The camera worked for eight months, then started rebooting at dusk. The cable tested clean, the link trained at gigabit, the certification report was a pass, and the plug looked fine to everyone who glanced at it. The camera was not failing. The termination was.
There is a reason this fault survives every test a contractor normally runs, and it starts with the connector.
The Plug Was Never Meant to Be Hand-Terminated
RJ45 8P8C modular plugs were designed to be terminated onto stranded copper cable, in a factory, by a machine. That is not a purist position — it is what the connector was engineered for, and it explains why hand termination onto solid copper creates problems that care alone does not remove.
The root problem is not the pass-through style. It is that RJ45 8P8C plugs should not be terminated by hand onto bulk cable unless there is no other choice — especially not onto solid copper.
Solid copper is what permanent horizontal cabling uses, because it carries signal better over distance and holds its geometry inside a jacket. Stranded copper is what patch cords use, because it survives being bent and re-plugged. That is why the professional method is "rack to jack": solid copper from panel to jack, factory cords at both ends.
Two Termination Styles, Two Different Physics
The reason a keystone jack outperforms a hand-crimped plug is mechanical before it is electrical.
| IDC (keystone jack, patch panel) | RJ45 8P8C three-prong contact |
|---|
How it connects | Prongs cut through insulation and bite into the copper at right angles | Three prongs pierce the insulation along the side of the conductor |
|---|
Seal | Creates an air-tight seal | Does not create an air-tight seal |
|---|
Under stress | Pierces from the sides, preventing longitudinal shifting | Longitudinal stability is not nearly as good |
|---|
Both are insulation-displacement in principle. The difference is what happens when the cable moves. An IDC termination resists the conductor sliding lengthwise because the contact bites across it. A three-prong contact displaces insulation along the side, so it tolerates far less movement — and a hand-crimped plug leaves the cable room to shift at the rear of the housing, exactly where a pulled cable applies force.
That is why a factory cord, made by a machine holding tolerances no hand tool can match, has effectively zero play at either end, and why the same plug fitted by hand can be intermittently fine for months.
The Tolerance Problem Nobody Controls
This part surprises installers, because it can be nobody’s fault.
Insulated conductors have a nominal diameter and a tolerance range. The plastic coating is soft and hard to hold to a precise diameter, so the tolerance is typically ±0.05 mm, and the copper underneath has its own. Cable at the outer limit of its range is still within specification. Plug compatibility, by contrast, is a narrow window:
- A given pass-through plug accepts insulated conductors from 0.95 mm to 1.05 mm.
- A given Cat6 cable states a nominal insulated conductor diameter of 0.97 mm, with a manufacturing tolerance of ±0.04 mm.
- That cable’s lower limit is therefore 0.93 mm — and a spool that lands there is 0.02 mm below the plug’s minimum, while remaining fully in specification.
Nothing here is defective. The cable met its tolerance and the plug met its range, and the combination is still unreliable, with symptoms appearing first at higher speeds and under PoE. That is why two installers using the same plug and tool report opposite results on cable from different production runs — and why conductors that felt loose during termination are a warning, not a nuisance.
Where It Shows Up First: PoE
Data failures are inconvenient. PoE failures are expensive, because the defect that degrades a signal also drains power and generates heat.
PoE is delivered as a common-mode voltage across a pair, and current divides equally between the pair’s conductors only when their DC resistance is equal. Any difference is DC resistance unbalance, and an inconsistently terminated pair causes it directly. Seat eight conductors by hand, under time pressure, with hand fatigue setting in, and one conductor of a pair will end up seated differently from its partner — at which point the pair stops being electrically symmetrical.
The limits are specific:
- IEEE Std 802.3-2012 specifies a maximum DC resistance unbalance of 3 percent between conductors — the difference between a pair’s two conductors must not exceed 3 percent of its total DC loop resistance.
- The proposed IEEE 802.3bt standard requires the DC resistance unbalance between two pairs to be no more than the greater of 7 percent or 50 mΩ.
