 |
| Technical Specification: | |
|---|---|
| Quantity: | |
Board to Board Connector
Industrial, IT Datacom
Two boards that have to talk to each other rarely line up: panel tolerance, paste volume, board bow and placement repeatability all push the same way. A rigid socket turns that error into contact wear and eventually into intermittent pins. The two CONSHARE codes here, AAC8F040G2049R1 and AAC8F060G2049R1, are built for that: 0.80 mm pitch, 40 and 60 positions, vertical SMT, a 4.90 mm LCP body and 3 µin gold on the contact area.
A 4.90 mm body sets only half of the spacing between your boards. The plug sets the rest:
Plug body | Board gap | What the gap typically has to hold |
|---|---|---|
4.80 mm | 6.50 mm | Air and a stiffener, sometimes shield foil |
6.80 mm | 8.00 mm | A folded FPC tail or a low board-level shield can |
7.80 mm | 9.00 mm | A row of 0402 or 0603 parts, or a thermal pad below |
Read the table backwards if you know what has to fit; if you do not, start at 8.00 mm, because shield cans appear late and 6.50 mm leaves nowhere to put one.
At 0.80 mm pitch it is tempting to give every position to a signal and sort out returns later. Work the other way. Reserving the four outermost positions per row for ground costs eight: 20 percent of the 40, 13 percent of the 60. On the 40 that hurts; on the 60 it buys room to reroute without a respin.
The rule that holds up: a frozen net list near 25 signals takes the 40; any chance of a second sensor or an upgrade board, and the 60 spends 8.00 mm of length to remove that risk.
Body width, pad geometry, land pattern and the 4.90 mm height are identical between the two codes:
Part number | Positions | Body length | Contact span | Commonly used for |
|---|---|---|---|---|
AAC8F040G2049R1 | 40 (2 x 20) | 25.00 mm | 15.20 mm | Camera and sensor mezzanines, compact display modules |
AAC8F060G2049R1 | 60 (2 x 30) | 33.00 mm | 23.20 mm | I/O mezzanines carrying power, control and high speed pairs together |
Length follows 0.80 x positions per row + 9.00 mm, so one footprint covers both codes and the count can be decided later.
The contact housing shifts inside the receptacle body while the plug stays rigid on its own board. Three error sources get absorbed:
Placement offset. Roughly ±0.3 to ±0.5 mm per axis for this class of part.
Board bow. A 33 mm body on a module that bows 0.2 mm spends half the allowance before you start.
Thermal growth. Over 33 mm, a 60 °C rise on FR4 adds about 0.03 mm, on top of the other two.
That split matters in service: pull a module for repair and the housing recentres itself on the next mate, where a rigid socket would have kept the offset as permanent wear.
At 120 mΩ maximum this part sits above the roughly 80 mΩ of a fixed socket at the same pitch, and that is worth pricing before you commit. Half an amp through one contact drops 60 mV; a signal out on one contact and back on another loses 120 mV. On 3.3 V nobody notices. On a 1.0 V core rail it is 12 percent of the noise budget, so parallel three or four contacts or give that rail its own connector.
The contact area gets the specified gold; the solder tails get flash only, over a nickel underplate that runs under the whole contact. Two things worth knowing before you write a spec: you are not paying for thick gold on a surface that never sees a mating cycle, and more gold on the tails will not give you stronger joints. Above about 5 µin, gold in a solder joint forms brittle intermetallics and weakens it. The flash keeps the tail solderable on the shelf and dissolves in the first second above liquidus.
Duty | Finish we suggest |
|---|---|
Assembled once, opened rarely | Gold flash |
Field service, roughly 20 to 50 mating cycles | 3 µin (both codes on this page) |
Fixtures and programming jigs, 100 cycles and up | 10 to 30 µin |
The nickel under it is what stops resistance creeping up over years of heat cycles.
Keep pads mask-defined at this pitch so the mask between neighbours does the anti-bridging work, and hold fabrication to ±0.05 mm. The two locating posts stand only 0.46 mm off the housing base: they register the part, they do not retain it. Retention is the solder joint, so size the holes as a slip fit, because an interference fit holds the body off its pads and tilts the interface.
A long, narrow body with two rows of pads is sensitive to uneven paste: more volume at one end tilts it, and a tilted body burns the float allowance before the plug arrives. Use a 0.10 to 0.12 mm stencil with apertures matched along the length, and if the ends lift, cut the ground tab apertures by about 10 percent rather than adding paste elsewhere.
The LCP body runs through lead-free reflow without the pre-bake a nylon housing needs, and it is rated -40 °C to +125 °C in service. Check that ceiling against local ambient near a heat spreader, not against the part's own 30 mW rise. Wave soldering is out: the floating housing has gaps that wick flux and you will not get it back out. Both codes ship on tape and reel; at 33 mm the 60 position body deserves a feeder pocket check, and pickup belongs on the flat top of the housing, not over the mating opening.
Build | Code | Gap | Why this connector |
|---|---|---|---|
Driver monitoring camera | AAC8F040G2049R1 | 6.50 mm | Housing tolerance plus placement tolerance, and a module that has to swap in the field |
Handheld instrument, display board | AAC8F060G2049R1 | 8.00 mm | MIPI, I2C, reset and 3.3 V in one part instead of a connector plus an FPC |
Industrial I/O mezzanine | AAC8F060G2049R1 | 9.00 mm | Room under the mezzanine for passives and a shield can |
The gap is already decided by something else, such as an optical bench or a heatsink contact. A fixed socket costs less and has lower contact resistance.
A net needs more than 0.5 A, or a 1 V rail has to pass through a single contact.
The product is mated once under a jig and never opened again.
Vibration is continuous with no secondary retention. Ask about a locking version.
You are routing precision analog references, where the floating path's resistance varies with position and that shows up in the measurement.
You need more than 9.00 mm. Take the 8.30 mm body, which opens 9.75, 11.75 and 12.75 mm, rather than stacking spacers.
Gold thickness is verified per production lot rather than per plating bath, contact resistance is sampled, and coplanarity is measured across the full body length instead of at the ends only. Each reel carries a lot code, so a field question traces to one build.
Both codes are held as finished stock, so sample and pilot quantities do not wait on a production slot. Pin counts from 10 to 120 positions, the 8.30 mm body and finishes from flash to 30 µin are built on the same tooling family, so they do not start from a new tool. Send us the stack-up and we will confirm the matching plug.
No. The floating geometry is a matched pair; with a rigid plug the float fights the other half instead of absorbing anything.
About ±0.3 to ±0.5 mm per axis once board bow and thermal growth are subtracted. If you need more, the problem is usually upstream in the mechanical design.
Yes, within 0.5 A per contact and with the drop accounted for. For a 1 V rail, parallel three or four contacts and check the drop at worst-case current.
For tens of mating cycles, yes. For a fixture cycled daily, step up to 10 µin or more.
Neither code covers it. The 8.30 mm body handles the taller range, and a custom height is possible if volume justifies it.
