wafer.space Community - Passive on-chip components? Inductor Spirals?

Hey everyone! I'm currently planning a ~900 MHz 802.11ah HaLoW RF front end for the gf180mcuD process. Since the open PDK ships no RF characterization; has anyone modeled on-chip passives on this process? I'm specifically looking at on-chip inductors.

Sadly no one has made any MIMO HaLoW basebands and the performance of a 2x2 or 3x3 MIMO radio would be undeniable

Yes, we had one project with some inductor test structures on ws-run2. See Code RLCM: https://github.com/wafer-space/ws-run2 These were simulated using openEMS as far as I know. @Ghaith Al Sabagh may be able to tell you more.

GitHub - wafer-space/ws-run2: wafer.space GF180MCU Run 2

wafer.space GF180MCU Run 2. Contribute to wafer-space/ws-run2 development by creating an account on GitHub.

Hi @Sypher I have modeled the stack in openEMS and Palace AWS. There u can simulate ur own passives. In RLCM I have made some test structures to validate the stack modelling of the PDK.

Sypher

Sadly no one has made any MIMO HaLoW basebands and the performance of a 2x2 or 3x3 MIMO radio would be undeniable

Oh, what kind of backhaul would that need? Like, bandwidth-wise but also required latency performance. I'm assuming soft MAC to not need too much buffer on the gf180mcuD die? Luckily inductors should work over SRAM if the latter is configured with a deliberate tree-only PDN topology to not capture induced currents.

Ghaith Al Sabagh

Hi @Sypher I have modeled the stack in openEMS and Palace AWS. There u can simulate ur own passives. In RLCM I have made some test structures to validate the stack modelling of the PDK.

Cool, thank you!

namibj

Oh, what kind of backhaul would that need? Like, bandwidth-wise but also required latency performance. I'm assuming soft MAC to not need too much buffer on the gf180mcuD die? Luckily inductors should work over SRAM if the latter is configured with a deliberate tree-only PDN topology to not capture induced currents.

I have more of a mesh use case in mind, so no backhaul, but if you were running it as an AP for IoT devices it could theoretically service well over 100Mbps on the BSS

2:13 a.m.

soft MAC will actually be on a sidecar crosslink-nx FPGA, only digital RF samples would be coming out of the custom ASIC

2:14 a.m.

And then the Linux device it's connected to could consume it as either a userspace device or use the kernel 802.11 module to have a standard 802.11 interface

2:15 a.m.

Luckily inductors should work over SRAM if the latter is configured with a deliberate tree-only PDN topology to not capture induced currents.

I'll keep this in mind, thank you!