Taconic TLY-5 PCB Parameters: Datasheet Values, Application Fields and Production Notes

TLY-5 is the ultra-low-loss point of the Taconic (AGC) portfolio: Dk 2.20 ±0.02 and Df 0.0009 at 10 GHz on a lightweight woven-glass PTFE construction. The useful question is not what the numbers are — it is what each one forces you to do in layout, in the drill file and in the press cycle. This guide reads the 2024-03 TDS line by line, converts Dk into trace widths you can use, and lists the six values that are still circulating wrong on distributor pages.

RF and microwave PCB category thumbnail showing a microwave circuit board

How to Read the TLY-5 Datasheet

The current AGC/Taconic TLY-5 technical data sheet is a two-page document dated 2024-03. The first thing to understand is what kind of document it is: the values are typical, not specification limits. That matters at the two ends of a project. In design, typical values are what you simulate with. In procurement, typical values are not what you reject a lot against — you need the purchased laminate certificate and your own incoming check for that.

The second thing to understand is that the grade is narrow. TLY-5 is specified at Dk 2.20 ±0.02 at 10 GHz. The TLY family spans a wider range (roughly 2.17 to 2.40 depending on grade and thickness), and 2.17 is TLY-5A, not TLY-5. Several of the parameter tables you will find online are actually TLY-5A rows or a blend of the two.

That single mix-up is the entry point to a cluster of errors. The table below lists the six that matter, with the TDS value beside each.

Value in circulationAGC/Taconic TDS (2024-03)Why it matters
Dk 2.17 ±0.022.20 ±0.02 @ 10 GHz2.17 is the TLY-5A nominal. A 1.4% Dk error moves 50 Ω width by about 1% — small, but it is the fingerprint of a copied table, and the same copy usually carries the CTE and peel errors below.
CTE X/Y 15, Z 23 ppm/°C26 / 15 / 217 ppm/°C (25–260 °C)The Z figure is off by a factor of ten. Design rules built on 23 are wrong by an order of magnitude.
“CTE closely matches copper”Z 217 vs copper 17 ppm/°C12.8×, not matched. The Z axis is a constraint to be engineered around, not a feature to advertise.
Peel strength 1.1 N/mm3.04 N/mm (17 lb/in), 1 oz C1, Condition AUnderstated by nearly 3×. Peel strength is one of the reasons TLY survives rework and connector loading.
“Halogen free”Br/Cl free to IEC 61249-2-21The resin itself is a fluoropolymer, so the honest statement is halogen-free by the Br/Cl definition, not “contains no halogens”.
“Tg” quoted as a numberNo Tg step; PTFE is semi-crystalline, Tm ≈ 327 °CThere is no glass transition to design around — modulus drifts gradually with temperature instead of stepping down once.

The application and processing sections of most TLY-5 write-ups are fine. The parameter sections are where the copies drift. If a number in your own specification was taken from one of these tables, it is worth re-checking against the TDS.

Electrical Parameters: Dk 2.20 and What the Tolerance Actually Buys

Dk 2.20 ±0.02 @ 10 GHz (IPC-650 2.5.5.5). The tolerance is ±0.9%, which sounds tight until you compare it with the other error sources in a real stack. For microstrip, impedance sensitivity to Dk is roughly 0.4% of Z0 per 1% of Dk — so ±0.02 on Dk contributes well under ±0.5% to impedance. Etch tolerance and dielectric thickness combined routinely contribute several times that.

The practical conclusion: do not buy tighter Dk for impedance accuracy; buy tight dielectric thickness. Dk tolerance earns its money in a different place — phase consistency across a phased-array aperture, where every element must have the same electrical length, and where a batch-to-batch Dk step between panels shifts beam pointing instead of just nudging a return loss.

Df 0.0009 @ 10 GHz (IPC-650 2.5.5.5) is the number the material is bought for. Two caveats that the datasheet condition implies:

  • It is a 10 GHz number. PTFE-class dissipation factors creep upward with frequency. A 77 GHz link budget built by scaling the 10 GHz figure will be optimistic — measure at band, or accept a design margin.
  • It is a material number, not a board number. Insertion loss on the finished board also carries copper loss, surface roughness, plating and connector launches. At 77 GHz those terms can exceed the dielectric term even at Df 0.0009, which is why foil choice and finish matter as much as laminate choice.

