Hi, I'm trying to design step up switch capacitor ...
# analog-design
p
Hi, I'm trying to design step up switch capacitor converter. Which takes 1.5V as input and gives 4.5V as output. Operating frequency is 1kHz with 50% duty cycle. I have attached pics for the design. the input frequency is fed to the non-overlapping clock generator and then used to switch mosfets. can someone help me with my master's project, please. I have tried to change capacitor values, adding some load but didn't got as expected.
m
@Pritesh Ps If you’re using devices that will see voltage differences greater that 1.8v, you should use the 5V version of the logic from the
hvl
library.
p
I'm using 20v pmos from 'fd_sc_hd' library which comes with xschem + ngspice+ sky130. I'm not aware of hvl library and also how to include it for simulation.
m
Great! The discrete mosfets look ok. How about the standard logic cells. There’s a property on the logic symbols indicating the library. You can change that from
hd
to
hvl
. The
hvl
library is included in the standard
volare
pdks. I imagine you can just add another include line, replacing the 2
fd_sc_hd
with
fd_sc_hvl
.
l
Use 5V devices. They are enough and more versatile. You will need buffers for your non-overlap clocks. If you want a 4.5 V output, you may need more rectifier steps.
Try to use ideal non-overlap clocks first. Sometimes, the gap between the non overlap clocks isn't enough for the simulation.
I recommend you trying a simple dickson charge pump first. I'm not sure your PMOS devices have properly biased substrates. It can adversely affect your cirucit.
👍 2
p
@Mitch Bailey Thank for the library suggestion. however After changing it, it doesn't getting recognized in simulation. Am I doing anything wrong?
m
I think you need to change the prefix in the xschem property window from
sky130_fs_sc_hd__
to
sky130_fd_sc_hvl__
.
p
@Mitch Bailey yes it worked. thanks again.
👍 1
@Luis Henrique Rodovalho thanks for the suggestions. I have tried to use nfet_05v0_nvt but it shows error during simulation as, model issue.
I also tried to remove VSS voltage supply node and put GND for VGND & VNB in digital cells. Also tried to change nmos L to 0.5 thinking sometimes it throughs error for 'could not find a valid model name', but still it is same.
ok. I will start with dicksons multiplier, along side with this.
l
You're using nvt transistors. Maybe it is the problem. I'm not sure also if native NMOS transistors have isolated substrates in this technology. Additionally, 5V transistors can't have lengths below 0.5 um. Is there a symbol with the bulk terminal available? you should use them.
p
@Luis Henrique Rodovalho Thanks again. I have used 5v nmos with bulk terminal symbol it, and it worked. Also, i started with Dickson charge pump and got the results, however it needs to be operated at very higher frequency. And its not feasible to incorporate it for few kHz frequency. What i'm trying to recreate is in the image. I have made all the blocks with proper results. But, not able to generate 3*VDD, as i have very less understanding for DC-DC switch capacitor, as shown. is there any alternative way while keeping the frequency at 1kHz or 2kHz?
I have attached schematics for the whole block in GenGateOvrDrvVltg, except Switch capcaitor which is in DC_DC_SW_Cap . I thought at first to have it working independently with ideal sources and then incorporate it with the rest.
l
Ok, it is good it is working now. What you need is more DC gain stages. And there are some tricks you can use. First, the 5V transistors are only for the transistors that do get 4.5 V. In another words, maybe you could use 1.8 V. The dickson charge pump is just a starter. You must find the bottleneck of your circuit. Is it the driver of your charge pump? Frequency? The diode size? The capacitors? I haven't simulated your circuit yet. I will see it soon.
p
In charge pump I think capacitor is the bottle neck in my case. Since I have to stick with 1 kHz frequency.
l
@Pritesh Ps I've tried your circuit with a few modifications and it didn't work... You shouldn't use a logic with a 4.5 V supply to make a DC-DC converter for 4.5 V. If you already have it, it doesn't make sense. At least, I don't understand why.