Exceed it and the transformer can saturate, distorting the waveform and producing bit errors, retransmits and eventually non-functioning links.
Here is the trap. Neither TIA nor IEC requires DC resistance unbalance testing as a field measurement — an omission dating to a time when no field tester could perform it. A link can therefore be certified clean and still be unfit for PoE. A cable vendor’s specification does not close the gap either, because it describes the cable, not the terminations someone made in a ceiling at four in the afternoon.
Field testing is the only way to confirm performance after installation, and it is worth specifying now that four-pair PoE makes pair-to-pair balance a failure mode in its own right.
A blunter problem sits alongside it. Copper-clad aluminum is not compliant with industry standards and does not support PoE, because its DC resistance can be 55 percent higher than solid copper of the same diameter — more heat in the cable, less voltage at the device. Some CCA even passes DC loop resistance testing on short links while still showing resistance unbalance. Both TIA and ISO/IEC require twisted-pair data cable to be 100 percent copper.
Heat Is the Second Problem, and Bundles Make It Worse
PoE raises the temperature inside the cable, and rising temperature increases insertion loss — enough to fail a channel that previously passed, or to force a shorter run. It compounds when many PoE cables share a tight bundle, and it scales with power.
PoE type | Delivered | Available to device | Pairs used |
|---|
Type 1 (802.3af) | 15.4 W | 13 W | 2 |
|---|
Type 2 / PoE Plus (802.3at) | 30 W | 25.5 W | 2 |
|---|
Type 3 (802.3bt) | 60 W | 51 W | 4 |
|---|
Type 4 (802.3bt) | 90 W | 71 W | 4 |
|---|
Above 60 W the rules become mandatory. The 2017 edition of the NEC contains ampacity tables specifying maximum ampacity for a given bundle size, conductor gauge and cable temperature rating at 30 °C ambient, and because the NEC is law, compliance is required. UL offers an alternative: after a fact-finding study into higher-power PoE in bundled cable, it introduced Limited Power (LP) certification, indicating a cable has been tested to carry PoE under worst-case installation conditions without exceeding its temperature rating.
Two cautions. LP is a certification, not a listing or rating, so unlike plenum or riser ratings it is optional — and the NEC accepts LP cable in place of the ampacity tables. And even with LP cable, poor workmanship can still cause excessive resistance unbalance, so test it for DC resistance.
Most PoE devices, including LED lights, draw well under 60 W — but nobody knows what a cable will eventually be asked to carry.
What to Do in the Field
Terminate to a jack or panel, not a plug. Keystone jacks and patch panels are the correct endpoint for solid copper. Where a male plug end is genuinely required, use a field termination plug built for solid conductors.
Buy patch cords; do not make them. Factory cords are assembled and tested by machines holding tolerances no hand tool can match, and solid copper does not survive repeated handling.
Keep both ends consistent. Do not put plugs on both ends of a run. Where a plug is unavoidable — a device housing that accepts nothing larger, common with outdoor cameras — the panel end should still be a jack and the plug should carry a strain relief boot.
Match the plug to the cable and check the fitment range. Pass-through plugs tolerate less fitment variation than standard body styles, and for Cat6A a standard load bar style plug is the better choice — pass-through designs may struggle to reach 10G.
Test the plug connection, not around it. Modular plug terminated links need the correct adapters so the final plug connection sits inside the measurement, and DC resistance unbalance belongs in the test set.
Never terminate a live passive PoE cable. Standard PoE negotiates before applying power, so a bad connection usually fails safely. Passive PoE does not, and a miswired plug on a live cable can destroy the device.
How to Specify
Write the termination into the specification, not just the cable. "Cat6 solid copper, punch-down at both ends, factory patch cords" removes the ambiguity that puts a hand-crimped plug in a ceiling.
Add DC resistance unbalance to the acceptance test. It is not required by TIA or IEC, so if it is not specified it will not be measured, and the failure surfaces after handover.
