ECE 477 Digital Systems Senior Design Project Spring 2005
§ Introduction
Constructing a compact real world device that would help people chat over the internet line is the core idea of our design project. The project shares a number of features offered in the commercially available VOIP(Voice Over Internet Protocol) phones. This particular project concerns with building two separate devices (phones) in order to prove the VOIP concept. Each of these devices would be able to detect calls, digitize the analog voice signals and vice versa and send the digitized packets over the internet line. There are several constraints pertaining to the efficiency of our design. Firstly, the device would need a bunch of [D/A]/[A/D] converters for digitizing the analog voice signals and vice versa. Secondly, a considerate thought should be placed on the methods of compressing the digitized voice signals. Compressing the signals would help us ease the process of storage and would lead to a proper data transmission. Thirdly, the idea of precise transmission is very much dependent on timing. Thus, speed/baud rate compatibility should be taken into consideration while selecting the hardware. Lastly, in order for the data to be sent over the internet line, an Ethernet module should be interfaced with the main controller. In view of these design constraints, we have chosen a list of components whose documentation appears in the following paragraphs.
§ Analysis of “real world” design constraints pertaining to the project
o Computation requirements : Successful transmission of data packets is the major requirement of our design and in order to accomplish this criteria a strong focus should be put on the sampling rates of the [D/A] and the [A/D] converters. Assuming that the normal telephone line operates at a frequency of about 4 kHz, the nyquist theorem suggests the sampling rate to be greater than 8 kHz. Thus, the codec to be used for digitizing the signal should have such a sampling rate.
o Interface (sensors/actuators/peripherals) requirements: Timing constraint should be the main focus while interfacing the codec with the microcontroller. Non-compatible baud-rates could lead to overwriting of data or an inclusion of more number of buffer registers. Fast transmission is a good feature of a device but a trade off to that feature would be the power consumption. The Data transfer between the Codec and the Microcontroller would take place using the SPI interface which further suggests us to choose a microcontroller whose SPI ports operate at a frequency greater than that of the codec’s PCM. Timing won’t be a limiting factor for the other interfaces that include Keypad with the DTMF and the DTMF with the LCD.
o Power supply constraints: The power supply is not a big problem as we intend our devices to be connected to a main DC supply all the time. But, power being the major trade off for the increase in speed of data transmission implies the need of some switching regulators that have a higher efficiency.
o Packaging (size/weight) constraints: We intend to constrain the size of our device to that of the normal cordless phone, thus indicating the use of smaller sized components wherever possible. Furthermore, it would be better to use the standard phone receivers instead of fitting the micro-phone and speaker in a compact case. Apart from the sizing factors, weight of the device is also an important contributor to the overall packaging. But, this shouldn’t be a problem as the design, dose not involve any gigantic hardware.
o Cost constraints: Implementation of features such as: Full-Duplex, Echo-Cancellation, Voice Activity Detector in our design could make our device a bit more expensive so a strong focus should be laid on having a balance between such features vs their implementation costs. Selecting components with matching voltage levels would be of great importance as this would reduce the need of level converters and thus reduce the expenses. Furthermore, a higher priority should be given to the components with greater number of in-built feature that we plan to implement. For example, getting a codec with features like: DTMF, VAD, etc.. would eliminate the necessity of getting separate components for each of these features. Such conditions while selecting components may help us reduce expenses in some cases. Hence the bottom line would be to make a cost/functionally efficient hardware selection.
§ Rationale for component selection
1.) DTMF chip: (Dual Tone Multi Frequency)
The DTMF is a pair of Generator and a Receiver. Depending upon the key pressed on the keypad, the generator would generate a specific tone which would then be decoded by the receiver and sent to the LCD display unit.
Selection Criteria:
1.) We plan to use a keypad with 12 different keys indicating a DTMF chip which can at least generate and decode 12 different tones.
2.) The Generator part would be interfaced with the keypad, while the receiver part with the LCD. The simplicity of these interfaces suggests us to have a DTMF chip that transmits data through the I/O pins and not use any kind of SPI interface.
Candidates choices :
1.) Motorola MC145740
Motorola MC145740 has both Encoding/Decoding functionality in one single chip which makes it good candidate for our component selection. It has the ability to encode/decode 16 different tones. But the fact that this particular chip uses an SPI interface to poll the data into some other device makes it a less tractable candidate.
2.) Jameco TCM5089N (Generator) / Zarlink MT8870 (Decoder)
The combination of both these chips would constitute the DTMF module. The combination is able to encode/decode 16 different tones plus it sends the decoded data using the I/O pins which makes much more tractable than Motorola MC145740.
Product Choice: Jameco TCM5089N / Zarlink MT8870D
2.) Voice CODEC
Candidate choices:
1) TI TLV320AIC22C
2) Winbond W6810, W682310
3) Oki Semiconductors ML7074-003
Selection Criteria:
The criteria that we were looking for when selecting a chip for the Voice CODEC included the Voltage at which it works, the sampling rate, the codec standards to which it complies, the kind of chip it was (LQFP, DIP) and the ease of implementation.
The different CODECS that we were looking at included:
1.) G.711: G.711 is a pulse code modulation (PCM) scheme operating at 8 kHz sample rate, with 8 bits per sample. According to the Nyquist theorem, which states that a signal must be sampled at twice its highest frequency component, G.711 can encode frequencies between 0 and 4 kHz.
