• Resolved

LMX2572: LMX2572 : Phase error measurement

Intellectual 260 points

Replies: 1

Views: 19

Part Number: LMX2572


This is a question related to following.

LMX2572: Phase error after synchronized

Thank you for the advice on how to adjust the phase using MASH seed.

But to determine the seed value, we need to measure the phase difference of multiple IC outputs.

I can only think of using one IC as a reference local and then IQ demodulating the outputs of the other ICs and getting them into ADCs, but if there is a better way?.

We can also install a test terminal to outputs and measure them with a high-speed oscilloscope or net-analyzer, but this will affect the cost of products due to the increase in inspection time.

Could you advise us on the configuration of a phase-detection circuit using your ICs?

I will have to compare the price and area of ICs with inspection cost.


  • Hi user6249907,

    There are multiple ways to approach this problem, depending on the level of alignment required, the number of devices used, the system-wide temperature stability anticipated in the application, and the complexity of the routing available from device to device. I can offer some suggestions, but keep in mind that the calibration techniques used by our customers are usually designed in-house, without TI involvement. So there may be better or easier ways to calibrate the system propagation delay that we haven't learned.

    One conceptually straightforward way to check the phase alignment of the devices is to merge the outputs with a power combiner, and check for the highest/lowest output power across MASH seed increments. The highest power will be at perfect phase alignment, while the lowest power will be at 180° alignment - and since the MASH is deterministic, either of these values could be used to determine the required MASH offset. This does require every synthesizer output to be connected in some way to a power combiner, and requires trust in the phase alignment and skew characteristics of the power combiner ports. On the other hand, it's a relative power measurement, so as long as the power meter is monotonic, the measurement device does not need calibration, and is insensitive to the line length of the connector (which could be helpful for hastening measurements using RF relays).

    For a large number of devices, the part-to-part variation and combined expense of individual RF combiners, switches, and related layout area can become prohibitive to getting high precision alignment between devices. It might make more sense to have an external inspection board with characterized and calibrated delays, which can be connected to many outputs at once to simplify the alignment process. This works best if the LMX2572 outputs are connected to their target devices through some cabling or connector interface; if the LMX2572 output connects directly to its target, and the target is a mixer, a similar scheme could be established farther down the signal chain.

    Hopefully that helps,

    Derek Payne

    Texas Instruments