After weeks of working on this page on and off, spending many hours, I have a deeper knowledge of T1 lines than I thought I would when I started this page. It is certainly a complex topic, and I hope this page is useful and informative.
A while ago, I went down the rabbit hole of trying to untangle T1 line build-out and equalization. Information on anything related to T1 circuits seems to be mired in unclear and sometimes contradictory jargon and terminology making it difficult to get to the bottom of things at times. This page is an effort to get to the bottom of T1 line build-out and do so in a well defined and consistent manner while referencing a variety of sources for the content on the page.
Once we build the foundation of what line build-out and equalization are, we'll discuss where it is used throughout a traditional repeatered T1 span and why. Along the way, we'll have to explore some other technical topics relevant to T1 signals.
In order to make sense of line build-out and equalization, we'll need to discuss some basic principles of how signals propagate in transmission lines. We'll talk about signal power levels, transmitters, receivers, loss and distortion in cables, equalization, line build-out, and crosstalk.
I will not go into all of the details of power and units of power. Instead, I will simply define the basics of T1 power levels as I will be using them in this document: 0dB in relation to a T1 signal is defined as a 3V pulse amplitude.
Power levels greater than 0dB of course have a higher pulse amplitude and levels lower than 0dB (negative values) of course have a lower pulse amplitude.
A 0dB T1 signal also has a standard pulse shape specified in various standards such as ANSI T1.403. We'll discuss in later sections why the shape is specific to a 0dB signal.
T1 transmitters are the circuitry that generate T1 pulses. They have to be designed to produce pulses of the correct shape and amplitude for the application.
T1 receivers are the circuitry responsible for receiving and processing T1 pulses. There are some key characteristics that need to be discussed relating to T1 receivers.
The first is sensitivity: The lowest signal level that a T1 receiver can properly receive and interpret is known as the sensitivity of the receiver. A receiver with a lower sensitivity can successfully process a lower received signal level. This can get confusing, as a lower sensitivity means a more sensitive receiver, and a higher sensitivity means a less sensitive receiver.
Designs of receivers vary, but suffice it to say that it is more difficult and costly to implement a more sensitive receiver.
The other important parameter is the maximum signal that the receiver can successfully receive and interpret. This is called the saturation point. At signal levels greater than this point, the receiver performance suffers. In the extreme case, damage can even result.
Receivers with higher saturation points can again be more difficult and costly to implement.
In other words, for the sake of complexity and cost, it makes sense to use the least sensitive receiver you can get away with, and with the lowest saturation point.
Cable loss is fairly straightforward on the surface. As a signal propagates through the cable, power is lost to various effects (conduction losses and dielectric losses to name a couple). For a cable that is uniform along its length, the loss can be uniformly attributed to its length, and the loss can be rated per unit length e.g. dB per 1000ft.
Simple loss decreases the signal level without affecting the shape of the signal.
Where T1 signals are concerned, most of the signal spectrum is centered around 772kHz. As such, loss is measured at 772kHz when rating loss for T1 signals.
Distortion is where things get more complicated. There are multiple types of signal distortion, but the general idea is that any factor which changes the shape or characteristic of the signal is distortion.
There are nonlinear forms of distortion, but for cables, the distortion we are concerned with is linear.
There are two main aspects that contribute to distortion: The first is that the cable loss actually depends on frequency, with higher frequencies being subject to more loss.
The second is that cables introduce delay as the signal propagates along them, and that the speed of propagation (and thus the delay) also depends on frequency. In the case of T1 signals, we can consider that the low frequency components of the signal are delayed more than the high frequency components.
The differences in loss and delay for different frequencies are what cause distortion to the pulse shape. Since loss and delay are both tied to cable length, distortion is as well.
The loss and distortion applied to a signal by a cable is ultimately dependent on the characteristics of the cable itself. In order for any signal transmission system to account for the two, the characteristics of the cabling used must be understood.
To counteract distortion applied by cable, we use a process called equalization. In essence, equalization is an attempt to perform the inverse of the distortion applied by the cabling. The desired end result would ideally be undistorted, but it may still be attenuated by loss.
