How to use phaser_drtio?

green

My current Sinara system includes a Phaser board, and the system has already been updated to ARTIQ-9.

In my device description JSON file, the Phaser is defined as:

{
    "type": "phaser",
    "ports": [6]
}

I would like to confirm whether converting this to a phaser_drtio setup is as simple as modifying the device type to "phaser_drtio" and then rebuilding the firmware following the standard build process.

In other words, after updating the device description and regenerating the firmware, can I directly flash it to obtain full support for phaser_drtio, or are there additional steps or hardware/firmware considerations involved?

bukehu

Hi!

phaser-drtio or Phaser-MTDDS is a different gateware. To use it with Kasli(-SoC), you would need to build and reflash both the phaser gateware and the Kasli (-SoC) gateware.

You may see instructions for Phaser MTDDS gateware here: https://git.m-labs.hk/sinara-hw/assembly/src/branch/master/src/hw/phaser_drtio_mtdds.md#building-the-gateware-and-firmware

To rebuild and flash Kasli gateware, please refer to the ARTIQ manual: https://m-labs.hk/artiq/manual/flashing.html#re-flashing-your-core-device

Best,
Buke

sb10q

There's also a Hydra build available. Flashing Phaser requires a JTAG cable or kasli-i2c (experimental): https://git.m-labs.hk/M-Labs/kasli-i2c

green

sb10q I currently have a requirement to generate multi-frequency outputs in two separate bands using a Phaser: one in the 70–100 MHz range, and another around 350 MHz.

With my current baseband Phaser, the first band (70–100 MHz) works well. However, for the higher-frequency band, the signal quality becomes significantly worse. In practice, I observe that once the output frequency exceeds ₂₀₀ MHz, the signal quality starts to degrade noticeably.

I have already tried configuring the DUC (Digital Up-Converter) and NCO (Numerically Controlled Oscillator), but these adjustments did not improve the situation.

My questions are:

Is it fundamentally possible for a baseband Phaser to support both of these frequency ranges with acceptable signal quality?
Would switching to phaser_drtio (Phaser-MTDDS) make any difference in this regard?
Or is an external up-converter board strictly required to achieve high-quality multi-frequency output around 350 MHz?

Any clarification or guidance would be greatly appreciated.

claytonho

green
looks like it's this issue: your board was probably built with the wrong antialiasing filters, the original design spec'd them for about 350 MHz corner freq.
looks like you'll have to swap out the filter components on your board for the correct ones

morgan

green

Like claytonho suggested, the Baseband Phaser output is limited by the 200 MHz corner freq AA filter. And with a modified 350 MHz corner freq AA filter, the only way to get >250 MHz output is to use the NCO mixer inside the 1 GSPS DAC34H84. As the single channel input data rate of the DAC is limited to 500 MSPS.
No. As it's a hardware issue, the new Phaser-MTDDS gateware will still be limited by the 200 MHz AA filter.
With a modified 350 MHz AA filter, the measurement shows a 10 dBm drop at 380 MHz. So RF signal power and quality depends on the tone frequencies that are mixed with the 350 MHz. Also compare to a purely digital upconverter, an external analog upconverter will introduce more spurs, LO leakage, etc, which is hard to filter out if theirs frequencies are near to the desired RF signals.

green

We have modified some circuit components and measured the frequency response. However, there is still a significant attenuation beyond 300 MHz.

Before modification:

After modification:

The modified components are highlighted in green in the schematic:

We checked the RF transformer datasheet and found that it shows noticeable attenuation above 300 MHz:

We are now selecting a suitable RF transformer for this application.
I would like to know if there were any additional considerations or practical constraints when selecting this RF transformer.

green

Our hardware engineer tried the capacitor configuration suggested in issue 158, and it indeed meets the requirement at 350 MHz.

Thanks for all the helpful suggestions.