Charge Transfer Devices Research Articles

Charge-transfer devices filter complex communications signals : J. J. Tiemann, W. E. Engeler, R. D. Baertsch and D. M. Brown. Electronics, November 14, 1974. p. 113

Charge-transfer devices filter complex communications signals : J. J. Tiemann, W. E. Engeler, R. D. Baertsch and D. M. Brown. Electronics, November 14, 1974. p. 113

Tunable matched filters for pulse-code modulation using charge-transfer devices

A tunable matched filter for the major PCM codes has been synthesized utilizing a developmental bucket-brigade delay (BBD) line. The variable delay feature of the BBD allows continuous tuning of the matched filter over a wide range of bit rates. By combining and weighting the outputs of two BBD lines, matched filters for nonreturn-to-zero (NRZ), return-to-zero (RZ), and biphase codes are realized with a signal-to-noise performance within 0.8 dB of theoretical.

Phase errors in transversal and recursive filters realised with charge-transfer devices

It is shown that nonideal amplitude transfer in charge-transfer devices can cause large phase errors in transversal and recursive filters using these elements.

A wide-band low-noise charge transfer video delay line

A new wide-band low-noise charge transfer video delay line is introduced utilizing a bulk charge transfer device (BCD) with a self-aligned electrode structure, and a simple integrated sample-hold circuit. Brief analysis of charge transfer characteristics of the BCD and design considerations for the proposed device are included. Several kinds of 128-element devices are designed and examined. Measured transferred signal bandwidth is 4 MHz at 10 MHz clock of voltage swing as low as 10 V, with signal-to-noise (S/N) ratio of 42 dB. The device itself can operate beyond 20 MHz clock rate. Moreover, the device features the use of an elaborate sample-hold circuit to eliminate the complex reset pulse and filtering circuitry, high packing density of 18 /spl mu/m per element with 3-/spl mu/m line spacings assuring a high fabrication yield, and process compatibility with conventional Si gate MOS technology.

Memory of the future

Speed, capacity and cost are the essential features on which any memory store is judged, and there was much talk of cycle times, access times and costs per bit when the Magnetism Group met to debate the topic 'Memory stores – magnetic versus the rest', on 20 September. The profusion of emergent technologies can often baffle the onlooker, with the exponents of semiconductors, charge transfer devices, magnetic bubbles and holographic stores, not to mention the good old fashioned magnetic discs and tapes, and ferrite cores all jostling for attention and a share of the market.

A charge-transfer-device logic cell

A logic scheme capable of performing AND-OR logic functions using charge-transfer devices is presented in this paper. Applications of the logic cell are discussed. A half-adder, a ‘flip-flop’, a majority logic gate, an analog-to-digital and a digital-to-analog converter are obtained by extending or modifying the basic scheme. An integrated MOS BBD realization of the AND-OR logic cell is described in detail. Experimental results are presented and shown to be in agreement with theoretical calculations.

Optimum linear filtering for charge-transfer devices

A method of filtering analog signals transferred through a charge-transfer device (CTD), which is capable of greatly reducing the distortion introduced by incomplete charge transfer, is suggested. By using transversal filters consisting in part of single CTD storage cells to achieve the necessary phase delay, the output charge packets can be combined in a manner which offsets the presence of incompletely transferred portions of charge acquired by the packet of interest. The transfer function of the filter is calculated directly from that of the CTD. Examples are given which indicate that in most cases of interest only a few filtering stages are necessary to reduce the distortion to 1 percent of its unfiltered size.

Frequency response of a multiplexed charge-transfer delay line

Analytic expressions for the frequency response of multiplexed charge-transfer devices (MCTD's) are derived. A brief discussion of the advantages and disadvantages of several CTD architectures is given.

