Document the file naming scheme

[proppy]

See https://github.com/google/skywater-pdk-sky130-raw-data/issues/1#issuecomment-1202386286

@miesli inferred that the naming scheme could be something like:

sky130_fd_pr__nfet_01v8_w0p36u_l0p15u_m1(8701_9_10_IDVD).mdm
               ^- Device name
                         ^-  width in m
                                 ^- length in m
                                       ^- metal contact
                                            ^- Here is 8701_9_10 and 8701_11_12. What is the difference ?
                                                     ^- measurement curve (either IDVD or IDVG)

@akinakturk can you confirm that this is the case?

I'd be good to add it to the README and/or the docs directory.

[mithro]

I tried to document the naming scheme for the base level primitives at http://bit.ly/open-source-pdks-naming and https://bit.ly/sky130-names

[mithro]

I believe that the numbers in brackets are connected to the test point names on the test tile.

At the moment we don't have permission to publish publicly the test tile documentation. I'm working with SkyWater to change that and there is a public effort to recreate the documentation from scratch at http://bit.ly/sky130-mpw1-testtile-notes

[akinakturk]

You are mostly right about the naming scheme:

sky130_fd_pr__nfet_01v8_w0p36u_l0p15u_m1(8701_9_10_IDVD).mdm ^- Device name ^- width in microns => 0 point 36 micrometers (u is commonly used for microns) ^- length in microns => 0 point 36 micrometers ^- multiplication factor =>

shows how many devices are in parallel, m=1 corresponds to a single device ^- Here is 8701_9_10 and 8701_11_12. What is the difference ?

These numbers come from the skywater chip. The first is the module number (each module has 12 pads associated with it). The following are some of the pad numbers.

Pads 9 and 10 correspond to the source and drain pad numbers of one of the discrete MOSFETs (gate and body are shared within this module. Module number, source and drain

pads describe the location of this device on the chip)

                                                 ^- measurement

curve (either IDVD or IDVG)

Akin

On Tue, Aug 2, 2022 at 10:29 AM Johan Euphrosine @.***> wrote:

See #1 (comment) https://github.com/google/skywater-pdk-sky130-raw-data/issues/1#issuecomment-1202386286

@miesli https://github.com/miesli inferred that the naming scheme could be something like:

sky130_fd_pr__nfet_01v8_w0p36u_l0p15u_m1(8701_9_10_IDVD).mdm ^- Device name ^- width in m ^- length in m ^- metal contact ^- Here is 8701_9_10 and 8701_11_12. What is the difference ? ^- measurement curve (either IDVD or IDVG)

@akinakturk https://github.com/akinakturk can you confirm that this is the case?

I'd be good to add it to the README and/or the docs directory.

— Reply to this email directly, view it on GitHub https://github.com/google/skywater-pdk-sky130-raw-data/issues/9, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACPNITQENKXVA2IXGUSVLSDVXEWDRANCNFSM55LMF4GA . You are receiving this because you were mentioned.Message ID: @.***>

--

---

Akin Akturk CoolCAD Electronics LLC 5000 College Avenue, Ste. 2105 College Park, MD 20740

[miesli]

Thank you for the clarification (I attributed the "u" to the number of the length and width and not as a unit. So in the end we both arrive at the same value).

In my new example, how to read the data using DMT, we encountered an unexpected behavior. The plot shows the charateristic ID(VG) for all nfet_01v8 with length 0.5um. The device from the module 2602 really differs from the other ones. Do you have an explanation for that?

[akinakturk]

The documentation for these test structures do not provide much information, but I believe the difference is due to some stress effects (assuming that these are the same type of devices). See below.

2602: "nfet w/ sep sb ; w=1.000; l=0.500;"

8436 9 10: "w=0.42; l=0.5; m=1;sa=2.5;sb=2.5;prox_count=0" 8436 10 11: "w=0.42; l=0.5; m=1;sa=1.11;sb=1.11;prox_count=0" 8436 11 12: "w=0.42; l=0.5; m=1;sa=0.68;sb=0.68;prox_count=0" 8701 7 8:"w=0.65; l=0.5; m=1; contact-gate=0.050um" 8701 5 6:"w=0.42; l=0.5; m=1; contact-gate=0.050um"

The main differences between these are w and sa/sb (stress parameters). By the way, currents do scale with w; however, they do not always scale linearly with w. Especially for "narrow" devices, the narrow width effects are pronounced.

Another and probably the most important difference is the device name used in device descriptions. Modules 8436 and 8701 are labeled as "nshort" devices. This is the name skywater uses. These are "standard" thin oxide n-channel FETs.

Mismatch module such as 2602 is labeled as "fet mismatch nfet". Based on w and l values, and other mismatch module labels, nshort is the best name to describe this device. Mismatch modules 2602, 2605, 2607, 2608, 2611, 2612, 2618, 2622, 2624, 2627 have these generic device names, as opposed to standard device names such as nshort. Based on w and l, and the device name used, you need to guess what each of these devices corresponds to.

Hope this helps.

Akin

On Wed, Aug 3, 2022 at 4:59 PM miesli @.***> wrote:

Thank you for the clarification (I attributed the "u" to the number of the length and width and not as a unit. So in the end we both arrive at the same value).

In my new example, how to read the data using DMT, we encountered an unexpected behavior. The plot https://dmt.semimod.de/examples/readin_dut_lib.html#the-generated-files shows the charateristic ID(VG) for all nfet_01v8 with length 0.5um. The device from the module 2602 really differs from the other ones. Do you have an explanation for that?

— Reply to this email directly, view it on GitHub https://github.com/google/skywater-pdk-sky130-raw-data/issues/9#issuecomment-1204477355, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACPNITSYUG6U2DDZO5RECU3VXLMTRANCNFSM55LMF4GA . You are receiving this because you were mentioned.Message ID: @.***>

--

---

Akin Akturk CoolCAD Electronics LLC 5000 College Avenue, Ste. 2105 College Park, MD 20740