Dk stability across frequency is the quiet advantage of the woven-glass PTFE construction: the dielectric constant holds closely enough from 1 GHz into the Ka band that a design captured at 10 GHz stays usable at 30 GHz without re-characterising the material. For phase-critical work, verify with a coupon at the operating band anyway — the material will not surprise you, but the stackup might.

What 0.0009 Means in Decibels

Dissipation factor is an abstraction. In physical units, dielectric attenuation in a homogeneous medium is αd ≈ 2.6 · fGHz · √Dk · Df dB per inch. For TLY-5 at 10 GHz that lands near 0.035 dB/inch. A microstrip puts part of the field in air, so the dielectric share falls to roughly 0.02 dB/inch; the rest of the measured loss is copper and roughness.

Put sideways against the materials TLY-5 competes with, the gap is the whole argument:

MaterialDkDf @ 10 GHzDielectric loss, dB/invs TLY-5
FR-4 (high-Tg)4.30.020~1.08~31×
Rogers RO4350B3.480.0037~0.18~5×
Rogers RO4003C3.380.0027~0.13~3.7×
Taconic TLX (family typical)2.500.0019~0.08~2.2×
Taconic TLY-52.200.0009~0.0351×
Rogers RT/duroid 58802.200.0009~0.0351×

Figures are computed from each material’s published Dk and Df on the homogeneous basis above, so read them as ratios rather than as predictions of your board. The comparison to RT/duroid 5880 is worth stating plainly: electrically the two are interchangeable, and the choice between them is mechanical — TLY-5 uses light woven glass (more dimensionally stable, better registration on large panels), 5880 uses non-woven microfibre (isotropic, no weave effect). That trade is worked through in the TLY-5 application and breakthrough guide and, from the other side, in the RT/duroid 5880 advantages breakdown.

CTE 26 / 15 / 217: the Z Axis Is the Constraint

The TDS gives CTE X 26, Y 15, Z 217 ppm/°C, measured 25–260 °C by TMA to ASTM D3386. Three consequences follow, and they are the ones that separate a working PTFE build from a failure analysis report.

1. The Z axis, not the X/Y plane, sets your plated-hole rules. Copper expands at 17 ppm/°C. At 217, the laminate expands 12.8× faster. Over a 200 °C excursion each plated barrel is asked to stretch about 4.3% of its length — roughly 69 µm on a 1.6 mm board — and copper’s own expansion only gives back about 0.3%. Whether the barrel survives is a plating-ductility question, not a laminate question: specify high-ductility copper with ≥20% elongation, keep PTH aspect ratio at or below 8:1, and treat microsections at the panel corners as mandatory rather than optional.

2. X and Y are not equal. 26 along the glass warp against 15 along the weft is ordinary anisotropy for a woven product, and it is why a large panel can bow differently in length and width. Lay out symmetric copper, keep the panel orientation consistent between prototype and production, and state the grain direction on the fabrication drawing if the design is dimensionally tight.

3. There is an exit ramp. When layer count or thermal cycling pushes past what 217 supports, TLY-5Z is the same nominal Dk with ceramic filler that cuts Z-CTE to 130 ppm/°C and density to 1.92 g/cm³, at the cost of Df rising to 0.0015. Specify TLY-5 when loss is the binding constraint and the stack is thin; specify TLY-5Z when via reliability or mass is.

For reference on how the Rogers equivalent behaves on the same axis, the 5880 family sits at Z 237 — the same order of magnitude, and the same set of rules applies. The parameters behind that comparison are in the RT/duroid 5880 laminate overview.

Thermal, Mechanical and Dimensional Parameters

These are the values that decide whether the board survives assembly and service. They are quoted less often than Dk and Df, and they are where the practical limits of the material live.

