can someone help me identify the layers i'm seeing...
# analog-design
m
can someone help me identify the layers i'm seeing here when i pull the MiM cap (metal 3) device when i pull from the list of devices 2 in Magic, i think i'm supposed to see metal 3 layer and the capm layer but the layers here look different and i can't find layers that look the same from the layer palate in Magic, so if anyone could help me identify these layers and how i could use this cap and thanks.
t
It would look a lot better if you were using cairo or OpenGL graphics. But anyway, the gray layer underneath everything is metal 3. The darker gray square is the MiM cap layer; this sits above metal3 (the capacitance is measured between this cap layer and metal 3). The smaller rectangle on top of that is the MiM cap contact up to metal 4. The stripe on the side is another contact, but that one goes directly between metal 3 and metal 4 (because you never want to connect a MiM cap bottom plate downward, for antenna effect reasons. Always connect it upward to the next higher layer, in this case metal 4).
m
How can i switch to OPENGL?
t
magic -d OGL
, although
magic -d XR
(cairo 2D graphics) is generally more reliable and essentially the same as OpenGL (they both use video card acceleration).
m
Do you know why the layers look different compared to when i use for example metal 3 from the layers palate?
I tried the magic -d OGL but it crashed, any idea what might be the reason?
t
OpenGL implementations can be finnicky. Did you try
magic -d XR
?
m
Yeah that did work, thanks very much
Sorry for asking too much but what advice would you give me if i wanna setup a large cap around 50p , like what is the best way to set it up?
t
No problem with the questions! My suggestion is (1) take advantage of the stacked MiM caps and double up all caps to get 2fF/um^2. (2) Use the maximum size of cap (30um x 30um) (larger metal areas require slotting, which you can't do under a cap). (3) A 30x30 cap at 2fF/um^2 is 1.8pF, so you would need 27.8 caps to get 50pF. (4) You preferably want the same number of the same size cap, so rework that to make a convenient array. So, say, use an array of 6 x 6 caps. Then 50pF / 36 = 1.39pF per cap, so at 2fF/um^2, each cap in the array needs to be 694.45um^2, or 26.35um x 26.35um.
m
Perfect, thanks so much
one last thing, here i have the layout of a lvt PMOS used in a rectifier circuit, i had to use large W for this transistor to minimize the voltage drop across it so that justifies its large size, i wanted to take your opinion on its layout here, i've connected all the gates with all the drains making it diode connected and i've connected them through metal1 and for the sources, i've connected all of them through metal2 and also i've made the drain area less than the source area, so what do you think of this layout, is it good?
t
I'm unclear on where the extents of the transistors are. It looks like the pdiffusion extends past where the transistor is supposed to be defined and under the gate contact, which would be illegal. Also, I don't see poly on the top and bottom rows, so those aren't transistors.
m
The transistors themselves i got them from the devices menu and when i do that the poly is not quite red like when i use the poly layer from the layers, so the brownish red you see is in fact poly not pdiffusion
t
Don't draw additional layers on top of the cell. If you want to modify the contents of the parameterized cell, then edit the cell itself. Keep the DRC checks turned on.
I should clarify that last statement. Don't draw additional diffusion or poly layers on top of the cell. You should only draw metal layers on top of the cell.
m
Oh i didn’t know that, thanks for telling me, but i wonder what’s the reason behind that?
t
Magic detects devices by layer types which are "derived" layers. For magic to create the correct derived layers for a device, all of the layers must be present in the same cell. So there is a device called "pmoslvt" which is a combination of poly and p-diffusion (and the LVT implant, although that does not get shown in magic's layout view). If you draw diffusion or poly on top of that in another cell, the extraction can get completely screwed up.
m
okay i got it, thanks so much for all the info