Equalization networks built from passive components will always have loss since they cannot add more power. However, amplifiers can be used in conjunction with, or incorporated into, the equalization if counteracting the loss is desired.
It should be noted that there are two general types of equalization: pre-equalization and post-equalization.
The two types depend on the simple principle that the effects of equalization and the cable are commutative. It does not matter if I apply the effects of the cable and then the equalization (post-equalization), or if I apply the effects of the equalization and then the cable (pre-equalization). I can even apply part of the equalization before the cable, and part of it after. The result will be the same as long as all components are used within their ratings.
Equalization must be designed around a particular type and length of cable with particular characteristics. If the characteristics of the cable vary too much from the design, the performance may become unsatisfactory. In the context of T1, equalization exists for ABAM cable, high-capacitance plastic insulated cable, low-capacitance MAT or ICOT cable, and 'LOCAP' cable. While multiple wire gauges of cable exist with differing loss within each of those groups, the bell system apparently did not think that wire gauge variation warranted different equalization.
Equalization in the context of T1 is often specified for a cable length that it compensates e.g. '0-220ft. of ABAM' but can also be thought of with respect to cable loss e.g. '31dB of high-capacitance plastic insulated cable.'
Line build-out networks are used to simulate the loss and distortion of cable.
You're probably wondering why we would want such a device when we just spent the last section discussing how to counteract cable loss and distortion. There's a few reasons.
The first has to do with equalization networks. As discussed, equalization has to be designed around the characteristic of the distortion it is to counteract. For example, if I were to design equalization to remove the distortion caused by 5000ft of plastic insulated twisted pair cable, then that equalization network is only good for that cable. If I were to attempt to use that equalization network with 3000ft of the same cable, the performance would not be as good and would likely result in a distorted signal.
If I add a line build-out network that simulates 2000ft of that cable, I can artifically 'build out' the line to 5000ft thus providing the correctly distorted signal (as odd as that sounds) to the input of the equalization network.
Another reason goes back to receiver saturation. If a receiver is designed for a signal with a nominal power of -31dB and a saturation point of -27dB, and I provide it a signal of -16dB, it won't work properly. I can use a line build-out network to add the appropriate 15dB of cable loss to meet the nominal -31dB input conditions.
Line build-out networks in regards to T1 lines are often rated in terms of dB of loss at 772kHz and are also designed to match a specific cable type. The most common is made to match pulp or plastic insulated 'high capacitance' cable, but some T1 lines were designed using low-capacitance cable which may require different line build-out to match.
Crosstalk is the tendency of one signal in a cable pair to be influenced by the signals of other cable pairs. It is important to note that crosstalk is a reciprocal phenomenon: any signal that crosstalks to another signal also receives crosstalk back from that other signal. Thus, it is important to think not only of how other signals crosstalk to one signal, but also how that one signal crosstalks to the others.
The amount of crosstalk is the proportion of an interfering signal in another pair that makes its way into the pair that is being interfered with. Crosstalk depends on cable construction and length. Generally, it is expressed in units of dB, or sometimes dB per unit length with regards to a specific type of cable.
There are different ways to measure crosstalk, but in general crosstalk becomes a problem if there is enough signal degradation to cause errors. The degradation that results from crosstalk depends on the amount of crosstalk, the strength of the interfering signal, and the strength of the signal which we are trying to receive. In summary, a stronger signal is degraded less, more crosstalk in a cable causes more degradation, and stronger interfering signals cause more degradation.
Now we'll start to look at the physical arrangement of T1 lines and T1 equipment to see where we actually use line build-out and equalization, along with discussion of why it is used in those locations.
We'll break things down into a few pieces which fit together to form the whole picture. There are many possible arrangements, which is why I'm breaking things down the way that I am. These pieces fit together to form many of those possible arrangments.
For these sections, I'll be referencing 235-200-100 (T1 digital line general description, in particular figure 23) and 314-645-100 (DS1 digital service description, in particular figure 6). I've reproduced those figures here for your convenience.
