Introduction

This text is not intended as a comparison between vendors, features and other aspects pertaining to certain product sales. This document is intended to give the reader a working knowledge of basic multiplexer technology. This is turn will allow the reader to make intelligent decisions in the multiplexer purchasing process. The term multiplexer was given to an electronic device that time multiplexes video pictures from numerous cameras onto one VCR. This means that one field or frame from one camera was switched to the VCR, then immediately following that picture was another field or frame from another camera and so on from each camera, then it started over again. This technique maintains full resolution as does normal video switchers but reduces the time between recorded images so the 2-3 second dead time between normal camera switching was reduce to 17ms (milliseconds) times the number of cameras.

Multiplexer History

The early multiplexers were basically video switchers that could mark each camera with a unique number in the vertical interval. This required the cameras to be genlocked or v-phased so the VCR would see a continuously composite sync signal so it would not loose servo lock on the switched incoming video signals. The playback mechanism merely switched to the correct camera only during its active period on the tape while switching to a flat field (usually a solid gray picture) for the rest of the time. This caused severe flicker but produced a viewable single camera image and was effective.

Later digital memory was used to save the active camera until a new picture was displayed. Early genlocking or v-phasing was poor and cameras drifted causing the VCR to record garbage, therefore resulting in poor playback regardless of digital memory or not. Next a two field digital memory was used to time base correct the incoming signal to guarantee continuous composite video to the VCR. The main benefit of this technology was that this device guaranteed continuous composite sync to the VCR regardless of the video quality. A side benefit was that non-genlocked or any camera could be used. This method is still the preferred method used among multiplexer manufactures. The problem of this approach is that the encoder and decoder is rather costly and is usually packaged together since they are basically the same internal hardware.

Since new cameras have quality genlocking systems and/or stable line locked vertical intervals, a reversion to the old technology is feasible. If time base correcting is not required, a low cost analog only encoder can be used that merely marks each vertical interval with the camera number. The decoder would then contain the digital field memories for playback. This system would be viable for retailers that have many small locations and one common location for review. The many locations would have the low cost encoder and the few common locations would have the high cost decoder.

Camera Encoding/Decoding

The most common type of camera identification is the insertion of digital information into the vertical interval. This is accomplished by dividing a line of video into say eight different sections. Each section is defined as a one or a zero by either the presence of black video or white video. By doing this with eight sections this can interrupted as one byte of digital data which can be converted to a number from 0-255. Some of these sections or bits are used for error correction since most multiplexers only go up to 32 cameras and the full use up to 255 is not required. This line of digital data is usually repeated on at least two adjacent lines to make sure the decoder reads the data correctly and to guarantee data is available in both the odd and even fields of video. Since every vendor uses a different technique incompatibility is a problem. If switching from one vendor to another your old tapes might be able to playback on the new unit but usually the recordings of the new unit can not playback on the old one. Check specifications on compatibility if you mix vendors.

Multiplexer Math

By definition NTSC video records 60 pictures or fields per second. One odd field and one even field (2 fields) make up one frame of video. This is important to multiplexers since they are basically a time base multiplexing device. In a true real time recording mode the above standard is adhered to but when we switch to virtual real time or time lapse video recording everything changes. In the CCTV world there really does not exist a standard to how many pictures per second are recorded in virtual real time or any time lapse mode. Normally in virtual real time you have 20 pictures per second. In a 24 hour time lapse mode you will have 5 pictures per second for a single density VCR. When you go to double or triple density VCR this number is increased. In other words all VCR vendors are somewhat different and you must consult the manual for this information. This is the main reason when setting up the multiplexer there exists a menu or dipswitches to tell the multiplexer this information. If this information is not set up correctly the multiplexer will switch between its input cameras either too slow or too fast resulting in lost video cameras recorded on the VCR or at best reduces the efficiency of the multiplexer. When the VCR changes speed and the multiplexer does not know this fact the above problem exists.

To eliminate all this problem some multiplexers use a camera switcher pulse. This pulse tells the multiplexer electronically when it has recorded the last video image on tape. The multiplexer then can switch to the next input camera, process it and have it ready for output to the VCR. The advantages of using this pulse is that the multiplexer is guaranteed to stay "in sync" with the VCR even if the VCR changes recording speed. Another big advantage is you do not have to learn multiplexer math and rest easy your system will record perfect images no matter what.

When the menu does not have a particular model of VCR in its menu than you plan to use you must use the manual setup in the multiplexer to define the timing. Basically you tell the multiplexer how many pictures per second your VCR is recording depending on your recording speed. Again this information is in the VCR manual. The multiplexer will then compute how often to switch its cameras.

The most important thing that is normally forgotten about multiplexers is that they still are basically video switchers. When you connect many cameras and the time lapse recording is too slow the time between recording a single image from a particular camera may be too long to catch any event. You basically take the number of camera inputs and divide that by the recording pictures per second.

# of cameras / record pictures per second = update rate

If you had a 16 camera system and were recording at a 168 hour time lapse mode then you will have 17.40 seconds from recording a new image from any one camera input. This of course would have little use in any application since someone could walk by and never be recorded. If we reduce the recording to 24 hr time lapse then we would have a new image every 3.2 seconds. This would be more acceptable but not very applicable in high traffic areas. If we went to 24 hr virtual real time at 20 pictures per second then we would have a new image every .8 seconds. This is more like it and will be useful in most applications.