Theory of Noise in Charge-Transfer Devices

The noise introduced into charge packets transferred through and stored in charge-transfer devices is calculated in a manner that includes all important relaxation, suppression, and correlation effects. First, the noise induced into each packet during each transfer phase from thermal, trapping, emission-current, and leakage-current fluctuations, whose statistics are nonstationary, and from clock-voltage fluctuations, whose statistics are stationary, is determined. Relaxation of the transferring charge to these fluctuations is found to suppress their size. Second, the accumulation (collecting) of the noise as each packet is transferred through the device is calculated neglecting the role of incomplete charge transfer. Attention is drawn to the significant differences between the collecting of storage-process noise, which is unsuppressed, transfer-process noise, whose spectral density is nearly totally suppressed at low frequencies, and modulation noise, which is nearly totally suppressed for digital and analog signals. Third, the role of incomplete charge transfer in suppressing the collecting of the noise is shown for digital signals and indicated for analog signals. We conclude with a numerical calculation of the maximum possible signal-to-noise ratio that can be expected from charge-transfer devices. The presentation is sufficiently general and detailed that, with a minimum of background in formal noise theory, one can use the approach to evaluate noise in many novel, solid-state devices.

Experimental Investigation of a Linear 500-Element 3-Phase Charge-Coupled Device

A linear, 500-element, 3-phase charge-coupled device, originally built as a high-resolution linear image sensor, has been chosen as a representative structure for single-level, 3-phase charge-coupled devices to exemplify the performance of such devices in detail. Charge-handling ability and transfer efficiency have been studied as a function of various design parameters and operating conditions. Most of the observed functional dependences are well understood and agree with the expectations based on model calculations. However, various problems are encountered in these structures. An unusually wide spread of the performance of different devices and slow instabilities are observed. They are attributed to a lack of control of the interface potential in the gaps between the transfer electrodes. Some emphasis is placed on a more detailed description of the various measurement techniques used. These techniques are of a general interest since they are applicable to other charge-transfer devices.

Charge-transfer device analogue delay lines: A building block for electronics

The impact of bucket brigades and charge-coupled devices on the fields of imaging and memory has been well publicised, but what of their use as simple analogue delay lines?

Error Rates of Digital Signals in Charge Transfer Devices

We calculate the probability of error in detecting digital signals transferred through a charge transfer device in the presence of incomplete charge transfer, random noise in the device, and detection uncertainty in the detector. The coefficient of incomplete charge transfer is assumed to be independent of charge-packet size, and both the device noise and detector noise are assumed to be Gaussian. Error probabilities for two-level and four-level codes are computed for the cases of both simple static and optimum dynamic detection. For rms detection voltage level fluctuations V <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</inf> of the order of tenths of volts (much larger than the random noise fluctuations in the device), a very rapid increase in error probability (from ≍ 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−20</sup> to ≍ 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−5</sup> ) is found to occur for a very small (20 percent) change in V <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</inf> . This indicates that detection level fluctuations will have to be held down to a few hundred millivolts at most. To achieve equal error rates with an error probability of about 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−14</sup> , V <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</inf> for the detection of four-level codes will have to be about 3.5 times smaller than for two-level codes. Comparison of error probabilities under static and dynamic detection shows that in CTD's improved detection has a greater potential for reducing error rates than improved coding.

Operational Limitations of Charge Transfer Devices

The incomplete transfer of charge and the existence of random noise lead to the primary operational limitations of charge transfer devices. Owing to the signal dependence of the residual charge, which accumulates as a result of the incomplete transfer, signal detection with static detection levels becomes seriously impaired before the onset of significant signal attenuation or noise degradation. A scheme using dynamic detection levels is found to greatly extend the operational range of CTD's and achieves the minimum possible error rate for detecting uncorrelated charge packet sizes. By contrast, simple coding procedures are found to be ineffective in overcoming signal degradation due to incomplete transfer. Shannon's expression for maximum information transmission capacity is transformed into an expression for maximum information storage capacity. It is found that significantly larger storage capacities are possible with CTD's than have been achieved.

Spectral density of noise generated in charge transfer devices

We calculate the spectral density of the noise added to an analog signal passed through a charge transfer device (CTD) in terms of the charge fluctuation associated with each transfer. A correlation between noise in neighboring elements substantially suppresses the energy content of the noise associated with charge transfer for frequencies much less than the clock frequency.