I'll be using some of my own figures based on these two to highlight the relevant parts and leave out the irrelevant ones.
Starting with the basics, a T1 line uses repeaters every so often to regenerate the digital signal. In this way, the cable losses are overcome allowing T1 trunk lines to traverse great distances.
In this diagram, the repeaters are denoted as triangles and the cable pairs used for the transmission of the T1 signals are denoted by the lines between them. The input of the repeater is the line entering the side, while the output of the repeater is the line leaving the corner.
Each T1 line requires one pair in each direction for full duplex operation (a 4 wire circuit) and thus two sets of repeaters are shown operating on two distinct pairs in each direction.
The dotted line in the middle denotes that more or less cable sections and repeaters may be present than the 4 repeaters shown here in each direction.
The length of cable allowed between the repeaters depends on the cable characteristics, and is really more of a function of loss. 365-200-100 states that the loss allowed between repeaters is 31dB which corresponds to roughly 6000 ft. of 22 AWG high capacitance pulp cable at 772kHz (for more info, see 855-371-101 table X). It also states that end sections (such as the first or last section bordering a CO) are purposely built shorter to account for the electrical noise produced from CO equipment.
The loss of one cable section is labeled. The output of a line repeater is nominally a 0dB signal, the cable section can be up to 31dB of nominal loss, and thus the signal received at the repeater can be as low as -31dB.
While we're at it, we should talk about what's in one of those T1 repeaters. For that, we refer to 365-200-101 figure 22.
In this diagram, we see that one of the first elements the received T1 signal enters is a line build-out network. As explained earlier, the line build-out network here is used to artificially 'build out' the length of the line to the proper amount of shaped loss (31dB at 772kHz for T1, with the shape corresponding to the type of cable in use, usually high-capacitance twisted pair cable.)
By using the appropriate line build-out network, we are able to ensure the signal is shaped the way it needs to be before it enters the amplifier section. While it's not specifically indicated in the diagram, 365-200-101 explains that the amplifier gain is shaped to provide line equalization. So here, we're looking at the element which provides line post-equalization as well. The amplifier is designed to compensate a nominal loss of 31dB and a signal with more or less loss will not be optimally processed.
At that point, the T1 repeater recovers the clock from the output of the equalizing amplifier and uses the clock along with the pulses themselves to appropriately trigger the output drivers to create the new pulses at full 0dB amplitude for the next section of cable.
T1 lines can pass through intermediate offices where they are simply repeated before passing through. A basic diagram of this is shown below. Two T1 lines are illustrated, but there could be more or less, and there could be more than just two neighboring offices involved.
Note that I am not showing the power feeding arrangement or test jacks here, since it isn't relevant to this discussion.
The general path of a T1 signal through the office in this case is as follows:
We can see in this example how line build-out networks are used to artificially add loss to bring the end section total loss to the nominal 31dB so that the repeaters can function properly. Both directions are more or less symmetric in how they traverse an intermediate office.
It's also worth noting that while I show the signal level being as high as -7.5dB by the time it makes it to the first line repeater outside of the office, this would effectively mean the repeater is right next to the office. In reality, there is generally some distance between the office and the first set of repeaters, so the signal would not be this strong.
A terminal office is one where a T1 line ends. Here, it will be connected into some kind of terminal equipment, such as a channel bank or a digital switch.
The diagram below illustrates this. You will notice some similarity to an intermediate office: in general, similar equipment is involved and, in fact, terminal offices for some T1 lines may be intermediate offices for other T1 lines.
Again, I have left out power feeding and test jack arrangements as they're not relevant to line build-out and equalization.
Here, the path of the T1 line in each direction is not symmetric, so we will go through each direction separately. In the direction of the signal being received by the terminal equipment:
That isn't all that different from what we've seen before in the intermediate office, except that the output of the receiving office repeater is connected to the terminal equipment.
