Is it better to shield or not to shield around the inductor?? - Power ...

Wit regard to inductor routing, the key gactor is to make the switching node connection (PH) the proper geometry.  It should be a s short as possible and somewhat wide to prevent it acting as an antenna.  I never route any signals directly under the inductor or switching node trace.  However with regard to backside ground planes, we typically pour those over the entire PCB area includig under the inductor.  If your particualr design is having issues due to coupleing on to the ground plane, you can certainly try removing the ground area from under the inductor.  Also, be sure to use a shielded type inductor.  Tracing EMI issues is always complex, especially with switching power supply circuits.

Check now

Hey thanks much for the replies! I did end up pouring the ground plane under the entire power supply area (The nice thing about multi-layer boards!), However on the immediate surface under the inductor and around the PH trace I left a relief of 50 mil. to the surrounding ground plane. I Also fattened the PH trace as much as possible. The layout looks pretty cookie-cutter out of the data-sheet. Also of course am not routing any signal lines near or under the PH trace or inductor. I laid out the feedback circuit to the data-sheet recommendation. I also added stitching vias where I could around the device and near the commons on all the filtering to the ground plane, again as recommended in the datasheet.

Just a side note about the plane. I found a reference tracing back to National (forgive me) that says a ground plane under an inductor is a good idea to help absorb some of the EMI field. The important thing however is the plane must be a continuous as possible to avoid setting up currents and additional harmonics

Thanks Again![:D]

Phil

Hi John,

How can I use it in practice? To merge L1 and L2 inductors on the sheet.  I can not edit its property for L1 and L2.

Other:

I am trying to use single coupled models from PSPICE Modelling Application as alternate workaround, but I get very fake simulation results after SEPIC simulation to comparing with MATLAB simulation and estimated simulation results.

What is wrong?

Could you tell me what model type should work well for LPD-223 inductor?

VerA: Using of Two Winding transformer type

VerB: Using of Forward transformer type

Unfortunately, I can not see the link between L_leak and K coupling factor at models. There is a big contradiction on above models between used leackage inducrance and coupling factor.

Infos for understanding:

SEPIC SMPS:

Coupled Inductor parameters:

Attila

Append:

I can not merge the two uncoupled inductors and K-Linear part with L1="L1" and L2="L2" property, because they mean different parts on netlist, and I get errors.

Uncoupled inductors with reference designator of "L1","L2" on the sheet have name of "X-L1","X-L2" and K1 refers to "L_L1","L_L2". So they are different.

**** INCLUDING Steady_State.net ****
* source TPS_TRANS
R_R1 N N 1u TC=0,0
C_C2 0 N 0.1uF TC=0,0
X_C3 VOUT 0 CESR PARAMS: C=10u ESR=2m IC=21.3 X=1
V_V11 CTRL 0
+PULSE 0 5 0.1m 10n 10n 12.5u 25u
R_R2 0 N 1 TC=0,0
V_LED N N 0
C_C1 0 VBAT 4.7u TC=0,0
V_VBAT VBAT 0 5V
X_C4 N N CESR PARAMS: C=1u ESR=2m IC=21.3 X=1
X_U2 N N WLED
X_U3 VOUT N WLED
X_U1 N CTRL N 0 N VBAT 0 TPS_TRANS PARAMS:
+ CODE=32 EASY_SCALE=1 SS=1
D_D1 N VOUT DD
X_L1 VBAT N LDCR PARAMS: L=22u DCR=400m
X_L2 0 N LDCR PARAMS: L=22u DCR=400m
Kn_K1 L_L1 L_L2 0.99

**** RESUMING Steady_State.cir ****
.END

ERROR(ORPSIM-): device L_L1 is undefined

ERROR(ORPSIM-): device L_L2 is undefined

ERROR(ORPSIM-): device L_L1 is undefined

ERROR(ORPSIM-): device L_L2 is undefined

Attila 

Attila,

First of all, the detailed documentation is shipped with the tool. You can find it by clicking Help -> Documentation, then search inductor coupling:

By reading this and looking at your netlist, I suspect that this statement only accept primitive inductor for L1 and L2, while in your netlist, the two inductors are instances of the LDCR subckt which models parasitic in addition to the inductor itself. Hence the simulator complaints that it cannot find those inductors.