Incomplete transfer in charge-transfer devices

The authors present a general analysis of incomplete charge transfer in charge transfer devices. By using a lumped-charge model to characterize the dynamics of the charge transfer, they calculate /spl alpha/, the small-signal coefficient of incomplete transfer, in terms of single-device small-signal parameters. They show that three contributions to incomplete transfer are common to all charge-transfer shift registers: an intrinsic transfer rate contribution, an output conductance or feedback contribution, and a storage-capacitance modulation contribution. The analysis suggests that most relevant measurements necessary to characterize incomplete transfer can be made on a single cell of a shift register that has input and output diffusions. This allows transconductance etc., to be measured on a small-signal basis as a function of transfer current, from which one can calculate transfer efficiencies. An analysis of the influence of interface states which may be important for some charge transfer devices, is also included.

Digital signal transfer in charge-transfer devices

The nonlinear properties of digital signal transfer through charge-coupled device and bucket-brigade shift registers are considered in terms of adjacent bit charge levels. A signal transfer efficiency is defined and shown to be a useful parameter for charge-transfer device shift register simulation. Approximate equations are developed for the worst case output bit levels and an approximate formula for the optimum input `fat' zero level, with respect to the worst case output signal `window', is obtained. The analysis includes only the intrinsic incomplete charge transfer properties of CTD's under square-wave clock pulsing conditions. Comparison of the theory and preliminary experimental results for CCD indicate good quantitative agreement.

Transversal filtering using charge-transfer devices

Techniques are presented for making transversal filters using charge-coupled devices and bucket-brigade devices. In a CCD transversal filter, the delayed signals are sampled by measuring the current flowing the clock lines during transfer, and the sampled signals are weighted by a split electrode technique. In a BBD transversal filter, the delayed signals are `tapped' with a source follower whose load determines the weighting coefficient. Examples are given of CCD and BBD filters that are `matched' to particular signalling waveforms, and the limitations of charge-transfer devices in matched filtering applications are discussed. Finally, the application of CTD transversal filters to other signal processing functions is discussed.

A Fundamental Comparison of Incomplete Charge Transfer in Charge Transfer Devices

Using a small-signal analysis, we present a general comparison of the more important contributions to the incomplete transfer of charge in the following charge transfer devices: (i) the polyphase charge-coupled device and the IGFET bucket-brigade shift register, where individual charge transfers are single-step processes, and (ii) the two-phase charge-coupled device, the conductively connected charge-coupled device, the tetrode bucket brigade, and the stepped-oxide bucket brigade, where individual charge transfers are two-step processes. A recently proposed lumped-charge-model approximation is made in order to estimate the time dependence of the transferred charge including both drift and diffusion. In this calculation we also include the effects associated with the injection of charge from a diffused source into the IGFET channel which modify the current-voltage behavior at low currents. Using this calculation of the time dependence of the transferred charge, the various contributions to incomplete transfer, including those due to trapping in the interface states, are derived and compared for each of the charge transfer devices of interest.

Differential mode of operation for bucket-brigade circuits

A differential circuit configuration is described that suppresses unwanted clock-frequency spectral energy at the output of a charge-transfer device. The differential mode of operation also eliminates the d.c. offset voltage that is characteristic of these devices. Experimental results are presented.

Charge Transfer Devices

This is a review describing the use of charge transfer devices for digital memory, analog delay, and image sensing. Short descriptions of different types of charge-coupled devices and MOS bucket-brigade devices are presented. Those factors such as transfer inefficiency, noise, and dark current which affect the performance of these devices in the above applications are discussed. Various possible organizations of charge transfer devices to serve different functions are described. Many charge transfer devices have been fabricated and already indicate a high degree of achievement. Transfer inefficiencies per transfer in the range 10−3 to 10−4 have been measured. Finally, it is projected that charge transfer devices will rapidly find their way into certain analog delay, image sensing, and digital applications.