Above 60 W, choose the compliance route in advance — NEC ampacity tables or LP-certified cable — because the two imply different products.
Our patch cords are factory-assembled and tested as complete assemblies, and the communication cable behind them is built to the conductor concentricity that resistance unbalance depends on, in Cat5e and Cat6 constructions. How Ethernet cables work covers the construction underneath, and our test and inspection process is listed by parameter. If a link is intermittent and the cable has been blamed, send us the details — the plug part number, conductor diameter and termination tool usually explain it.
Frequently Asked Questions
Can I crimp a standard RJ45 plug onto solid Cat6 cable?
You can, and it may work — but it is the last option rather than normal practice, because modular plugs are designed for stranded cable terminated by machine. Terminate to a jack or panel, or use a field termination plug built for solid conductors.
Do pass-through plugs cause PoE problems?
Not specifically. The PoE failures traced to plugs come from poor fitment, and standard body styles fail the same way. Pass-through plugs simply tolerate less variation, which makes an existing mismatch more likely to surface.
My link certifies clean but the camera reboots. How?
DC resistance unbalance is not a required field test under TIA or IEC, so a link can pass certification and still be unbalanced enough to cause problems under PoE. Test for it explicitly.
The Short Version
RJ45 plugs were designed for stranded cable, terminated by machine. Hand-terminating them onto solid copper is the last option, not a method — the physics says so: IDC terminations bite across the conductor, while three-prong contacts pierce along the side and tolerate far less movement.
Then the tolerance window: plugs accept a narrow band of conductor diameters and cables are made to a wider one, so a spool at its legal minimum can fall below a plug’s legal minimum with nothing out of specification on either side.
Where it hurts is PoE. Inconsistent terminations unbalance the DC resistance of a pair, IEEE 802.3-2012 caps that at 3 percent, and 802.3bt caps pair-to-pair unbalance at 7 percent or 50 mΩ — yet TIA and IEC do not require the measurement in the field, so a clean certificate can sit on top of a link that cannot carry power.
Terminate to jacks, buy cords rather than making them, keep plugs off both ends, and put DC resistance unbalance into the acceptance test. Above 60 W, choose between NEC ampacity tables and LP-certified cable before the cable is ordered.
Sources
- trueCABLE — Terminating Pass-Through RJ45 Connectors onto Solid Copper Ethernet Cable, covering the history of solid-nosed standard 8P8C RJ45 plugs from 1995 and pass-through plugs from about 15 years ago, the three-prong contact pin construction inside an RJ45 connector that pierces insulation along the side of the insulated conductor, the statement that IDC terminations are far superior for mechanical and therefore electrical stability, that RJ45 8P8C modular connectors are actually supposed to be terminated onto stranded copper Ethernet cable at a factory by a machine, and that the root problem is not pass-through plugs but that RJ45 8P8C plugs should not be terminated by hand onto bulk cable unless there is no other choice, especially solid copper, the description of IDC prongs cutting through conductor insulation and biting into the copper from right angles creating an air-tight seal and piercing from the sides to prevent longitudinal shifting of the conductor under stress, the statement that the three-prong contact pin pierces insulation from along the side of the conductor and that longitudinal stability is not nearly as good as IDC terminations, that the termination strategy used for RJ45 8P8C connectors is susceptible to stress from conductor movement when the cable shifts at the rear of the plug housing, that the golden contact prongs do not create an air-tight seal and displace insulation along the side of the copper conductor, the rack to jack method for terminating bulk solid copper Ethernet cable meant for permanent installation, the recommendation that patch panels, keystone jacks or field termination plugs are the preferred way to terminate solid copper Ethernet for maximum performance and reliability, the instruction not to construct Ethernet patch cables with solid copper and to buy factory pre-terminated patch cables assembled and tested before purchase, the guidance that if there is no other choice the connection should be permanent, terminated to field termination plugs at both ends, and not plugged and unplugged often because solid copper conductors do not hold up to repeated handling, the