Motion Detection / High Density VCRs

Many other features come into play to improve the above results. Many multiplexers have video motion or activity detection. This is where electronically the multiplexer can determine if something has change in the video scene of any camera. If nothing has changed the multiplexer records less pictures per second of that camera therefore increasing the pictures per second of the other cameras that have motion or something happening. Do not be fooled by trade show demos of playback where 15 images are of still pictures and one has a lot of motion. The multiplexer in this type of demo will appear to have an update rate 15 times greater than reality. Also newer VCRs have double and triple density recording. These products can actually record more pictures per second for any given time lapse speed than standard or single density VCRs. Also some vendors of multiplexers have limitations on the update rate due to the design of the unit itself. Some can record at field rate (60Hz), some only at frame rate (30Hz) and some even slower. The slower ones are usually the PC based systems that rely on PC software to do the multiplexing rather than hardware in most stand alone multiplexers. Be careful here, if it is a field recorder the above math equation stands, but if frame then divide the time by two, if slower then divide by the update rate.

In retail applications where many people are moving all the time care must be given to the above limitations of multiplexers. In industrial and residential uses most cameras view the same scene with little motion and a larger number of cameras can be used and utilize the video motion detection more efficiently.

Multiplexer Displays

Most multiplexes have the ability to display 4, 8, 16 or 32 pictures simultaneously on the screen. Usually the image update rate is same as the output multiplex rate but some vendors have up to real time multiscreen. This feature is useful since this screen is for viewing only since the multiplexed output is a full screen high resolution image. Selection of single images or customized viewing screen is available but mainly for show only. In reality a multiplexer is designed as a stand alone unit for recording images, not for a guard station for viewing live. The multiscreen is mainly useful in playback of all recorded images and later single camera selection when something interesting is viewed.

Multiplexer Architectures

The low cost encoder would consist of a standard vertical interval switcher. Each camera when homed, would be electronically marked in the vertical interval with a camera number. Each camera would have to be genlocked and if color was used it would have to be VBS locked for color stability between cameras. B&W cameras can be genlocked with twisted pair while color must be genlocked with RG59U. Preferably the video switcher would be controlled by the VCR camera switcher pulse so as to sequence to a new camera upon recording the previous one. When video is lost from a camera it must be excluded from the sequencing as to not upset the genlock nature of the system.

If the playback signal from the VCR is guaranteed to be in a predetermined sequence, it is not necessary to have a two field memory playback unit. When the user desires to monitor only camera two, the field memory could be controlled to update the viewable picture at the end of camera ones video. By using a one field frame memory cost is greatly reduced. This is the technique used in the JVS 925 VCR which has this feature built in and can be used with a low cost analog encoder with no ID such as the 4PSSMT.

Most current encoders do a time base correction on the incoming switched video signal. This is accomplished by holding the the last picture digitized in a field memory until another field memory has been filled with the new camera video picture. The multiplexer then switches to the new field at the vertical inter-val of the previous field thus genlocking the two electronically. This requires two field memories on the encoder. Since the decod-er is integrated into this unit a more practical design is to store the desired field in one memory and constantly update the second until the next camera field is present and then switch to that field and so on.

A/D Conversions

All current high cost decoder/encoder designs utilize analog to digital (A/D) converters to convert the standard video signal into a digital format for use with common digital logic devices. It is then later converted from digital to analog (D/A) for output back to the video world. Although a multiplexer only requires only one A/D most designs us a minimum of two to increase the multiplexing rate. The number of A/Ds is usually not important to the end user but is defined here to show that a field rate multiplexer is more costly to produce that slower ones. Multiplexer front ends are of two types, analog color space decoding/encoding or completely digital using new computer multimedia chips utilizing the DMSD (Digital Multistandard Decoder) technology. That later produces better quality color and makes one unit changeable from NTSC to PAL usually by software control.

Multiplexer Types

The simplex multiplexer is basically the lowest cost since it has the least amount of features. A simplex multiplexer has the ability to time base multiplex to the VCR and all the timing setups required. The main thing a simplex multiplexer does not have is the ability to record and show a multiscreen display at the same time whether it is 4,8,16 or 32 camera screen. Normally simplex multiplexers have no capability for multiscreen display at all.

A duplex multiplexer has the ability to display the multiscreen and record to the VCR the multiplexed data. Some duplex multiplexers can playback from one VCR while recording on another but you give up the multiscreen viewing at this time.

A full duplex multiplexer has the ability to record the multiplexed output to one VCR, playback from another, and view the multiscreen at the same time.

A triplex multiplexer has all the features of the full duplex but the multiscreen output can be substituted for a third VCR if desired.

As discussed above in low cost encoders this is basically what an analog system is comprised of. The decoders can either be analog (merely the tape playback or utilizing the JVC digital VCR) or a low or high cost decoder.

Dave System

Summary

Hopefully now you have a working knowledge of the basic concepts used in the design and construction of multiplexers. Since the product offering is so large and new entries every month cost, features and performance are increasing very quickly. Purchase the best unit to meet your CCTV needs and not some specification that is usually written to meet sales goals rather than requirements. As a reminder the VCR is as important as the multiplexer so make the decision together to optimize system performance.