High Definition Time Domain Crosstalk (HDTDX) is used to troubleshoot links that fail Near-End-Crosstalk (NEXT). Measurements for certification are made in the frequency domain (MHz). However, for diagnostic purposes, this is of little use. We need to be able to see where the crosstalk is happening, so the failing link can be fixed / replaced. This is where time domain data comes into play. Time Domain Reflectometry is the analysis of a conductor (wire, cable, or fiber optic) by sending a pulsed signal into the conductor, and then examining the reflection of that pulse. Hence frequency domain is in Hz, time domain is in meters (ft). Using a time domain measurement will allow us to see the sources of crosstalk and troubleshoot the links. HDTDX is a very good example of a time domain measurement.
What is a good termination?
We would prefer you not find out how useful HDTDX can be! You will be looking at re-terminating connectors and in some cases laying new cable runs profit margins are compromised and time penalties are inevitable. Fluke Networks vision of a good termination is not always that of the installer. So before we explain how to use HDTDX, lets set the standard for termination.
From the standard ANSI/TIA-568-C.0
It is essential to maintain the design performance of connecting hardware when terminated to a balanced twisted-pair cable, and this shall be achieved by terminating the appropriate connecting hardware for that balanced twisted-pair cable in accordance with the connecting hardware manufacturer’s instructions. Where no connecting hardware manufacturer’s instructions exist, then the cable geometry shall be maintained as close as possible to the connecting hardware, and its cable termination points and the maximum pair un-twist for the balanced twisted-pair cable termination shall be in accordance with table 1.
Table 1 states 13 mm (0.5 in) for Category 5e, 6 and 6A.
Figure 1. Correctly terminated connector
In reality, if you have up to 13 mm (0.5 Inch) untwist, the performance of the cabling system will be reduced resulting in marginal passes and / or failures. You should be aiming to achieve no untwist in the pairs for Category 6, as shown in Figure 1. Our experience shows that allowing 13 mm (0.5 inch) untwist will result in NEXT / Return Loss failures for Category 6 and reduced margins for Category 5e.
From Cenelec EN50174 Part 2
The connecting hardware used for copper cabling shall be installed to provide minimal signal impairment by preserving wire pair twists as closely as possible to the point of mechanical termination (by not changing the original twist), see EN 50173. In addition only a minimum of cable jacket shall be removed.
Will adding extra twists degrade the performance?
No, despite what it says above - by not changing the original twist. You will always find a pair with a loose twist and its not easy to maintain the twist to the punch down contact. Adding an extra twist in that pair will not only improve NEXT, but Return Loss as well. Try it for yourself. Particularly important for block wiring with SYSTIMAXs 110 IDC and ADC's Ultim8. Don't overdo it! Be careful that you dont cause one wire in the pair to wrap around the other wire while the latter remains relatively straight. Youll introduce NEXT and may cause a failure. As always, you should adhere to the manufacturers installation instructions.
Looking after the cable
The manufacturers go to great lengths to control the geometry of the individual pairs within a cable.
For Category 6, it is common to use a separator as shown below. However, if you create large bundles, the cables at the bottom of the bundle will inevitably be compressed, also shown below.
Figure 2. Affects of crushing a cable
Note how the distance is reduced between the brown and orange pair under compression. The cross member here is of a helixical design to reduce such compression effects. However, if the cable is compressed over a significant distance, it will have a significant impact on the test results and will be visible using HDTDX. Crushing the cable at just one point will normally have little effect this is not a license for abusing the cable in one or two places! We have all tried tying a knot in the cable. Its only when the cable is abused over the length of a cable do you see the NEXT and Return Loss margins degrade.
In large successful installations, we have observed documentation stating that no cable bundle shall be greater than 24 or 48 cables and that they are stacked no more than 3 high. This comes at a price with the need for wider cable trays and multiple cable routes. The price comes even higher if you decide to load the cable trays to the point of overflow. Damaged cable cannot be repaired, it can only be replaced.
Before you start each day, grab a box of cable, run 90 meters and terminate with the connectors you are going to use. Run the Autotest. This will give you an increased confidence that when installed, the links will PASS. Plus, minimal waste, you can still use this cable. Good test of termination practices for the installers as well. 90 meters is preferred because this will better show if there is an issue with the cable.