single acceptable scenario of terminating a plug onto solid copper where the endpoint device will not accept any other size connector with the example of a surveillance camera water-tight housing, the recommendation to use cable strain relief boots and to keep the switch end as a keystone jack or patch panel rather than putting plugs on both ends, the guidance to use pass-through style only for Cat5e and Cat6 and not for Cat6A because reaching 10G may be difficult, deferring to the Cat6/6A standard load bar style RJ45 plug with a proper strain relief boot, the note that pass-through plugs are less tolerant of fitment problems than standard RJ45 connectors, the tolerance discussion covering insulated conductors having a nominal diameter and a stated tolerance range typically 0.05 mm plus or minus with the copper thickness also having a tolerance and variability from one production run to another able to place cable at the outside maximum or minimum while still in specification, the statement that soft plastic coatings are difficult to hold to a precise diameter, that RJ45 8P8C plugs are hard plastic with a strict compatibility or fitment range for insulated conductors, the worked example of a pass-through RJ45 plug with a strict insulated conductor compatibility range of 0.95 mm to 1.05 mm against a Cat6 cable with a stated nominal insulated conductor diameter of 0.97 mm and a manufacturing tolerance range of 0.04 mm plus or minus where a lot dropping to the minimum side now measures 0.93 mm and falls below the strict minimum by 0.02 mm, the statement that the combination may work but will more likely show issues at higher speeds or with PoE, and that neither the cable nor the plug is defective but the cable is within specification while being on the low side of acceptable, the PoE notes that issues sometimes seen with Power over Ethernet are not due to pass-through RJ45 plugs but to poor fitting RJ45 plugs generally, that a poor fitting plug affects both speed and the ability to power a camera or WiFi access point, that actual damage to the endpoint PoE device is rare because most PoE devices use the 802.3af, at or bt protocol and the switch will not supply power unless the endpoint device requests it, and that danger dramatically increases with proprietary passive PoE where power is always applied whether the end device requests it or not, with the author’s account of killing a PoE surveillance camera by terminating a plug upside down on a live cable — https://www.truecable.com/blogs/cable-academy/terminating-pass-through-rj45-connectors-onto-solid-copper-ethernet-cable-a-really-bad-idea
- Fluke Networks — DC Resistance Unbalance Testing: Easy, Low-cost Insurance For Your PoE Systems (white paper), covering the ratification of gigabit Ethernet and PoE by IEEE in 1999 and 2003, the estimated 85 percent of the installed cabling base supporting both technologies, the requirement of all four cable pairs for gigabit Ethernet bidirectional transmission against two pairs for 10/100BASE-T, PoE as phantom power applied as a common-mode voltage between two pairs, the statement that DC resistance unbalance is not required in TIA or IEC performance field testing while being specified in IEEE PoE standards, the move to four-pair PoE delivery under IEEE 802.3bt with pair to pair DC parallel resistance unbalance as an additional source of problems, the power figures of IEEE 802.3af at a maximum of 15.4 W with 13 W available over two pairs and IEEE 802.3at at 30 W with 25.5 W available and the proposed 802.3bt at 100 W, the Alternative A and Alternative B delivery methods where Alternative B delivers power over spare pairs 1 and 4 compatible with two-pair data applications and Alternative A delivers power simultaneously with data over pairs 2 and 3 compatible with both two-pair and four-pair applications including 1000BASE-T, the statement that when the resistance of each wire in the pair is equal DC resistance unbalance is zero and current is split evenly achieving common-mode current, that too much unbalance causes the potential for saturation of the transformer and can distort the waveform of Ethernet data signals causing bit errors, retransmits and even non-functioning data links, that with a four pair PoE system some DC resistance unbalance between the pairs can be tolerated but excessive unbalance stops PoE functioning, the causes of DC resistance unbalance including transformer offsets at PSE and end devices but more often poor workmanship, inconsistent terminations and subpar cable quality, the importance of consistency in individual conductor terminations and of punching down conductors to the proper IDC tower of a network jack to displace insulation and