DE19513988A1 - Band limited analogue bit pattern reproduction on digital plane - Google Patents

Abstract

In this process the characteristic data of the signal amplitude, form, data rate and coding for the analogue bit pattern can be freely defined. The condition for the transmission of the digital data in an analogue video signal is the control of the band width of the channel, that is the rate of rise of the edges of the bit pattern must be limited. This is possible if the bit sequences are represented as (cosine)<2> pulses, so that more than 95 % of the spectral energy lies below the frequency 1/Ti.

Description

For the transmission of additional signals, such as VPS (Video Program System) is the vertical blanking interval of the Video signal used. The additional signal represents any one Bit pattern represents that in the analog video signal as luminance formation is transmitted in a white-black pattern. For the Generating the white-black pattern is the bandwidth of the to comply with the analog transmission system. The band limit characterizes the waveform of the bit pattern and adjusts the filtering represents the bit pattern according to a sin² function.

State of the art is that all additional signals to be transmitted in the vertical or horizontal blanking interval on analog Paths are keyed. This happens in the form that the Data feeder has an internal clock frequency, which in the Numerical value of the data rate of the bit pattern or an integer Corresponds to multiples of it. With this clock frequency it will Bit patterns generated as "1.0 / 0.1 / 1.1 / 0.0" sequences, with a sin² filter band-limited and as an analog, value and Continuous-time signal stamped into the blanking interval.

With the introduction of the digital standard ITU-R601, the Video information within the studio as more value and time discrete Data stream processed, the transmission to the viewer takes place as an analog signal. The adaptation of the digital data to the analog transmission system has the following problem: Additional data in the blanking gaps can be in the digital level should not be added because the sampling frequency is not integer multiple in the transmission grid of the analog Inserts additional data. The consequences of digital keying are Jitter of the data signal and an incorrect waveform.

The invention is therefore based on the object of a method with which it is possible to develop any changeable analog bit patterns, taking into account the Waveform and band limit to be keyed in on digital level.

To solve this task, those in the license plate of the Claim 1 measures proposed. The advantage this method lies in the high-precision generation of a analog bit pattern that can be found anywhere in the digital Studio complex can be added without the video data in one convert analog signal. Another advantage lies in the connection of a digital system to an analogue environment.

Other advantages are:

• 1. generation of high-precision analog waveforms,
• 2. any band limitation,
• 3. any waveform (sin², cos², trapezoid, triangle, etc.),
• 4. any bit duration of the simulated analog waveform.

This allows data carriers with digitally simulated analog Additional data are generated for transmission to the viewer converted into an analog signal only once at the studio output Need to become.

The vertical and horizontal blanking interval of a video signal are used to transfer additional data such as B. Video program system (VPS), wide-screen signaling (WSS) and vertical Timecode information (VITC) used. Acting on these signals it is a bit sequence (coded according to NRZ or biphase pre font), which is a white-black pattern in the luminance signal be keyed in. A logical "1" corresponds to the white level (500-700 mV, depending on the application), a logical "0" dem Black level (0 mV).

A prerequisite for the transmission of digital data in the analog video signal is compliance with the bandwidth of the transmission channel, ie the edges of the bit pattern must be limited in terms of rise times. This can be achieved by representing the bit sequences as cos² pulses and causing more than 95% of the spectral energy to be below the corner frequency 1 / T _i (see Fig. 1).

The generation on the analog level is carried out by passing a data signal with the data rate frequency f _B over a cos² filter. The bit duration T _B corresponds to 1 / f _B and is therefore directly dependent on f _B.

At the digital level, samples must be used so that after a digital-to-analog conversion the required There is a white-black pattern, which is a desired bit pattern represents and the above requirements for Compliance with the bandwidth met. Here is the sampling theorem by Shannon-Nyquist, which states that the data rate must be less than half the sampling frequency.

The present invention provides a method of generation any band-limited white-black pattern at free selectable sampling frequency and sampling width The prerequisite is that the given sampling frequency in one rational (i.e. by an integer quotient representable) division ratio to the data rate frequency.

The invention is described below using an exemplary embodiment explained in more detail.

An arbitrary bit sequence is given, which as a value-discrete, band-limited signal with any data rate and one arbitrary sampling frequency in a digital data stream should be keyed in.

The given data becomes a sequence of samples generated, which are inserted into the digital data stream. First an integer quotient Q must be found, the corresponds to the ratio of data rate and sampling frequency. This quotient gives the ratio of oversampling to Downsampling again:

f _A = sampling frequency of the digital system
f _B = data rate frequency
x, y ∈ integers
x = oversampling factor
y = downsampling factor

In the present method, the desired analog one Waveform with the required amplitude x times Oversampling process calculated as a discrete-value signal. These discrete values are in a value table for another Processing available.

