⚠️ this README has not been updated to reflect the final state of the bandgap circuit submitted to efabless.

1. summary of design metrics
2. test results
3. layout pictures
4. transient reset power
5. what's a bandgap circuit?
6. how's the bandgap circuit work?
7. prerequisites
8. steps to set up the repository
9. design files
10. running all tests
11. test descriptions
12. viewing results
13. viewing unscripted results

0 <= Temperature <= 70

a circuit which provides a voltage that's (ideally) independent of temperature. the final circuit in this repository is shown below.

A bandgap circuit consists of the following blocks

a bipolar junction transistor (BJT) given some current, will produce a voltage between its base and emitter. This voltage is dependent on temperature and that dependence is typically around -2mV/degC. Notice that the dependence is negative (the negative dependence is usually called "complementary" to absolute temperature, or CTAT). the Vt generator is typically produced by taking the difference in base-emitter voltages between a pair of BJTs with different current densities, which is achieved by either supplying the BJTs with different currents, or by making the area of one BJT larger than the other. The temperature dependence of this difference, i.e. the output of the Vt generator, is typically +0.085mV/degC, much smaller than the CTAT voltage's dependence. Because this Vt dependence is positive, it's often reffered to as a proportional to absolute temperature (PTAT) voltage. The PTAT voltage is scaled (by M in the figure) such that when the CTAT and PTAT voltages are added, the resulting temperature dependence is zero, since the positive and negative temperature dependencies cancel out.

this circuit has its nuances

the bandgap reference core generates both the CTAT and PTAT voltages. assuming Va=Vb, the currents in R1 and R2 will be the same and they will correspond to the CTAT voltage Veb (since V=IR). The generated Vt appears across R3, so the current through R3 corresponds to the PTAT voltage. These PTAT and CTAT currents are summmed at node Vb and mirrored to the output, where that current generates the reference voltage across R4.
the amplifier works to keep Va=Vb so that the above process can happen (we assumed Va=Vb). the self-bias circuit supplies the amplifier with a bias current.
the startup circuit helps the circuit find its stable operating point after the circuit is powerd on.
the current mirror mirrors current!

it's assumed that you've installed ngspice, skywater-pdk, Xschem_sky130, and Xschem. if you're missing one of these, please follow the steps in this video. the written documentation for Xschem is available here.

note: Xschem_sky130 is different from the skywater-pdk but is entirely necessary. make sure you have it installed.

1. enter bash and source caddy.bashrc
2. in project-paths.json fill in the absolute paths for each of the entries. my paths are left in that file as an example.
3. python edit-xschemrc-paths.py
4. xschem &. see if xschem opens without error. important: please launch xschem from this top directory. xschemrc must be in the current working directory, i.e. only launch xschem from the root of this repository.
5. File -> Open -> /path/to/repo/schematics/tsmc_bandgap_real.sch
   hopefully, you can now see the full bandgap circuit schematic. try generating a netlist from this schematic; click netlist in the top right corner. to verify this worked, click the simulation tab in the top toolbar and then click edit netlist. a nano editor window should pop up, showing the various circuit components in the netlist.

the bandgap circuit is compact and the entire design schematic is contained in the following file
schematics/tsmc_bandgap_real.sch
note: you will need xscheme to open the above! if you just want to run the tests for this design, see the next section.

first, each .spice file must be generated through xschem. all tests are listed in tests.json.

1. xschem &
2. one by one, open the schematics in table below
3. click Netlist in the top right corner of the screen
4. repeat steps 2,3 for all tests
5. in sims/ you should see the different tests in their .spice format
6. python run-tests.py
7. all of the tests will run. to view the relevant results, follow the steps of "viewing results"

as an example, the following commands simulates only one SPICE netlist:
ngspice -b -r ./sim/test-name-here.raw -o ./sim/test-name-here.out ./sim/test-name-here.spice

There is a script to interpret the ngspice data generated from each test, i.e. to get the most salient information you can run these scripts

a printout of relevant operating point voltages is available at sims/tsmc_bandgap_real_op.out.

if you want to play with simulations yourself, use schematics/tsmc_bandgap_real.sch as your sandbox. the difference between this schematic and the schematics in the test descriptions table is that the schematics in the table have specific simulation commands attached to them. schematics/tsmc_bandgap_real.sch has many ngspice commands in the schematic, which are enabled and disabled by the "spice_ignore" variable in the command listing. if you want a set of simulation commands to be ignore, set spice_ignore = True, but if you want to use a set of simulation commands, set spice_ignore = False.