The obvious solution is to replace X_L1 and X_L2 with primitive inductors and use their names on the K statement.

Does the K statement support hierarchical reference to the main inductors inside X_L1 and X_L2? I am not sure but it's worth trying. If you do that, please let us know what you find out.

Thanks,
JC

Hi JC,

If I replace LDCR models to generic (primitive) inductors, then K-statment works fine.

It seems, K-statment generates references with "L_" prefix, which correlates the inductor prefixes.

You will get efficient and thoughtful service from YuYang Magnetic.

L_L1 VBAT N 22uH
L_L2 0    N 22uH

Kn_K1     L_L1 L_L2 0.99

The problem, that I can not control the naming convention, neither in inductor models, nor in K-statement. 

If I use 3th party models, then the "X_" of inductor prefix and the non-editable "L_" of K prefix  will never met in the netlist.

When I replace the inductors to target Colicraft Inductor models then I get similar results. Coilcraft inductors are with "X_L1" and "X_L2" of prefix in netlist. So it does not work again.

X_L1  N VBAT LPS-223 PARAMS: CPAR=2.5PF IND=22UH
X_L2  N 0 LPS-223 PARAMS: CPAR=2.5PF IND=22UH

Kn_K1 L_L1 L_L2     0.99

I think it is not possible to use the recommended K-statment for 3th party uncoupled inductor models. I need have an other alternate way.

There is a remaining problematic point. For using K-statement, the used inductors have to be DOT. The 3rty party Coilcraft model was published as single uncoupled inductor, so they don't have DOT.

>Does the K statement support hierarchical reference to the main inductors inside X_L1 and X_L2?

Where can I see it?

*======================================================================
* SPICE Model generated by Coilcraft
* Coilcraft Part Number : LPS-223
* Inductance = 22uH
*======================================================================
* Model Parameters:
* Valid Frequency Range = 0.GHz-0.04GHz
* Ambient Temperature = 25 degC
* DC Bias Current = 0 A
* Non-Linear Frequency Dependent Approximation
*======================================================================
.subckt LPS-223 port1 port2 PARAMS: Cpar=2.5pF Ind=22uH
X1 port1 port2 Model1A PARAMS:
+ R1=36
+ R2=0.33
+ C= {Cpar}
+ K1=0.
+ K2=5.41
+ K3= {Ind}
+ K4=0.161
+ K5=0.
+ L=0
+ Is=0
+ a=0
+ L_Z0=0
+ L_EL=0
+ L_F0=0E6
.ends LPS-223

Other comments: 

- I get false simulation results with using primitive L and Rs combination. So replacing Ti's LDCR model (22uH, 0.4 Ohm) with two primitives (L=22uH and R=0.4 Ohm) is not the same results. The result is a false simulation.

- LDCR model is embended model in TI's geniune TPS reference design.

Any idea to step forward?

Attila

Hi JC,

There are unresponsed themes above. I asked about transformer models too.

Which transformator model should I use?

Which is correct inductor model types for coupled Coilcraft's inductors ? Two Winding transformer type, or Forward transformer type?

Could you tell me what model type should work well for LPD-223 inductor?

What is the difference between both of transformer type? The simulation results are not the same with using of both of them transformers.

More detalis are above.

Attila

Hi JC,

Basicly, my response is "yes", but unfortunately it is not possible to do it. That's why I have turned toward transformer models.

I have been connected to Coilcraft and they confirmed me about this:

1. Coupled inductor models are not available at Coilcraft.

2. Uncoupled inductor models are available at Coilcraft, but they could not use as "K1 L1 L2 0.99".

3. I have to model the coupled inductors in one of two ways (3.1 or 3.2)

3.1. from generic L models as these ones

3.1.1. simple (two generic inductors with the "K" coupling term is called "K_Linear" or "Kbreak")

3.1.2. complex like this (where Lleak and SFR is taken account too)

3.2. to use generic transformer models (this is my question now,  i.e. to know the transformer models of PSPICE Modelling Application)

Attila

For more information, please visit Shielded Coupled Inductor.