PropertyValueCondition / methodDesign consequence
Thermal conductivity0.22 W/m·KASTM F433Lower than FR-4 (~0.3). Heat leaves through copper and vias, not the dielectric — power designs need thermal vias or a heatsink interface.
Glass transitionnonePTFE semi-crystalline, Tm ≈ 327 °CNo modulus step to design around; properties drift gradually with temperature, so analyse the worst-case temperature rather than a Tg crossing.
Peel strength3.04 N/mm (17 lb/in)1 oz C1, Condition A, IPC-650 2.4.8High for PTFE. 1 oz CL1 gives 2.86, ½ oz ED 1.96, and after elevated temperature 2.32 — still more than enough for rework.
Flexural strength96.91 / 89.32 N/mm²MD / CD, IPC-650 2.4.4About 8% anisotropy; both directions comfortably above what handling and connector insertion demand.
Young’s modulus~9.65 GPa (1.4×106 psi)MD, ASTM D3039Roughly 2–3× softer than FR-4. Support large boards, avoid point loads, and think twice about press-fit connectors.
Poisson ratio0.21MDNormal for a glass/PTFE composite; relevant only to FEA of constrained assemblies.
Density2.19 g/cm³ASTM D792Heavier than FR-4 (~1.85). Not a lightweight material — if mass matters, TLY-5Z at 1.92 or a thinner dielectric is the lever.
Moisture absorption0.02%IPC-650 2.6.2.1An order of magnitude below FR-4. No humidity-driven Dk drift, no bake-out ritual in service — still bake before press on multilayer.
Dimensional stability~0.04 mil/inchMD, after bake and thermal stressThe payoff of the woven reinforcement: registration holds on large panels and across production lots.
Arc resistance>180 sASTM D495Relevant to high-voltage RF supplies and antenna feeds.

Two of these deserve emphasis because they run against the usual sales narrative. First, 0.22 W/m·K is worse than FR-4: low loss and good heat spreading are opposing requirements in a PTFE laminate, and a 50 W amplifier on TLY-5 has a thermal problem the laminate will not solve. Second, the material is soft: 9.65 GPa means a 1.6 mm panel flexes visibly under a cable pull, and repeated flexing is how thin PTFE boards crack at the plated holes.

Dielectric Strength, Insulation and Flammability

Dielectric breakdown >45 kV (IPC-650 2.5.6), with a short-time dielectric strength on the order of 2,693 V/mil (ASTM D149) for the raw dielectric. Both are material-level numbers measured on a coupon, not a clearance specification for your board. Creepage and clearance for the finished assembly still come from IEC 60664-1 or the applicable product standard — use the TLY numbers to confirm the material is not the weak link, not to size a gap.

Flammability: UL 94 V-0. V-0 is expected at this price point and is worth confirming rather than assuming, because it constrains what else can be added to the resin. The reason PTFE gets V-0 without brominated additives is the same reason RO4003C cannot: halogenated flame retardants introduce polar groups that raise Df, and a material targeting 0.0009 has no budget for them.

On “halogen free”. The honest wording is Br/Cl free to IEC 61249-2-21 (bromine ≤900 ppm, chlorine ≤900 ppm, total ≤1500 ppm). PTFE’s backbone is a fluoropolymer, and fluorine is a halogen — but fluorine is outside that definition and outside the RoHS restricted list. So the material is compliant and the claim is defensible, provided it is phrased as a Br/Cl threshold rather than as an absolute.

From Dk to Trace Width: Thickness Options and 50 Ω Geometry

This is the step most parameter tables skip. Dk 2.20 is only useful once it becomes a width, so here is the conversion for the standard dielectric thicknesses and 1 oz copper, from the Hammerstad–Jensen microstrip solution.

