passes from 80 ns to 110 ns.
To connect doors MOS to doors TTL, it can pose a problem related to different supply voltages.
In addition, an exit TTL with the state H can be at 2,4 volts, whereas an entry MOS with the state H must be with a potential from 3 to 3,5 volts (for Vcc = 5 volts).
To solve this problem, one connects exit TTL to the food through a resistance called «pull-up» (“to draw upwards”).
That makes it possible to record the level of tension when the exit is with the state H.
Figure 24 represents two doors TTL and CMOS fed under 5 volts, connected according to this principle of assembly.
An exit MOS can be connected to an entry TTL of the type Schottky, low power (74 LS) with the TTL Low-Power (74 L).
In all the other cases, one can resort to circuits MOS having a bufferized exit and providing a higher current.
If circuits MOS are fed starting from a tension different from 5 volts, it becomes necessary to intercalate a buffer between exit CMOS and entry TTL, as well as a buffer with collector open between the exit TTL and entry MOS (figure 25).
Figure 26 represents the electric diagram of a buffer TTL to open collector.
External resistance is connected to the exit of the circuit and the food. This supply voltage can be higher than 5 volts.
A door with open collector can also order a relay directly.
Another case to be considered is that of figure 27 where a circuit MOS is fed between - 5 volts and + 5 volts.
The pin of circuit MOS, generally cabled with the mass, in this case is connected to - 5 volts. In such a situation, it is necessary to intercalate between exit TTL and entry MOS a transistor MOS.
When exit TTL is on the level H, the led transistor and the entry of circuit MOS are carried on the level H (+ 5 volts). If exit TTL is on the level L (0 volt), the transistor is blocked and entry MOS passes to the potential - 5 volts.
The fact of feeding door CMOS between - 5 volts and + 5 volts makes it possible to have a speed of operation higher than with a food of 5 volts.
This theoretical is now finished. The next one will treat converters analogical / numerical and numerical / analogical.
|
Click
here for the following lesson or in the synopsis envisaged to this end. |
High
of page |
 Preceding
page |
Following
page |