The necessary stock of values in the value table results from the Quotient Q and corresponds to the counter value x (see formula 1). Any multiple of the value x can also be used, it must be noted that y also multiplies accordingly must become.

Oversampling achieves a virtually higher resolution than the original signal. This creates the prerequisites for downsampling to read out intermediate values that correspond exactly to the system frequency in step size. (System frequency = probing frequency of the digital system)
In one embodiment, the following are given:
f _A = 13.5 MHz
f _B = 5 MHz or 5 Mbit / s
It is being searched for:
x (number of values for the oversampling process

⇒ X = n · 27, y = n · 10

For the given probing frequency 13.5 MHz and data rate 5 Mbit / s the value set in the value table is 27 (or a multiple from that). So there must be at least 27 values of the trainees Waveform can be calculated for each data bit.

If the value table were now read out with the sample frequency f _A = 13.5 MHz, a data rate of 13.5 MHz / 27 = 0.5 Mbit / s would result.

With the calculated set of values, data rates of 0.5 Mbit / s n are generated in a downsampling process, whereby only every nth value of the value table is read out. in the For example, the required bit rate is 5 Mbit / s, i. h., n must have the value 10 and matches y. For another Data rate outside the 0.5 Mbit / s grid are the values n or to calculate y according to Formula 1.

Calculation of the table of values:
Only the transitions between the individual bits of the given bit sequence are important for the digital simulation of a bit sequence. This results in four options according to Table 1.

Only the edges are to be calculated, since if the level remains, the table of values only consists of a constant value. The maximum value (500-700 mV as a digital code word) stands for a permanent "1" level, the minimum value (0 mV as a digital code word) for a permanent "0" level. The values for the edges depend on the choice of the desired analog signal form. In order to reduce the bandwidth, instead of a square-wave signal, a sin²-shaped (often also cos², sin, cos, etc., can be converted into sin² by the known addition theorems) is used.
If the signal form is an axisymmetric function (e.g. cos, sin, cos², sin² function etc.), the calculation of an edge is sufficient, the values for the additional edge result from the reverse order to calculated edge.

General formulas for calculation (here for sin² function, however, any periodic function can be used), linear quantization curve required (see Fig. 2)

z (n) = values for the rising edge
z _max = maximum quantization value corresponding to the maximum analog value
z _Offset = minimum quantization value corresponding to the minimum analog value
n _max = oversampling factor = x from formula (1)
n = 0, 1, 2, 3,. . ., n _max -2, n _max -1

If there is no linear quantization curve of the DA converter the z (n) must be in accordance with the quantization specification be modified.

With formula (2) the values for a rising edge were calculated. The values of the falling edge result from the z (n) are read in reverse order:

• z (n) ⇒ rising edge
• x (m) = z (n _max -1-m) ⇒ falling edge

m = 0, 1, 2,. . ., n _max - 2, n _max -1

For the two states, low level and high level, the values z _offset and z _max must be used for all quantization values.

After all required quantization values are calculated the data to be inserted can be generated:

Example: Bit sequence 011010010110 should be inserted.

First the bit sequence must be in the states according to table (1) being transformed:

01 ⇒ rising edge
11 ⇒ high level
10 ⇒ falling edge
01 ⇒ rising edge
10 ⇒ falling edge
00 ⇒ low level
01 ⇒ rising edge
10 ⇒ falling edge
01 ⇒ rising edge
11 ⇒ high level
10 ⇒ falling edge

Then the states are shown in a list of values corresponding quantization values replaced.

The values for the lowest data rate are in the list of values with the longest rise time and narrowest spectrum of the Signals. Downsampling does not make every value read out and keyed into the data stream, but every kth Value, the data rate increases by k times the rise time is shortened by k times and the spectrum of the analogue converted signal expands to k times. (k = x off Formula 1)).

Are over- and downsampling processes according to formula (1) applied, you get the desired digital values, which in the data stream are to be used to achieve the required waveform in the to achieve analog signal.  

1. Processor-supported implementation

In the process-oriented implementation of the method, all processing steps are implemented independently of the processor by software in accordance with the block diagram of FIG. 3.

Sampling frequency f _A and required data rate frequency f _{B are} entered in input field ( 1 ) and the resulting quotient Q is calculated according to formula (1). The input field transfers the x value (oversampling factor) to the oversampling block ( 2 ).

In the oversampling block ( 2 ) the required oversampling values z (n) are calculated according to the formula (2) according to the required signal form (e.g. sin²) and made available in a table for the subsequent downsampling block ( 3 ).

In the downsampling block ( 3 ), every yth value is read out according to the input field ( 1 ). Downsampling is directly related to the sampling frequency of the digital system (see Fig. 2).

The data from ( 3 ) are output in the output field ( 4 ) with the system frequency f _A.