2.) G.726: G.726 is for the conversion of a 64 kbit/s A-law or µ-law pulse code modulation (PCM) or 128Kbits linear PCM channel to and from a 40, 32, 24 or 16 kbit/s channel and is based on Adaptive Differential Pulse Code Modulation (ADPCM) technique. The ADPCM technique applies for all waveforms, high-quality audio, modem data etc.
3.) G.729: The G.729 encoder functionality includes Voice Activity Detection and Comfort Noise Generation (VAD/CNG) and the decoder is capable of accepting silence frames.
The table below compares a few VOIP candidates in terms of the CODECS they use,
sampling rate and the ease of their implementation.
Part Number / Voltage (V) / Sampling Rate (kHz) / CODEC Standard / Kind of Chip / Implementation EaseTLV320AIC22C / 3.3 / 16 / G.711 / 48-pin QFP / More difficult to implement since it’s a bigger chip
W6810 / 5 / 8 / G.712 / PDIP20,
SOP20, SSOP20 / Easier to implement since it’s a PDIP with only 20 pins
W682310 / 3 / 8 / G.712, G.711 / SOP24, SSOP20 / Easier to implement since it’s a PDIP with only 20 pins
ML7074-003 / 3.3 / 8, 16 , 32 / G.711, G.729A, G726 / 64-pin QFP / More difficult to implement since it’s a much bigger chip
The above comparison compels us to use the Part ML7074-003 in spite of the difficulty in using it. This is because of its ability to compress the data and thus send it more efficiently. But due to its non-availability, our next choice goes for W682310. Easier implementation, plus a lower power consumption makes W682310 a viable choice.
Product Choice: Winbond W682310
3.) Microcontroller
Candidate choices:
1.) Freescale MC9S12NE64
2.) Rabbit RCM 3000
Selection Criteria:
1) Microcontroller must have a 10Base-T Ethernet Module.
2) Large Memory.
3) A development environment that would be fairly reliable and easy for us to develop and debug on.
4) Should be cheap.
5) The rate of data transmission should be greater than the frequency at which the codec’s PCM operates.
The Freescale and Rabbit Microcontrollers, both meet our main requirements of having a 10Base-T Ethernet Module, having a large memory and a compatible baud rate. However, Rabbit does not meet the other 2 requirements. As per discussed in class, it was recommended not to use Rabbit since it does not have a very reliable development environment. Also, Rabbit was found to be considerably expensive as compared to the Freescale Microcontroller. Hence, we chose to use the Freescale.
Product Choice: Freescale MC9S12NE64
4.) POTS
Data Access Arrangement (POTS) Chip:
Candidate choices:
1. Analog Devices (AD1803/AD1804)
2. SiLabs Si3034
3. Zarlink MH88422
The purpose of this chip is to detect incoming calls, data transmission, etc.
All the above chips meet our requirements and some of the chips have been
successfully used in the previous projects. Since samples from Analog Devices
and Silicon Labs were not available, Zarlink was selected.
Product Choice: Zarlink MH88422
5.) LCD
A complete gamut of LCD was available, but LCD which meets our needs of 16X2 characters was chosen. Aesthetics was also a major factor while selecting the component. The choices were filtered down to the following two LCDs:
Candidate choices:
1)_Basic Micro LCD 16characters X 2 lines LCD
2) Hitachi LM018L
Basic Micro LCD 16characters X 2 lines LCD, Parallel Display with LED backlite.
Features:
· Maximum input voltage: 5.3VDC
· Operating input voltage: 5VDC
· 8-bit interface data bus
· Controller: HD47780 equivalent
· Character font size: 0.125"W x 0.200"H
· 14 pin/terminals
· Display size: 2.5"L x 0.7"W
· Module size: 3.4"L x 1.2"W x 0.5"T
The LCD from Hitachi LM018L is a 40 character x 2 line reflective type Liquid Crystal character display module. It includes a built-in LSI controller HD44780 which provides a range of features, including display clear, cursor shift etc.The module can be directly interfaced to a 4-bit or 8-bit MPU
We have chosen the first one because of its aesthetics and its robust operating conditions. It also has backlighting along with less power consumption. The Hitachi LCD was not chosen because though it was definitely much fancy, but the first one is simple and meets all the requirement of displaying the IP address.
Product Choice: Basic Micro LCD 16X2
6.) Keypad
Candidate Choice
1) KEYPAD 16 KEY BLK REAR PNL MNT by GrayHill Inc.
2) KEYPAD INDUSTRIAL 16 KEY CALC
Since the keypad is not one of the most complicated components, price is the most critical factor. Even the current rating also influences the choice, as it would determine the power dissipated which should be minimized. The current rating is 5mA for Grayhill Keypad. The aesthetic value of Grayhill keypad was more appealing.
Product Choice: KEYPAD 16 KEY BLK REAR PNL MNT by GrayHill Inc
§ List of References
http://www.jameco.com/wcsstore/Jameco/Products/ProdDS/32803.pdf
http://products.zarlink.com/selector_guides/Telephony___Dtmf_Receivers.htm
http://www.winbond-usa.com/products/codecs/
http://rabbitsemiconductor.com/products/rab3000/
http://www.freescale.com/files/microcontrollers/doc/fact_sheet/MC9S12NE64FS.pdf
http://www.clare.com/home/pdfs.nsf/www/CYG2217.pdf/$file/CYG2217.pdf
http://www.analog.com/UploadedFiles/Data_Sheets/500586613AD1803_0.pdf
http://www2.silabs.com/public/documents/tpub_doc/anote/Wireline/ISOmodem/en/an93.pdf