| T O P I C R E V I E W |
| ind029 |
Posted - Aug 25 2012 : 10:13:13
can anybody tell how to interpret the R+ jwL,
im designing a meander structure with 5 turns,
want to find out the self and mutual inductance and the total inductance of the meander line inductor.
In confusion over the result generated by fasthenry, in terms of R + jwL, the value of L in the result table intrepret what, self inductance,is mutual inductance included in the L value.
plz help. |
| 5 L A T E S T R E P L I E S (Newest First) |
| Enrico |
Posted - Nov 16 2012 : 18:28:28
I checked your reference and I think something is wrong.
Let's first clarify: FastHenry2, for your convenience, gives you directly the simulation result in the format R+jL, as written in the output. This is not an impedance, as you notice, since this should be R+jwL (also dimensionally correct); but it is the format you usually need when you must use the result, otherwise you always end up dividing by w. Therefore, simply put, what you see on the screen is the same result you get using the utility 'ReadOutput'.
However, if you need the full impedance data, without multiplying back by w, you can check the Zc.mat output file, that FastHenry2 always generates. Now, you can read this file with the 'ReadOutput' utility, but then you already had this result on the screen; or you can open the file with a text viewer and use the information within.
Now let's come to the value of the inductance of the loop defined by the FastHenry input file you pasted. Looking at the pictures in the book, it is clear that here we are not considering an inductance per unit length, but really a loop inductance value. This is clear because the scale is H and not H/m, and also because one of the pictures shows the trend of the inductance when increasing the line length. However, while the trends shown in the graphs are valid, the absolute values reported for the inductance are not correct. Such a structure is too small to have a loop inductance of 12pH. As a matter of fact, as you noticed, FastHenry2 calculates 50fH.
Let's do an order of magnitude estimation to be convinced of the result. The inductance of a wire loop is well known with a closed formula that is (as you can check also at en.wikipedia.org/wiki/Inductance#Self-inductance_of_a_wire_loop) r*(ln(8r/a-2), where r is the loop radius and a is the wire radius, and ln is the natural logarithm. If we substitute r=25nm (averaging a bit between the 15nm spacing of the line to the gnd plane and the 100nm of the line length) and a=2.5nm, we get about 75fH, which is very far from 12pH anyway.
I hope this helps you
Best Regards,
Enrico
|
| indr29 |
Posted - Nov 16 2012 : 06:30:48
quote:
Originally posted by Enrico
You have to consider H per unit length I guess; are the results you report referred to a loop over a length of 0.1 um?
Can you quote the paper you are referring to ?
Enrico
I am refering a IEEE Press Book ,"High-speed vlsi interconnections", by Goel, 2nd ed., and the simulation result are taken from a MS Thesis of student as given in the book. J.K.Parambil,"Extraction of on-chip inductances in nano to micro scale range", Michigan Tech University, 2004.
I need your advise on the result obtained after running fasthenry.These are my few queries
1. The computed matrices given in the form of (R + jL) instead of (R + jwL) which is dimensionally correct. Whether we have to divide the result obtained by omega or not ?
2. Yes results are refered to a single conductor of length 100 nm, so to get the total inductance we have to divide it by unit length or by total length of conductor. If i divide the above result (5.03107e-014j) by omega and then divide result obtained by e-12. I am able to get the desired result (16 pH) as quoted in the plot and also all the other values. got highly confused, please help
|
| Enrico |
Posted - Sep 25 2012 : 14:26:41
You have to consider H per unit length I guess; are the results you report referred to a loop over a length of 0.1 um?
Can you quote the paper you are referring to ?
Enrico
|
| ind29 |
Posted - Sep 07 2012 : 08:34:44
Fast Henry code
***************************************************
*6th Sep 2012
* Setting the unit for all dimensions
.Units um
*Defining the ground plane
g1 x1=0 y1=0 z1=0
+x2=0.1 y2=0 z2=0
+x3=0.1 y3=0.1 z3=0
*Thickness of ground plane
+thick=0.025
*Discretization
+seg1=5 seg2=5
* Nodes for latter reference
+nin (0.1,0.05,0)
+nout (0,0.05,0)
* The straight conductor
* The nodes
N1 x=0 y=0.05 z=0.03
N2 x=0.1 y=0.05 z=0.03
*Elements connecting the node
E1 N1 N2 w=0.01 h =0.005 nhinc=1 nwinc=2
* Shorting the end of the conductor with a corresponding pt on the gnd plane beneath it
.equiv nin N2
*Computing the loop inductance from N1 to a pt directly underneath on the gnd plane
.external N1 nout
*computing the impedance for one frequency
.freq fmin=5e8 fmax=5e8 ndec=0
.end
***************************************************
1.Program studies inductances for a single conductor placed at a certain distance above a ground plane.
2.Thickness and are of ground plane are choosen randomly because these values do not effect inductances in general.
3. Current return path for the conductor is assumed to flow through the ground plane.
4. Layout consists of a single conducting interconnection in 3D space above a ground plane.
The default values of the conductor dimensions in the layout and other parameters are as follows: length 100 nm, width 10 nm, thickness 5nm, distance of the conductor from the ground plane 30 nm, and freq 500 MHz.
The problem is i am not been able to get the value of inductance as shown in the graphs of inductance vs width and inductance vs length of a given paper. according to them at 10 nm conductor width, value of L should be 15-16 pH.
Now is there some method to calculate the inductance or it is straight forward, i.e. 5.03107e-014j
where is am commiting mistake ?
|
| Enrico |
Posted - Aug 31 2012 : 16:52:26
It depends on where you put the ports.
If you have defined, as I understand, only one port, with one node at one side of the inductor and the second node at the other side of the inductor, you get a total inductance value, which takes into account of course also the fact that there is coupling between the inductor turns.
If on the other hand you are interested also in the single coupling between the turns, you should define one port for each of the turns.
Best Regards,
Enrico
|
|
|