expose copper, the statement that a certain amount of force is required to seat conductors and that inexperience, hand fatigue and larger conductor gauge sizes can impact the ability to maintain consistency, that when two conductors of a pair carrying PoE are terminated inconsistently DC resistance unbalance can occur, and that using the correct termination tool helps increase termination consistency and avoid DC resistance unbalance in PoE systems, the role of precision manufacturing including selection of copper conductors and stringent controls to maintain the physical geometry of the cable, the risk of DC resistance unbalance where poor quality cable shows variations in diameter, concentricity or roundness, contour and smoothness of the copper conductors, the concern over copper coated aluminum, copper coated steel and other non-standard conductors masquerading as Category 5e or Category 6 cable, the statement that CCA cables are not compliant with industry standards and do not support PoE applications due to increased DC resistance that can be 55 percent greater than for solid copper cable of the same diameter resulting in greater heating of the cable and lower voltage at the powered device, the note that testing for DC resistance is not always enough because some CCA cables pass DC loop resistance testing for shorter links while CCA cable will typically feature DC resistance unbalance on pairs regardless of link length due to lack of consistency across conductors, the requirement of both ANSI/TIA and ISO/IEC standards that twisted pair data cable be 100 percent copper, the specification in IEEE Std 802.3-2012 of a maximum DC resistance unbalance of 3 percent between conductors meaning the difference in DC resistance between two conductors is no more than 3 percent of the total DC loop resistance of a pair, the note that TIA and IEC standards do not require DC resistance unbalance testing within a pair or between pairs as a field measurement historically because no field tester was capable of it, the requirement in the proposed IEEE 802.3bt standard that DC resistance unbalance between two pairs be no more than the greater of 7 percent or 50 mΩ, the explanation that DC resistance unbalance testing verifies that conductors in a pair have equal resistance enabling the common-mode current needed to support PoE and avoid distortion of data signals, and the DSX-5000 measurement of DC loop resistance as the sum of two conductors in a pair against DC resistance unbalance as the difference in resistance between the two conductors with DC resistance unbalance between pairs as the absolute difference in parallel resistances — https://www.flukenetworks.com/content/white-paper-dc-resistance-unbalance-testing-easy-low-cost-insurance-your-poe-systems
- ISE Magazine / Fluke Networks (Mark Mullins) — Can PoE Take the Heat, covering Type 1 PoE delivering up to 15.4 W with 13 W available and Type 2 PoE Plus delivering up to 30 W with 25.5 W available both over 2 pairs using Alternative A or Alternative B, the delivery of power in Alternative A simultaneously with data over pairs 1-2 and 3-6 and in Alternative B over spare pairs 4-5 and 7-8 with Alternative A compatible with both two-pair and four-pair applications while Alternative B is compatible only with data signals using two pairs, the proposed 802.3bt standard for 4-pair PoE including Type 3 delivering up to 60 W with 51 W available and Type 4 delivering up to 90 W with 71 W available, both delivering power over all 4 pairs simultaneously with data, the requirement in Type 1 and Type 2 PoE using Alternative A that the DC resistance of each conductor in the pair be balanced and equal with any difference referred to as DC resistance unbalance and too much unbalance distorting data signals and causing bit errors, retransmits and nonfunctioning data links, the additional concern in Type 3 and Type 4 that excessive DC resistance unbalance between multiple pairs can wreak havoc on data transmission and cause PoE to stop working, the role of both poor quality cable with variations in conductor diameter and concentricity and inconsistent terminations where individual conductors are not properly and consistently seated within IDCs as causes of DC resistance unbalance, the statement that while a DC resistance unbalance specification may appear on a vendor’s cable, field testing is really the only way to ensure DC resistance unbalance performance after installation, the heat rise discussion covering that when PoE is delivered over twisted-pair copper cabling increased temperature within the cable can increase insertion loss, which can cause a channel to fail insertion loss testing or require the length of the cable to be