Email or username:

Password:

Forgot your password?
Attie Grande

Join me, while I disassemble a 100BASE-ZX SFP module (1550nm, 80km)

Surprisingly, these appear to be quite happy running at 1Gbit/s over ~3m of fiber! The reported Rx power doesn't even saturate for the FS.com modules (though it does for the Startech module)

Oh: sorry for the bad photos... I wanted to share, but also couldn't be bothered to get the better camera (and my phone camera is terrible).

A fully assembled FS.com SFP module on a faux-wood desk. The release handle is marked yellow.
14 comments
Attie Grande

These modules are constructed in a delightfully simple way - a solid "chassis" is wrapped in thin steel, and these press-tabs are pushed in once assembled to hold the outer shell in place.

There is one of these push tabs on each side, though they're not quite aligned. They're tricky to get out, but if you're careful, you can poke something thin all the way through!

A close-up photo of a tab, pushed into the body of the module. If you look closely, the internal structure is visible, hard up against the tab.
A photo of the module with a pin sticking through it!
Attie Grande

The best way I've found to open these up, is by inserting something from one side, and pushing these tab out on the other side.

I've previously used a paperclip, but it bent and was a little large... today I modified a needle by cutting the eye. I actually left the fingers a little long - I have three modules to disassemble, and one of the fingers broke off part way through the operation... they can be much shorter!

A photo of two needles. One has the eye cut in half, leaving a V shape, like two fingers.
Attie Grande

Using this tool, insert it, and try to put a finger either side of the tab... then just push!

You will need pliers or a sacrificial surface to push against - it's quite a stiff operation.

Once you've got the tab flush with the outer body, you've actually blocked yourself from getting to the other tab. So revert to a screwdriver to open the tab the rest of the way.

A photo of the needle next to the tab, which is now mostly flush with the outside edge (insdead of pressed into the module)
A photo of a small flat screwdriver bit next to the tab.
A final photo, showing the tab sticking out from the body by nearly 90 degrees.
Attie Grande

Once you've done one tab, and cleared the opening... do the next tab!

The sheath will now slide a little more than it did before. To fully release it, you'll need to take care of the main retention mechanism that keeps the module in the parent equipment's cage. Depress the triangle, and slide the sheath further.

This final clip is actuated by the main release handle, but that doesn't generally get it far enough out of the way to remove the sheath (only remove the module from the cage).

A photo of the module from the top, showing the label, and the two tabs sticking out of either side.
A photo of the bottom of the module, showing the retention mechanism. The PCB is also visible. The photo is a little over exposed, sorry!
Attie Grande

The final hurdle is probably the label... I've taken apart some modules where the label is not an issue, and others (like this one - oopsy), where the label interferes and gets scuffed.

A screwdriver can be inserted under the EMI fingers to just lift the sheath slightly away from the label.

A closeup photo of the label, with some damage. The top clear lamination is coming away in both front corners, and one corner is damaged a little deeper, affecting the print.
A phot of the module, with a screwdriver posed to lift the sheath.
Attie Grande

We're in!

The retention mechanism (handle & clip) will almost certainly fall on the table, so make sure you rescue them.

There may also be a plastic shim to keep the ROSA / TOSA securely in place, but not all modules have this.

Finally, a single screw will usually allow the PCB to lift away from the chassis. There are sometimes metal features on the chassis to "hold" the PCB at the far end from the screw, but not always - just be gentle and it'll come out.

Watch out for a thermal interface!

A closeup photo of the PCB, with a phillips screw visible. In the background (and very out of focus) is the head of a phillips screwdriver.
Attie Grande

Finally, here are all the bits.

No thermal interface in this module, which was a little unexpected. If I remember correctly, the GPON modules I've opened had it... 80km is definately further, but perhaps 100Mbit/s is the key factor here for power dissipation.

A photo of all the pieces, showing the bent-metal shroud, release handle, plastic strut (for ROSA/TOSA), metal release pivot/clip, screw, PCB, and main chassis with a red/white/black FS.com sticker.
A photo of the same things, but the PCB and main chassis have been turned over.
Attie Grande

Look at the laser weld marks on the TOSA!

A photo showing the TOSA ont he left, and ROSA on the right. The PCB is visible below the ROSA.
A photo showing the other side of the assembly - ROSA and TOSA swapped, PCB visible in bottom left.
Ψ*Ψ

@attie ah c’mon you’re not going to take those apart?

Ψ*Ψ

@attie do I really only count two pins on the TOSA?? I guess they must be doing direct modulation on the laser.

Attie Grande replied to Ψ*Ψ

@psistarpsiii No, it's 4x pins total - 2x pins on each side of the PCB.

Attie Grande

@psistarpsiii 😂 I'd love to, but sadly these need to stay operational! 😭

Richard "RichiH" Hartmann

@attie everything up to and including 15 Gbit/s is basically an AM radio, but for light. Stands to reason that you can push them, and also I don't think anyone still makes low enough yield parts to even have "real" 100M in stock any more

Attie Grande

@RichiH Indeed, it was just a little surprising to put a couple of "nominally 100 Mbit/s" modules together and see the iperf3 results far exceed my expectations!

Go Up