In the direction of the signal being transmitted by the terminal equipment:
Much of this is similar to things we've already seen, with the exception of the equalization network present inside the terminal equipment, which warrants a closer look. This network is used to compensate for the cable installed between the terminal equipment and the office repeater bay. Since it is a passive network, it can only perform the equalization by adding loss, with the goal being that the combined loss of the cable and the network produces a 0dB signal by the time it makes it to the office repeater bay.
The issue is that we can't start with a 0dB signal, add loss, and then end up with a 0dB signal at the other end. For that reason, the terminal equipment actually outputs a higher amplitude, more like 6dB (6V peaks rather than 3V) to compensate for the additional combined loss of the cable (0 to 3dB) and equalization network (6dB to 3dB).
The equalization network cannot have 0dB of loss at 772kHz because it must have less loss at higher frequencies in order to accomplish equalization of the line (which will have more loss at higher frequencies than at 772kHz). For that reason, the terminal equipment outputs more than just 3dB higher signal, to allow for the additional loss budget needed to produce the required response in the equalization network.
Where a T1 line terminates at a customer premise, things look a little different.
We see some new pieces here.
The DS-1 connector mostly provides surge protection, power termination/looping, and maintenance loopback functions. It is the last piece before the network interface (NI) transitions from network to customer. Sometimes the DS-1 connector is referred to as a smartjack.
The CSU incorporates line build-out and equalization functions for each side, acts as a repeater, and enforces T1 facility requirements on the signal transmitted from the customer terminal equipment (e.g. 1s density, no bipolar violations, etc.).
It is important to note that there are many different manufacturers of CSU, and as such their internal designs vary widely. For the sake of this example, I'm going to depict the internals of the CSU a lot like I have been depicting the internals of T1 repeaters.
In the direction of the T1 signal being received at the customer premises:
In the direction of the T1 signal being transmitted by the customer terminal equipment:
Now that we've seen the arrangement of the customer premises, we will take a look at the T1 line end section. The end section near the customer poses a couple unique challenges that we don't see near the CO. Below is a simple diagram of a common situation, based on figure H.1 of of ANSI T1.403.
The figure illustrates two T1 circuits, circuits A and B, which run in a feeder cable together. At some point, some distance away from the preceding line repeater, circuit A and circuit B branch off from the feeder in two different directions. This is called a non-repeatered route junction (NRRJ).
Where the T1 signals for circuit A and B are repeated at repeater location 1, the signal level is nominally 0dB at the output of the repeater and both circuits experience the same loss (loss 2) up to the NRRJ before they branch off. Then their individual loss before the customer premise may differ, and are illustrated as loss A1 and B1. By the time the signals of circuit A and B reach the customer equipment, they may be different signal levels due to the asymmetric loss.
Ultimately, this doesn't cause any issues. We simply build out the loss of the line, if needed, and everything functions as normal.
Let's have a look at the other direction. Circuit A and B are transmitted and first traverse each through their individual loss A1 and B1. As these losses may not be equal, each circuit has now undergone different loss by the time they combine back into a single cable at the NRRJ. This is where problems may arise.
To understand why, we need to remember back to the discussion of crosstalk. Suppose, for the moment, that we have the following conditions:
Under these conditions, we would see that the signal level of circuits A and B at the NRRJ would be -12dB and -3dB respectively. The large difference in signal level here makes circuit A susceptible to far end crosstalk (FEXT) in the remaining section of cable between the NRRJ and repeater location 1.
Looking back at our diagram of the customer premise, we can see that we have a line build-out in the transmit path. Using this to make some adjustments, we can come up with some altered parameters:
Using these new parameters, circuit B has an additional 7.5dB of loss due to line build-out. Now, the signal levels of circuits A and B at the NRRJ would be -12dB and -10.5dB, much closer than the 9dB difference we had before. Under these conditions, the far-end crosstalk is much less significant and we prevented the issues by setting the line build-out of the CSU to compensate for the lower loss of circuit B.
ANSI T1.403 states that the difference between any two signals at an NRRJ should be less than 7.5dB in order to prevent these issues. In practice, variation in customer wiring as an unknown variable can make this difficult to achieve, as T1.403 allows the customer to use between 0 and 5.5dB of wiring (or more, if they are knowledgable in line build-out settings and can compensate appropriately by making their own adjustments).