reduced, that heat generated by PoE is even more of an issue when multiple cables delivering PoE are together in a tight bundle and that the higher the power the greater the heat, that the National Electric Code specifies the number of cables allowed in a bundle based on conductor size and temperature rating for 60 W or higher PoE and that the TIA is developing guidelines for limiting temperature rise in a bundle, the introduction by Underwriters Laboratories of a Limited Power (LP) Certification following a fact-finding study into the effects of higher levels of PoE applied over cables in a bundle, the LP certification indicating a cable has been tested to carry PoE under worst-case installation scenarios without exceeding the temperature rating of the cable and accounting for large bundle sizes, high ambient temperatures and other environmental effects such as enclosed spaces or conduits, the statement that LP is a certification and not a listing or a rating so that unlike other UL listings or plenum or riser ratings required by NFPA 70 National Electric Code, LP-certified cable is an option and not a requirement, the note that the 2017 edition of the NEC contains new requirements addressing heat rise but only when power is greater than 60 W or Type 3, that for these cases the NEC includes ampacity tables specifying the maximum ampacity allowed for a certain cable bundle size, conductor gauge and cable temperature rating installed in an ambient temperature of 30 degrees C or 86 degrees F, that because the NEC is law complying with these ampacity tables is required, and that the NEC does allow the use of an LP-certified cable as an alternative to following the ampacity table, the observation that most PoE-enabled devices including LED lights require less than 60 W but that it is never known how much power might eventually be delivered so following ampacity tables or using LP-certified cable is a good method for future proofing along with larger diameter conductors, higher temperature ratings, shielded construction or choosing not to use cable bundles, and the statement that while DC resistance unbalance is not typically an issue in higher quality cables with a higher temperature rating or LP certification, poor workmanship can still cause too much resistance unbalance so LP cabling should still be tested for DC resistance — https://www.isemag.com/core-legacy/article/14268351/can-poe-take-the-heat
- ISE Magazine / Fluke Networks (Mark Mullins) — Modular Plug Terminated Links, within the same article, covering the growth of IP-enabled devices connecting to copper horizontal cabling including LED lights, security cameras, building automation controls and Wi-Fi access points with integrated RJ45 ports, the scenario where a typical 4-connector channel is not used especially in ceiling space where installing a faceplate is impractical, with a single patch cord in the telecommunications room and the permanent link terminated at the other end with a plug to plug directly into the device essentially eliminating the equipment cord, creating what is known as a Modular Plug Terminated Link or MPTL, the benefits including improved security and aesthetics by avoiding exposed patch cords that can be inadvertently or deliberately disconnected such as on a security camera, and the ability to adhere to the code requirement of placing only plenum-rated products in air-handling spaces, the recognition of the plug-terminated link in the BICSI 005 Electronic Safety and Security standard, the BICSI 033 standard and the TIA-862 Building Automation Standard which allows for the use of a plug-terminated link when eliminating an equipment cord is deemed unfeasible or unsafe, the former terminology of direct attach connection, the previous absence of specific test requirements for a plug-terminated link specified by the TIA leading to testing using Modified Single-Connector Permanent Link testing with a Permanent Link Adapter at the patch panel and a Channel Adapter at the far end and the problem that using a Channel Adapter at the far end excluded the mated connection at the far end from the test, the recognition by standards bodies with the proliferation of field-terminated plugs and the potential for a poor plug termination of the need for a test procedure including the final plug connection at the far end, the inclusion of the MPTL configuration in the current draft of the ANSI/TIA-568.2-D standard and its integration into leading field test systems, and the statement that the Modular Plug Terminated Link configuration is tested using a Permanent Link adapter at one end and a Patch Cord adapter at the other end — https://www.isemag.com/core-legacy/article/14268351/can-poe-take-the-heat
