Alternative to Fakra connectivity

[Costas99]

Dear Stereolabs Support Team,

We are aiming to use a ZED X stereo camera with a ZED Box Orin NX for our applications. The system relies on the GMSL2 FAKRA-Z extension cable for the connection between the camera and the computational unit.

In our field environment, operators need to connect and disconnect the extension cable on a daily basis. As you know, FAKRA connectors are not designed for frequent mating/unmating (typically only 25–50 cycles). This makes handling difficult for our staff and creates a risk of connector wear or damage over time.

To solve the connector wear issue, we are considering the following custom solution:

1. Keep the 4× FAKRA-Z cable permanently attached to the Jetson (no daily disconnections).
2. Build a short transition cable from FAKRA-Z to LEMO push-pull plug (50 Ω coax).

* Coax type: RTK031 (50 Ω, low-loss, 2.8 mm Ø, DC–6 GHz) or equivalent.
* Connector: LEMO 00 series coax push-pull (≥ 5000 mating cycles).

3. Use a 15 m extension cable with LEMO ↔ LEMO in the middle (for daily plug/unplug) and FAKRA-Z on the camera side (to maintain compatibility with ZED X).

This way, all daily connections would happen through the LEMO push-pull connectors (designed for >5000 mating cycles), while the FAKRA connectors remain untouched.

Before moving forward with this modification, we would like to ask:

* Do you see any technical issues with replacing the FAKRA extension cable in this way?
* Will the signal integrity and PoC (Power-over-Coax) for GMSL2 still be guaranteed if we use high-quality 50 Ω coax (RTK031 or equivalent) and LEMO push-pull connectors rated up to at least 6 GHz?
* Do you already offer, or plan to offer, an alternative extension solution (with more robust connectors) that would suit this type of field use?

Your feedback and recommendations would be extremely valuable before we proceed with ordering materials.

Thank you in advance for your support.

[jay1]

Hi,

I saw your proposal for a custom FAKRA-Z to LEMO cable for ZED X / ZED Box Orin NX to address FAKRA low mating cycles. We have the identical problem on our field system.

May I check if you have implemented this cable solution? If yes, did you encounter any signal, PoC or mechanical issues during testing?

Many thanks

[Myzhar]

Hi @Costas99, @jay1,

thank you for the very detailed description of the problem, and sorry for the delay in answering on this thread.

@jay1 Welcome to the StereoLabs community.
I cannot speak for @Costas99’s implementation, but I can share the StereoLabs position on the design and the points that we would check.

Do you see any technical issues with replacing the FAKRA extension cable in this way?

Electrically the topology is sound in principle: each GMSL2 lane is a single ended 50 Ω coax link with PoC, so any true 50 Ω connector can carry it. The concern is not the connector family, it is the number of impedance discontinuities that you add; the proposed chain introduces two extra mating interfaces per lane (FAKRA-Z to LEMO transition, plus the LEMO to LEMO field coupling) on top of the interfaces already present at the camera and at the 1-to-4 adapter. Please also note that our validated maximum is 15 m of total GMSL2 length, and every cable to cable coupling injects noise and further reduces the usable length, as described here: What type of GMSL2 cable should I use with my ZED X camera?. In your scheme the transition cables and the LEMO segment must be counted inside that 15 m budget, not added on top of a 15 m extension.

Will the signal integrity and PoC (Power-over-Coax) for GMSL2 still be guaranteed if we use high-quality 50 Ω coax (RTK031 or equivalent) and LEMO push-pull connectors rated up to at least 6 GHz?

Honestly, we cannot guarantee it: a custom cable assembly is outside the configurations that we validate, so it is not covered by StereoLabs support or warranty. From a purely technical point of view, the parameters I would qualify are: the return loss / VSWR of the LEMO interface up to at least 3 GHz, since a 6 Gbps GMSL2 link has significant spectral energy well above 1 GHz; the quality of the coax termination itself, which in our experience is the dominant failure cause, far more than the connector body rating; and, for PoC, the DC resistance of centre conductor and shield over the full chain, so that the camera still receives its supply within tolerance after the drop. One practical limitation: the deserializer link status and eye margin are not exposed to the user on the ZED X, so validation has to be empirical, for example running both sensors at 1200p60 for long sessions, at the temperature extremes of your field environment and under vibration, while monitoring frame drops and link re-initializations.

Do you already offer, or plan to offer, an alternative extension solution (with more robust connectors) that would suit this type of field use?

We currently offer GMSL2 connectivity only with Fakra-Z RTK031 assemblies, listed in the store accessories section, and I am not able to comment on the roadmap. There are two approaches that keep you inside the validated setup:

1. Sacrificial coupling. Keep a short GMSL2 Fakra Extension Cable M-F (shortest available length, or the right angle version if space is tight) permanently mated on the ZED X, and perform the daily connection at the junction between that short cable and the 15 m extension. This does not increase the mating cycles, but it moves all the wear onto two inexpensive and field replaceable cables, while the camera connector and the 1-to-4 adapter never cycle.
2. Move the daily disconnection out of the GMSL2 domain. Install a ZED Box Mini close to the cameras, so the GMSL2 link stays short and permanently mated, which is also the best condition for signal integrity, and let the operators disconnect on the Ethernet and power side, where industrial M12 X-coded connectors are rated for several hundred cycles and are properly sealed. The images can then be sent to the main unit with the ZED SDK streaming module, or consumed directly on the ZED Box Mini. Both the ZED X and the ZED Box Mini are designed for outdoor field use, so this keeps the whole perception chain within products we support.

If you do proceed with the LEMO assembly, I would be glad if you could report your results in this thread; it would be valuable for other users with the same field constraint.