So far, we've been focussed on T1 lines that span long distances from one building to another. However, there's still one more case to look at that isn't covered: T1 lines which stay entirely within one building with terminal equipment nearby to each other.
This sort of scenario might be present if two telephone switches located near each other are to be joined with a tie trunk, as one possible example.
Each direction is principally symmetric in structure (although cable losses may differ) and appears as follows:
Following this example, we can start to see why it is beneficial to establish the cross connect field as the 0dB point.
The cross connect field is designed to be the place where changes are made in what is connected to what. The wiring between this field and the equipment is more or less permanent, but the connections within it may not be.
Since the cable loss is associated mostly with that permanent wiring rather than the cross connections themeselves, if we establish every line equalization to produce identical 0dB signals at the cross connect field, then we do not have to worry about adjusting equalizations or build-outs when we make changes to the cross connections. Any cross connection, by design, will be connecting a 0dB T1 signal to another cable terminal that is already setup to accept a 0dB T1 signal.
We now have all the pieces to put together a complete picture.
We've gone through a lot of building blocks and seen line build-out and equalization applied at various points inside them. We can start to form a general picture of how all of the elements work together. The basic principles are as follows:
That is everything boiled down to it's essence. I'm going to go into a little bit more detail on each of these 3 points, just to reiterate the finer details.
Remember back to the section on T1 receivers. T1 receivers have a minimum signal they can reliably receive, as well as a maximum signal. This is due to the design of the receiver and depends on a number of factors such as method of equalization, design of signal input amplifiers, etc. Additionally, receiver post-equalization is designed to work with a signal which has been shaped by the loss of the cable and works optimally when the pulses being equalized are the expected level and shape.
As a brief summary, early T1 repeaters were designed to work with an input signal in the range of -31dB +/- 4dB. T1 terminal equipment supported a similar range, designed to work with signals in the nominal range of 0 to -3dB (with some margin, of course, for tolerances of equipment and design).
T1 repeaters were usually designed to output 0dB T1 signals.
Terminal equipment had higher output signal levels designed to overcome the loss of the cable used to connect it to the office repeater bay. The output would be coupled with line equalization networks such that the signal level and pulse shape was nominally 0dB with the standard shape by the time it reached the office repeater bay in order to allow handling of the signal as if it were produced by a repeater.
In any case, by the time the signal arrived at the next receiver, it needed to match the expected levels and pulse shape for that receiver. If it was too high of a signal level, a line build-out network could be applied somewhere in the signal path (either at the transmitter, at the receiver, or a combination of both) to put the signal level and pulse shape into the correct range.
Line build-out is also used to aid in control of crosstalk. Where T1 signals leave a CO, they are all at similar signal level and as such, the crosstalk from one signal to another is not significant. However, where multiple customer signals join together, they may be at differing signal levels which can allow crosstalk to become significant.
In these instances, line build-out at the customer transmit end is used to reduce the signal level as needed to keep adjacent T1 signals at similar power levels in order to control the impact of crosstalk.
It is useful to establish a standard signal level present at a cross connection field. This allows cross connections to be made without requiring thought about what signal level a given pair has, and how line build-out might need to be adjusted as a result. Instead, permanent wiring has line equalization applied to ensure that its effect is canceled, and the result is that every signal arriving at a cross connect field has the same signal level.
The standard level to be used is 0dB (with corresponding nominal pulse shape).
One way to ensure this requirement is met, is to ensure that T1 transmitters are close enough to the cross connect field that the cable loss is negligible. This is a good strategy for equipment which can only output a 0dB signal level.
The other way is to transmit a higher level signal which has been pre-equalized to compensate for the cable loss between the transmitter and the cross connect field, such that the signal which appears at the cross connect field is 0dB and has the correct pulse shape.
With all of the theory out of the way, we'll take a look at some common themes seen among line build-out and equalization settings.
The most common line build-out settings on equipment for T1 will be the ANSI/FCC standard settings: 0dB, 7.5dB, 15dB, and 22.5dB. These settings may be software controlled or set by switches.