The HDTDX time domain plot needs to be obtained along with the NEXT frequency plot. In cases of dispute, it is better to download the result(s) to a PC. For the DSP-4100 and DSP-4300 CableAnalyzers, the HDTDX plot data is stored automatically for links that fail the Autotest. However, with the DSP-4000, you can get the graphical information of the last link measurement only. To do this, make your measurement using the DSP-4000. Connect the DSP-4000 to a PC with LinkWare running.
In LinkWare, click on Utilities > DSP-4x00 CableAnalyzer > Detailed Test Data
Figure 3. Accessing Detailed Test Data from a DSP-4000
Once you have saved the CSV file, you can open it by clicking on File > Open and change the file type to CSV. CSV is an acronym for comma separated value; it is a common format used by computers since the early 1980s. If the cable is compliant and has been laid without damage to it, you will see only peaks in the HDTDX plot at the point of the connectors. However, it is not unusual to see small peaks along the cable. These should be few in number and certainly less than 5%.
How can I view the HDTDX plot?
In LinkWare, open the file that contains the failure(s). An example is shown below. By clicking on the column header Headroom, LinkWare will put the worst result at the top and the best result at the bottom of the summary screen.
Figure 4. LinkWare summary screen
This example of an installation was submitted by a customer to the Fluke Networks Technical Assistance Center for evaluation. The following HDTDX traces are taken from the submitted results with the permission of the customer. However, all references to the customer have been removed. It is a policy of the Technical Assistance Center that all data is treated in the strictest confidence.
Now that you have your file open, double click on one of the test results you wish to analyze. A new window will pop up. Click on the NEXT tab as shown below to see which pair has the worst margin in dB. Note this pair, in this case 36-45. If the pairs are not the same for the main and remote, note down both. You will then need to look at the HDTDX responses for both.
Figure 5. Viewing NEXT results in LinkWare
Now click on the Pair Data tab. Then click on HDTDX Analyzer Plot to view the HDTDX graph.
Figure 6. Viewing HDTDX plots in LinkWare
In this example of a bad / damaged cable, using the HDTDX conversion chart is pointless. Its clear that the cable is bad and needs replacing yes replacing! Re-terminating or changing the connectors for a different vendor will NOT help. Our advice is that if you are unfortunate to see such a trace, you SHOULD contact your distributor or cable supplier immediately.
Figure 7. Bad / damaged cable viewed using HDTDX
To look at just one pair as above, press the Tab key on your keyboard. Whilst this is an extreme example, the trace found in Figure 8 is also of concern and can be deemed unacceptable due to the level of activity in the cable. This link resulted in a NEXT FAIL. Re-terminating the connectors may turn it into a marginal PASS, but you should take the hint given in this trace; this is a warning that there are issues with the installation.
Figure 8. Bad / damaged cable viewed using HDTDX
Lets now look at another link in Figure 9 which has also failed NEXT. There is no activity in the cable on this link, so well look to the connectors. We assume the correct Personality Module is being used in this example.
Figure 9. Poorly terminated connector viewed using HDTDXPoorly terminated connector viewed using HDTDX
In this example, we see a peak of 48% at 1 meter. In this example, our test lead is 1 meter, so we know this peak represents the connection to the patch panel. Using the tables below, we can see that, 48% represents 44.3 dB @ 100 MHz, thats almost 10 dB worse than the Category 6 connecting hardware standard (54 dB). The remote connection at 36 % is better although does no meet the desired 29-30%. As a rule, less than 35% will result in an acceptable pass for the link under test. Values highlighted in yellow represent the standard requirement. Solution for this example: re-terminate the patch panel.
There are four tables in total to choose from. Please ensure you are looking at the correct table.
Lets look at another example (Figure 10), a Channel test. This trace is a little more confusing at first. We established that the HDTDX spike should ideally be under 30% and never over 40% for connectors.
Looking at the link below, we see NEXT passes with a good margin of 5.7 dB. However, you can see two large spikes in excess of 40% at 0 and 34 m. How can this be a pass?