Dielectric thickness50 Ω microstrip widthAvailable copperWhere it is used
5 mil / 0.127 mm0.44 mm½ oz, 1 ozMillimetre-wave antenna elements; tight bend radii
10 mil / 0.254 mm0.84 mm½, 1, 2 ozThe 77 GHz workhorse and the most common TLY-5 build
15 mil / 0.381 mm1.24 mm½, 1, 2 ozFeed networks with mechanical margin
20 mil / 0.508 mm1.63 mm½, 1, 2 ozFilters, couplers, mixed analogue/RF sections
30 mil / 0.762 mm2.42 mm½, 1, 2 ozHigh-Q resonators, power combiners
40 mil / 1.016 mm3.21 mm½, 1 ozAntenna substrates needing a rigid panel
60 mil / 1.524 mm4.78 mm½, 1 ozLarge aperture arrays; connectorised test fixtures
62 mil / 1.575 mm4.93 mm½, 1 ozLegacy 1/16″ form factors

Because W/h stays near 3.1 for every row, the effective dielectric constant stays near 1.87 as well — so electrical length barely depends on thickness: λg is about 21.9 mm at 10 GHz (λg/4 = 5.5 mm) and 2.85 mm at 77 GHz (λg/4 = 0.71 mm) across the whole table. Thickness buys impedance match and mechanical stiffness; it does not buy a shorter line. If you need shorter elements, that is a Dk question, not a thickness question.

The corollary is a fabrication warning: at 5 mil the 50 Ω line is 0.44 mm wide, and a photolithography-plus-etch window of ±25 µm is already ±5% of the width. On thin dielectrics, etch tolerance becomes the dominant impedance error term, which is exactly when a controlled-impedance coupon and a ±10% acceptance band stop being paperwork.

Widths here are first-order calculations: solve them again in your field solver once the solder mask, the copper profile and the actual stackup are known. Taconic’s own parameter sheets for the sibling RF-30 grade follow the same logic at a different Dk, in the RF-30 parameter and production guide.

Application Fields, and the Parameter Behind Each One

TLY-5 is specified across six industries, and in each case one parameter is doing the deciding. Mapping them makes the selection argument much shorter than a list of markets.

ApplicationDeciding parameterWhat actually goes wrong without it
77 GHz automotive radarDf 0.0009 + Dk tolerance ±0.02Array elements drift out of phase; range resolution and angle accuracy degrade before the link budget fails outright
Phased-array and satcom antennasDk tolerance + CTE X/YBeam pointing error that cannot be calibrated out per-unit
Satellite payloadsOutgassing TML/CVCM/WVR 0.01% + moisture 0.02%Condensation on optics and sensors; Dk shift in orbit. The data supports screening — qualification remains programme-specific
Avionics and UAVDensity 2.19 + Z-CTE 217Mass budget and via reliability under thermal cycling; this is where TLY-5Z earns its place
RF power amplifiersDf 0.0009 — and thermal conductivity 0.22 as the counter-termThe electrical side is easy; the thermal path is the design work
High-field medical RF (MRI coils, RF ablation)Low loss + non-magnetic constructionCoil Q and SNR, plus field distortion. Biocompatibility is a finished-device question, not a laminate property

The family context around this grade — where TLY-5 sits against TLX, RF-30, RF-35, RF-10 and the rest — is covered in the Taconic properties and applications overview and the TLX series analysis. For the cross-brand decision, including where domestic laminates are now competitive, see the Rogers / Taconic / domestic selection guide.

Production Notes the Parameters Imply

None of the process steps below is exotic — but each follows from a specific number in the tables above, and each is a place where an FR-4 shop will lose yield if it treats TLY-5 as a drop-in.

