# memory cell simulation

Discussion in 'Homework Help' started by miklee, Mar 27, 2011.

1. ### miklee Thread Starter New Member

Mar 27, 2011
8
0
Good morning. I'm looking for a LTspice schematic of a cell memory to simulate. I was thinking in a cmos 6t sram. Memory simulation is a work for my thesis and I need a big help. Thank you so much

2. ### Georacer Moderator

Nov 25, 2009
5,151
1,266
This isn't a free homework service. You must post your attempt at your problem in order to receive help.

Aside from that, a Google Images search of "6t sram cell" will produce many result. I think Wikipedia has an article about sram cells too.

3. ### narasimhan Member

Dec 3, 2009
72
6
There are separate ltspice websites having those simple simulations and tutorials

4. ### miklee Thread Starter New Member

Mar 27, 2011
8
0
My problem is how to set the simulation. On my book, on internet is easy to find a general schematic, but i don't know which kind how nmos/pmos use, which kind of source. My teacher said: "just find a schematic on internet and only made a parametric simulation"

5. ### Georacer Moderator

Nov 25, 2009
5,151
1,266
I don't really think you care what NMOS/PMOS you will use. You might as well use an ideal device, or any foundry you like, just to prove the theory of operation.

6. ### miklee Thread Starter New Member

Mar 27, 2011
8
0
Ok I tryed to make the circuit: what do you think about? Vdd should be equal to V2? Vdd=V2= Pulse(...)?

File size:
85.5 KB
Views:
71
7. ### Georacer Moderator

Nov 25, 2009
5,151
1,266
Vdd is a constant power supply in all the digital circuits you will encounter.

V2 is the controlling voltage "Word Line" that says whether this memory cell will be used or not. If it is HIGH (Vdd) you can read from or write to this cell.

8. ### miklee Thread Starter New Member

Mar 27, 2011
8
0
I setted values and started simulation, but don't work.
Between the first and the second cycle of V(wl), V(q) should keep the high value, shouldn't it?
V(q) sounds cut...

settings:
V1: PULSE(0 1.8 6ns 0 0 2ns 5ns)
V2: PULSE(0 1.8 30ns 0 0 30ns 50ns 2)
Vdd: 1.8V

File size:
161 KB
Views:
50
9. ### Georacer Moderator

Nov 25, 2009
5,151
1,266
Do the following corrections and re-post the graph, so we can analyse it better:

• Reduce the input (bit) frequency. It is too fast for effective measurements. Half or 1/3rd frequency would be nice.
• Make the wl signal steeper. A rise and fall time of practically zero would be good.
• Finally tell us in which node you measure the internal states q and q'.

10. ### miklee Thread Starter New Member

Mar 27, 2011
8
0
I send a more detalied schematic that I done. I corrected the M5 pass-transistor and M2, M3 mos (source-drain inverted).
Now the simulation works but I don't now how connect the M5 and M6 substrate. In this way works but without substrate connection...

File size:
111.2 KB
Views:
77
11. ### miklee Thread Starter New Member

Mar 27, 2011
8
0
I solved the question, I connected correctly the substrate and now works.
I have a question: which parameters I can change for a parametric simulation? The W and L of the inverters' MOS?

File size:
35.3 KB
Views:
41
12. ### Georacer Moderator

Nov 25, 2009
5,151
1,266
I think it's universal that all PMOS substrates are connected to Vcc and all NMOS substrates on Ground. Try it out and see if the problem with the voltage reference persists.

edit: I missed the thread's second page so this post is out of date.

Last edited: Apr 1, 2011
13. ### Georacer Moderator

Nov 25, 2009
5,151
1,266
It depends on what you want to study. If you want to find the maximum feasible frequency you should play with it.
If you want to study the response foundry-based, you could reduce L and W.
If you want to study the response based on W, you could do it, to show how W affect rise and fall times.

Keep in mind that as a general rule the two Word Line transistors should be twice as wide as the rest of them, in order to be able to write the bit state fast enough.