| Author |
Message |
shlomo22
Joined: 31 Jan 2007
Posts: 28
Helped: 3
|
 10 Jun 2008 21:38 HFSS RF IC transformer issues |
|
|
|
I want to use HFSS to simulate a RF-IC spiral transformer.
I have succeeded in simulating a single coil spiral inductor, and get the right inductance using the following formula: L=-1/2/pi/Freq/im(Y11)
Based on this, I want to use the following formula for mutual inductance: M=-1/2/pi/Freq/im(Y12)
But the results seem not quite right. The M values I get are negative, and worse, they have a larger magnitude than the L values! That should be impossible since M=k*sqrt(L1*L2) and 0<k<1, right?
Any ideas about what’s going on? Thanks!
My file is attached
|
|
| Back to top |
|
 |
dacaili
Joined: 04 Dec 2007
Posts: 2
|
| 22 Aug 2008 9:46 Re: HFSS RF IC transformer issues |
|
|
|
Use im(Y12)/2/pi/freq instead!
Added after 8 minutes:
Are you sure the equation to caculate mutual inductance is correct?
In the addition,the transformer has four ports ,is it proper to simulate it with two ports?
|
|
| Back to top |
|
|
shlomo22
Joined: 31 Jan 2007
Posts: 28
Helped: 3
|
| 23 Aug 2008 20:10 Re: HFSS RF IC transformer issues |
|
|
|
Here is an update, now that I have figured out the issues a little better.
I used the formula: L=-1/2/pi/Freq/im(Y11)
This came from the following considerations: Impedance Z=jwL so Admittance Y=1/(jwL)=-j/w/L
Y11 means the ratio I1/V1 with V2=0.
I thought that was correct, but since then I have had doubts: Is it the right assumption that V2=0? Perhaps it is better to use the Z-parameter Z11, which assumes that I2=0! After all, a changing current in coil 2 will induce voltage in coil 1 which will change the ratio between V1 and I1. Forcing I2 to be zero should eliminate that effect, right?
So I started using L=im(Z11)/2/pi/Freq instead. And similarly, M=im(Z12)/2/pi/Freq to get the mutual inductance. (Note: these formulas are from memory and it's remotely possible I wrote them wrong.) With this formula, the M values were perfectly reasonable.
The L values were almost identical whether using Y or Z parameters, though the self-resonant frequency did change somewhat. All in all I see more reason to trust the Z parameters. I did a separate calculation based on the ABCD parameters (lots of complicated formulas which I got from somewhere), and the results were exactly the same as with Z parameters.
There seems to be a sort of chicken-and-egg relation between voltage and current which I don't quite understand (anyone want to explain?), but when using Z11 and Z12 the results seem to be correct.
I also decided that the minus sign issues were trivial. I think I fixed them by changing the direction of the integration line on the lumped port (between the two terminals of each inductor)
|
|
| Back to top |
|
|