  • Drilling. Glass-reinforced PTFE is softer than FR-4 in the resin and more abrasive in the glass. Use sharp carbide, high spindle speed and controlled feed, with entry and exit material and a stack height kept low so hole-wall quality holds. Taconic lists the dielectric as laser ablatable — that is useful for window and microvia formation, not a substitute for mechanical through-hole drilling in production.
  • Hole-wall preparation. This is the step that decides whether the plating adheres. Permanganate desmear alone will not activate PTFE; specify plasma (O2/CF4) or sodium-naphthalene treatment before electroless copper, and confirm it with a hole-wall microsection rather than with a pull test alone.
  • Etching. Wide traces are the gift of Dk 2.20 — 0.84 mm at 10 mil — so undercut is a smaller fraction of the width than on FR-4. The exception is thin dielectric: at 5 mil, ±25 µm of etch variation is ±5% of impedance, so ask for a coupon.
  • Lamination. PTFE does not melt and flow, so there is nothing to “flow into” a hybrid stack. Bonding runs through FEP film at roughly 290–315 °C or a low-flow prepreg at 180–190 °C. The 300 °C press cycle often quoted for TLY-5 is the FEP route, and it needs a slow ramp and a slow cool — the Z-CTE of 217 is generating real stress while the stack is still hot.
  • Soldering and assembly. PTFE tolerates lead-free reflow (Tm ≈ 327 °C) and the laminate is compatible with Pb-free assembly. What hurts it is repeated rework, point loading on a 9.65 GPa material, and thermal isolation on large planes — use thermal reliefs, preheat, and limit reflow cycles.
  • Quality control. Verify what the TDS describes: Dk/Df on a coupon at the design frequency, peel strength to IPC-650 2.4.8, thermal cycling for via integrity, dimensional check after bake, and microsection at panel corners on any build with a high aspect ratio.

XCE PCB runs TLY-5 and TLY-5Z on a PTFE-capable line — plasma and sodium-naphthalene desmear in house, controlled-impedance coupons on every RF lot, 2–72 layers, impedance ±10%, panels to 600 × 1200 mm, and samples in as fast as 24 hours with laminate in stock. If you are converting an FR-4 design to TLY-5, send the stackup and we will return the widths, the press cycle and the acceptance plan with the quote.

Frequently Asked Questions

What is the dielectric constant of Taconic TLY-5?
Dk 2.20 ±0.02 at 10 GHz, per the AGC/Taconic data sheet dated 2024-03. The 2.17 ±0.02 figure that appears on many distributor pages belongs to TLY-5A, a different grade in the same family — the two should not be used interchangeably in an impedance calculation.
What is the CTE of TLY-5, and does it match copper?
CTE is 26 / 15 / 217 ppm/°C in X / Y / Z (25–260 °C, ASTM D3386). It does not match copper: copper is 17 ppm/°C, so the Z axis expands about 12.8× faster. Over a 200 °C excursion a plated barrel sees roughly 4.3% elongation, which is why high-ductility plating and an aspect ratio of 8:1 or less are specification items rather than preferences.
How much loss does Df 0.0009 actually represent?
About 0.035 dB/inch of dielectric loss at 10 GHz on a homogeneous basis, or roughly 0.02 dB/inch in a microstrip where part of the field sits in air. That is about 31× better than FR-4, 5× better than RO4350B and the same as RT/duroid 5880. On a finished board at 77 GHz, copper and surface roughness usually contribute more than the dielectric.
What thicknesses are available and what trace width do they give?
Standard dielectric thicknesses run 5, 10, 15, 20, 30, 40, 60 and 62 mil (0.127–1.575 mm) with ½, 1 or 2 oz copper. The 50 Ω microstrip width ranges from 0.44 mm at 5 mil to 4.93 mm at 62 mil with 1 oz copper. Because W/h stays near 3.1, electrical length is essentially independent of thickness.
What processing does TLY-5 require?
Three things differ from FR-4: hole walls need plasma or sodium-naphthalene activation before electroless copper (permanganate alone will not activate PTFE); bonding uses FEP film at 290–315 °C or low-flow prepreg at 180–190 °C, since PTFE has no melt flow; and the soft laminate needs support during assembly. Etching and imaging otherwise follow standard high-frequency practice.
Is TLY-5 halogen free and UL 94 rated?
It is UL 94 V-0, and it is halogen-free by the IEC 61249-2-21 definition (Br ≤900 ppm, Cl ≤900 ppm, total ≤1500 ppm). The resin is itself a fluoropolymer, so fluorine is present by construction but sits outside that definition and outside the RoHS restricted list. It is also RoHS and REACH compliant.
RF & Microwave PCB Fabrication

Quoting a TLY-5 Build?

Send us your Gerber files and requirements for a manufacturability review and a tailored quotation.

2–72 layers ±10% impedance tolerance Manufacturing since 2009