On my Cisco ISRs, one of my Adtran CSU Aces, and my LARSCOM CSU/DSUs, the line build-out is set through software configuration. In contrast, on my Kentrox DSU/CSU and my other Adtran CSU Ace, the line build-out settings are controlled via dip switches. As an odd case, the external 120A2 CSUs for my Definity PBX have line build-out controlled by software on the PBX.
The case is similar for line equalization settings. The commonly seen values in my experience are 0-133ft, 133-266ft, 266-399ft, 399-533ft, and 533-655ft. However, these settings do vary and I have seen different ones referenced in datasheets and documentation regularly. Again, these settings may be software controlled or set by switches.
On my Cisco ISRs, one of my Adtran CSU Aces, my LARSCOM CSU/DSUs, my Definity PBX, and my Meridian 1 PBX, the line equalization is set through software. On my other Adtran CSU Ace, and on my SL-100, the line equalization is set through dip switches.
In general, if you see settings rated in feet of cable (or other distance) these are line equalization type settings. They represent signal levels >0dB which have been precompensated to account for cable loss.
On the other hand, line build-out values are rated in dB, and correspond to an artificial loss added to the output, making the resulting output 0dB or lower.
The proper setting for line equalization is to match the length of 22AWG ABAM cable between the output of the device and the cross connect field. If you are using a different type of cable, then you will need to match the loss (at 772kHz) against the equivalent length of 22AWG ABAM to determine the correct equalization setting.
If you are not using a cross connect field, you should still ensure that the line equalization setting is configured to produce a signal in the range of 0dB to -3dB at the next receiver (noting that the lowest length line equalization setting will be close to 0dB signal level output). Generally, it is acceptable to set the line equalization based on the length of cable between transmitter and the receiver on the far end. However, if the far end is farther away than the highest line equalization setting, you can use the highest setting and as long as the far-end receiver is sensitive enough for the signal level that it receives, this will work fine.
In that case, as a general rule of thumb, your signal should be able to reliably transit twice the distance of the line equalization setting (again noting these are usually related to 22AWG ABAM cable, and may require adjustment if you're using other cables based on loss equivalence at 772kHz).
For line build-out settings, generally the phone company provides the proper setting to use based on the design of their cable plant. In some cases, when using T1 lines that are delivered over other transport such as HDSL or fiber optic cable, the actual electrical T1 circuit only begins at the transport device (e.g. H2TUR or fiber mux) and in those cases, the correct setting is often 0dB line build-out.
If the phone company has advised you to use a line build-out setting of more than 0dB, then you have additional freedom in how long your inside cable can be. In general, the cable between the network interface and your CSU can be 0 to 5.5dB under ANSI T1.403. If you use more cable than 5.5dB, you can adjust your line build-out setting to compensate for the additional loss by decreasing the line build-out by the appropriate amount.
If you are engineering your own T1 line, you will need to consider a number of factors relating to cable loss and crosstalk considerations such as those detailed in 855-351-101.
However, if you're dealing with a single T1 line and using relatively modern equipment, you can usually just set the line build-out to 0dB. In some cases, where the cable loss is low and older equipment is in use, the signal level could be too high for proper reception by the far-end receiver (although most modern receivers do not have this issue). In those cases, you may change the line build-out setting to 7.5dB or even 15dB.
The T1 technology I described up until now is based on a generally earlier generation of equipment than is available today. I chose this approach because it allowed a more detailed look at the original technology that T1 was designed around, and the principles used within without overcomplicating the explanations with any more details than I already had.
Over the decades, many improvements have been made as better techniques were discovered and invented, and as cost-effective technology became available to implement them. I will list some of these improvements here so that they are known to the reader.
While early T1 repeaters used manually chosen and installed line build-out networks, later T1 repeaters contained automatic line build-out circuits. These circuits would automatically detect the signal level and adjust the line build-out applied to achieve the correct level and shape automatically. In these repeaters, the equalization applied to the signal was still fixed, relying on this adjustable line build-out to achieve proper function.
