Hello,
We are planning to separate an ARTIQ master crate and a satellite crate by approximately 15 meters for our experimental setup.
The two crates are currently connected through a shorter optical-fiber DRTIO link and are working normally. We would like to replace only the optical-fiber cable with a longer, approximately 15 m cable, while keeping the same Kasli boards, SFP transceivers, ARTIQ configuration, device database, and DRTIO topology.
Our planned connection is:
Kasli master SFP port
|
| approximately 15 m optical fiber
|
Kasli satellite SFP0
Could you please clarify the following questions?
- Maximum or recommended fiber length
Is there any official maximum or recommended optical-fiber length for a normal Kasli master–satellite DRTIO connection?
Would a 15 m link be considered a normal and reliable length for DRTIO?
Are there any known systems using 15 m or longer DRTIO fiber links continuously in laboratory experiments?
- Replacing only the fiber cable
If we keep the same SFP transceivers at both ends and replace only the fiber cable with a longer cable of the same type, should the DRTIO link normally work without rebuilding the gateware or changing the ARTIQ configuration?
For example, if the current link uses:
the same SFP transceivers;
the same optical wavelength;
the same single-mode or multimode fiber type;
the same duplex LC connectors;
the same master and satellite hardware;
would changing the cable length from approximately 2.5 m to 15 m normally be transparent to DRTIO?
- Propagation and total link latency
The propagation delay of standard optical fiber is approximately 5 ns/m. Based on this estimate:
2.5 m fiber propagation delay is approximately 12.5 ns;
15 m fiber propagation delay is approximately 75 ns;
the additional propagation delay is approximately 62.5 ns.
Is this estimate applicable to the physical fiber portion of a Kasli DRTIO link?
What is the expected total additional latency after including:
the SFP transmitter and receiver;
FPGA transceiver latency;
8b/10b encoding and decoding;
clock recovery;
DRTIO protocol and pipeline latency?
Is this total latency fixed and deterministic after every power cycle?
- DRTIO synchronization
Does DRTIO automatically account for the increased link length when synchronizing the RTIO timestamps of the master and satellite?
After replacing a 2.5 m fiber with a 15 m fiber, should events scheduled at the same RTIO timestamp still occur simultaneously at the master and satellite RTIO domains?
Or should we expect a fixed physical output offset between channels located in the master crate and channels located in the satellite crate?
- Output timing calibration
If we require TTL output edges from the master and satellite crates to align within a few nanoseconds, should we measure the actual edge timing with an oscilloscope after replacing the fiber?
Is there an officially recommended way to compensate for a fixed master–satellite output delay?
For example:
adjusting event timestamps in the experiment;
applying a channel-specific delay;
configuring a DRTIO delay parameter;
modifying gateware timing;
or calibrating the external cable lengths.
- Phase-sensitive devices
Could increasing the fiber length affect phase-sensitive devices or operations, such as:
Urukul DDS phase synchronization;
Phaser channels;
coherent microwave control;
synchronized TTL triggers;
distributed DMA or remote DMA operations?
Would the effect be only a fixed latency, or could it introduce an uncertain phase or timing offset after startup?
- SFP and fiber requirements
Are there specific SFP transceiver specifications recommended for Kasli DRTIO?
In particular:
Does the transceiver need to support 2.5 Gb/s operation explicitly?
Can a standard 10G SFP+ transceiver be incompatible because of its internal CDR or rate-locking behavior?
Is multimode 850 nm preferable for a 15 m laboratory connection?
Is single-mode 1310 nm also acceptable?
Are there SFP or AOC models officially tested by M-Labs for 10–20 m links?
- Reliability and diagnostics
After installing the longer fiber, what tests would you recommend to confirm that the DRTIO link is fully reliable?
Would the following tests be sufficient?
Confirm that the DRTIO link and destination remain up.
Run repeated RTIO events on the satellite.
Monitor the core logs for link, AUX, sequence, CRC, or timeout errors.
Compare master and satellite TTL outputs on an oscilloscope.
Run the system continuously for 24–72 hours.
Are there specific commands, counters, or diagnostic experiments that should be used to evaluate the DRTIO link error rate?
Hardware and software information
We can provide the following details if required:
master hardware: [Kasli/Kasli-SoC version]
satellite hardware: [Kasli/Kasli-SoC version]
ARTIQ version: [version]
master gateware target: [target]
satellite gateware target: [target]
current SFP model: [model]
current fiber type: [OM3/OM4/OS2]
current fiber length: approximately 2.5 m
planned fiber length: approximately 15 m
topology: one master and one satellite
Our main concern is whether changing only the fiber length from approximately 2.5 m to 15 m can affect DRTIO reliability, deterministic timing, or synchronization accuracy.
Thank you.