2. Realization with hardware for real-time applications

Please refer to the block diagram in Fig. 4. In it are:

• (1) mod x counter with step size y:
  x = oversampling value from formula (1)
  y = downsampling value from formula (1)

The counter ( 1 ) is a mod x counter with the step size y, where x corresponds to the number of oversample values and y to the downsampling factor from formula (1). The counter is clocked with the system clock (sampling frequency of the digital system). The current counter value addresses the value table ( 3 ).

If the counter ( 1 ) exceeds the maximum value x, a carry signal is generated.

(2) 2-bit shift register

The bit sequence that corresponds to the analog signal to be generated runs through the 2-bit shift register ( 2 ). The carry signal from (1) functions as a clock signal. The outputs Q₀ and Q₁ represent the current state (rising, falling edge, high, low level) and select one of the four sub-tables from the value table ( 3 ) via the demultiplexer ( 4 ).

(3) Table of values with sub-tables ( 3 a), ( 3 b), ( 3 c), ( 3 d)

The x quantization values for the four states according to table (1) are stored in the value subtables ( 3 a) - ( 3 d). The sub-tables are addressed by the counter ( 1 ). The outputs of all sub-tables are on the multiplexer ( 4 ).

Block ( 3 a) contains the values for the rising edge corresponding to a 0 → 1 transition.

Block ( 3 b) contains the values for the falling edge corresponding to a 1 → 0 transition.

Block ( 3 c) contains the values for a 1 sequence corresponding to a 1 → 1 transition.

Block ( 3 d) contains the values for a 0 sequence corresponding to a 0 → 0 transition.

(4) 4 to 1 multiplexer

Depending on the two bits in the shift register ( 2 ), the multiplexer switches a sub-table from ( 3 ) through to the output. The data provided at the output can be inserted into the digital signal with the system clock.

Claims (6)

1. A method for digital reproduction of band-limited, changeable analog bit patterns, characterized in that the keying in the digital data stream is carried out with the given sampling frequency under real-time conditions and the data rate of the bit pattern is in a rational division ratio to the given sampling frequency of the digital data stream, the characteristic data for Signal amplitude, signal shape, signal data rate and signal coding of the analog bit pattern can be freely defined for any bit pattern to be transmitted. 2. The method according to claim 1, characterized, that replicate the video program system information technical Guideline No. 8R2 of the public law Broadcasters (short: VPS) in a digital video data stream according to ITU-R601 / 656 or EBU Tech. 3267-E with required Signal amplitude and bit cell duration in the vertical blanking interval is keyed and after coding in a composite signal as band-limited analog bit pattern is present. 3. The method according to claim 1, characterized, that the replica of the television text after techn. Directive No. 8R4 of public service broadcasters into one digital video data stream according to ITU-R601 / 656 or EBU Tech. 3267-E with the required signal amplitude and bit cell duration in the  vertical blanking interval is keyed and after coding in an FBAS signal is present as a band-limited analog bit pattern. 4. The method according to claim 1, characterized, that the simulation of wide screen signaling (WSS) according to ETS 300 294 into a digital video data stream according to ITU-R601 / 656 or EBU Tech. 3267-E with required signal amplitude and Bit cell duration is keyed into the vertical blanking interval and after coding into a composite signal as a band-limited analog bit pattern is present. 5. The method according to one or more of claims 1 to 4, characterized, that further analog additional signals in the blanking gaps and on anywhere in the analog CVBS signal in one digital video data stream according to ITU-R601 / 656 or EBU Tech. 3267-E with required signal amplitude, bit cell duration and Band limit can be keyed. 6. The method according to one or more of claims 1 to 5, characterized in that the value x of the counter ( 1 ) corresponds to the number of oversample values and the value y of the counter ( 1 ) corresponds to the downsampling factor and addresses the value table with the current counter value, whereby when the maximum value x of the counter ( 1 ) is exceeded, the counter ( 1 ) generates a carry signal which drives the 2-bit shift register ( 2 ), the outputs (Q₀ and Q₁) the current state (rising, falling edge, high -, low level) of the bit pattern and controls the demultiplexer ( 4 ), which evaluates one of the four sub-tables of the value table ( 3 ) and inserts the required data word into the data stream, whereby in the value sub-tables ( 3 a) - ( 3 d) the x quantizing values of the counter ( 1 ) for the four states according to table (1) are stored and connected to the multiplexer ( 4 ), block ( 3 a) entering the values for the rising edge accordingly em 0 → 1 transition and block ( 3 b) contains the values for the falling edge corresponding to a 1 → 0 transition and block ( 3 c) contains the values for a 1 sequence corresponding to a 1 → 1 transition and Block ( 3 d) contains the values for a 0 sequence corresponding to a 0 → 0 transition, the characteristic data such as signal amplitude, signal shape, signal data rate and signal coding being freely definable.