The early repeaters that used fixed equalization networks were designed around a -31dB +/-4dB signal. They could be adjusted to accept a higher signal level by manually using a line build-out network. The advantage of the new design did away with manually choosing a line build-out network and also allowed the T1 receiver to automatically compensate for variable line loss (as would occur due to temperature swings). The new design supported signal levels ranging from -7.5dB to -35dB without manual settings or adjustments.
Still later, T1 receivers began to implement variable equalization. Rather than apply loss using a fixed equalization network, many receivers today adjust the equalization curve instead. T1 receivers such as the Intel LXT310 designed for long-haul applications commonly have input ranges down to -35dB and up to 0dB.
Early T1 transmitters in repeaters and terminal equipment had fixed level outputs and used passive equalization networks and line build-out networks to adjust the signal levels. Later, T1 transmitters began to control the pulse shape and amplitude electronically. Today, modern T1 transmitters electronically support signal outputs which are pre-equalized to produce 0dB after up to 655ft of ABAM cable (3dB of loss) by electronically increasing the signal level and shaping the pulse. T1 transmitters intended for long-haul applications such as the Intel LXT310 often support electronic control to produce 0dB, -7.5dB, -15dB, and -22.5dB output signal levels rather than use physical line build-out networks. Many long haul capable transmitters, such as the Renesas 82P2281, support the short-haul equalization settings as well (0 to 655ft of ABAM cable). Some T1 transmitters even support software programmability for custom line build-out or equalization settings using arbitrary waveform generation, like the Renesas 82P2281.
Some CSU type devices have an automatic line build-out setting which can automatically adjust the transmit line build-out. I can't find a lot of detail on this, but it is supported by the Adtran CSU Ace line of CSUs, for example. I assume that the CSU measures the signal at its input, and adjusts the output line build-out to match, but this is speculation.
Due to the complexity of the topic, I've left out a lot of detail in an effort to stick to the points that I think are most important to the discussion. There are a few considerations worth noting which I did not mention that I will note here in passing.
In a number of places within this page, I mention 'nominal' loss. I make this distinction because loss can change with a few factors, primarily with cable temperature. There is a lot of very detailed information on engineering T1 lines in 855-351-101, including information on how to appropriately consider temperature variations of cable. The general principle of course is that the system must be designed such that normal variations in cable loss are accounted for.
In a number of places, I reference cable lengths which may be based on the '6000ft' number that is thrown around a lot discussing T1 cable section length between repeaters. As I briefly touched on, cable loss is dependent on the exact cable type, and this distance seems to correspond to the correct amount of loss for 22AWG pulp insulated cable somewhere around 140 degrees Fahrenheit. Again, there is a lot of very detailed information on engineering T1 lines in 855-351-101. Suffice it to say that these days there are many cables which have better performance than the old pulp-insulated ones, not to mention lower capacitance cables which (when used with T1 repeaters that implement the correct line buildout and equalization) can go substantially farther than 6000ft.
I briefly mentioned far end crosstalk (FEXT) which is where signals transiting the same cable in the same direction crosstalk to each other to produce signal degradation at the far end. There is another type of crosstalk known as near end crosstalk (NEXT) which is where near end transmitters crosstalk to near end received signals.
NEXT can be a real problem if you don't consider it at all. For short distances, it's not as much of a concern. However, for longer distances, it must be accounted for to prevent errors.
While lots of information appears to be present in 855-351-101, there are a couple simple solutions that are practical for typical small-scale T1 applications.
The first is to use cabling that separates the send and receive pair groups with shielding such as metal foil or braided wire shielding. Cable with individually shielded pairs is available which would meet this requirement.
The other is to use two cables, one for the send pairs and the other for the receive pairs. Depending on the physical distance between the two cables, shielding of each cable as a whole may be useful, but would not be required at the individual pair level since each cable only carries one direction of transmission.
In either case, the general idea is to decrease crosstalk between send and receive directions by either adding shielding or increasing physical separation.
The following is a list of references with links that are used in this page as supporting information.