Figure 10. Channel measurement viewed using HDTDX
The second spike at 2 meters is the patch panel; the first spike is the connection to the tester. The standards require us to remove the effect of the connection to the tester. The DSP-4x00 makes the measurement as above, and then converts it into the frequency domain (MHz) as required for certification. Before it does that, the first and last spikes (connection to the tester) are removed; then the conversion to the frequency domain is applied. Since we are mathematically removing these connections, were not too interested in them.
Given the spikes exceed 40%, doesnt that suggest the adapters for the DSP-4x00 are not Category 6 compliant? Using this measurement technique they do not need to be. The reason is Far End Crosstalk (FEXT). You cannot remove FEXT from the measurement no one can. So what we have done is created a channel adapter that has very good FEXT properties but not so good NEXT properties which we can cancel.
What can we learn from this trace?
Choice of patch cords is critical. Using HDTDX, you can clearly see how your cords interact with the cabling system. Dont forget, with the early Category 6 solutions, you must use vendor specific patch cords. If you dont, expect to see the spike(s) exceed 40%.
We talked about activity in the cable, but what is acceptable and not acceptable? This is best represented in the following two examples. The spikes seen at 5 and 7 meters in Figure 11 are an indication of the cable being kinked. The reality is that this will happen from time to time and we have to live with it. It is almost impossible to avoid. However, there is no reason not to inform the project manager that this has shown up. This is not the case in Figure 12. The cable has been damaged severely in six locations and the unauthorized addition of a consolidation point appears to be added at 45 meters.
Figure 11. HDTDX shows a little activity within the cable
Figure 12. HDTDX shows significant activity within the Cable
HDTDX Conversion Utility
This utility will convert the HDTDX graph into a new format, where the vertical axis represents the NEXT @ 100 MHz using the lookup tables shown in this article. You will need Microsoft® Excel 2000 or later to run this utility. Click here to download the utility. (1.5 MB) New Version 001A. With the HDTDX plot data saved as a CSV file, open the utility where you will be prompted to open the HDTDX CSV file. You may received a warning that you are about to run a macro. In order for the conversion to work, you need to click on Enable Macros. If you are unsure about doing this, please contact your IT Manager.
This utility supports the following standards:
- TIA Cat6 Channel
- TIA Cat6 Perm. Link
- TIA Cat5e Channel
- TIA Cat5e Perm. Link
- ISO11801 Channel Class E
- ISO11801 PL Max Class E
- ISO11801 Ch Class D-2002
- ISO11801 PL Max Class D
You will also need to ensure that the Numeric Format in LinkWare is set to 00.0 and not 00,0.
Below is an example conversion.
Figure 13. Example conversion.
How do I read this?
You will note from “What is HDTDX?” that the connecting hardware limit for Category 6 is 54 dB @ 100 MHz. The first spike represents the mated connection in the patch panel. The last spike represents the mated connection in the outlet. On this occasion, the news is good. Both spikes are above the minimum 54 dB requirement for Category 6 connecting hardware – and hence the worst case margin for this pair was 8.5 dB.
What happens if it falls below 54 dB? This 54 dB is not easy to achieve. The good news is that achieving this in the field is becoming possible with the very latest generation of connecting hardware in conjunction with the Personality Modules DSP-PM25 / DSP-PM06. Users need to make sure they have the very latest product from the connecting hardware manufacturer.
Here are some guidelines:
- Links less than 15 meters (50 ft) MUST have connecting hardware spikes better than 54 dB in order to achieve a PASS for the Autotest
- Links greater than 30 meters (98 ft) can have connecting hardware spikes as low as 49 dB; so long as no consolidation point is used
- Any activity in the cable will require the mated connection at each end to be better than the guidelines in 1) and 2)
Figure 14. Example conversion.
Looking at the graph above, we see no crosstalk in the patch panel but a significant amount of crosstalk at the outlet. This resulted in worst case margin of just 1.8 dB. The pair combination did not fail due to the lack of crosstalk in the patch panel. If this link were less than 15 meters (50 ft), it would most certainly fail.
Here are some reasons for such a trace:
- Category 5e outlet used
- New person / temp terminating outlets and failing to keep the strict “no untwist” requirement for Category 6
- Incorrect Personality Module – check with connecting hardware vendor
- Non component compliant connector