Update of the “Digital EMC project”

(1)

MICAS Department of Electrical Engineering (ESAT)

Update of the “Digital EMC project”

December 12, 2006

AID–EMC: Low Emission Digital Circuit Design

Junfeng Zhou Wim Dehaene

KULeuven ESAT-MICAS

(2)

Part I: Additional TF measurement – high frequency up to 2 GHz, – current injection method.

Part II: New structure ready for FIB (Focusd ion beam) work

• frequency domain,

• time domain ,

• conclusion.

Part III: Future plan and related work

Outline

(3)

Measured TF up to 2 GHz

1GHz 100MHz

10MHz 1MHz

> 30 dB

(4)

Measured TF –

AC - current injected by current probe

100MHz 10MHz

1MHz

> 30 dB

(5)

New structure for FIB work

stimulus

1. Req : moving the output pole to high frequency, improving the dynamic di/dt rejection

2. R_ndl: parasitic resistance introduced by FIB

3. C_offchipC_offchip: :

a.a. Make the Vctrl dominant,Make the Vctrl dominant, b.b. Emulate the reduction of Emulate the reduction of

GmGmOTA_OTA

(6)

New structure for FIB work -cont.

V_ref -

+

RLoad

Ctank Vctrl

VDD_input C_aux

Mp

Raux

Rc OTA

Req

(7)

Current TF vs. R _ndl

C_tank: 100 pF I_load: 1 mA C_off-chip: 100 nF R_eq: 10 K

30 Ohms

20 Ohms

10 Ohms

TF vs. R

_ndl

-3dB

-^{10 dB}

-^{20 dB}

-30 dB

-40 dB

-50 dB

100k 10M 1G

1K

150 KHz

In real situation,

R

_ndl

is gone !!

(8)

Current TF vs. I _Load

C_tank: 100 pF R_ndl: 20 ohm C_off-chip: 100 nF R_eq: 10 K

40 mA 9 mA

100 uA

TF vs. I

_Load

-3dB

-^{10 dB}

-^{20 dB}

-^{30 dB}

-40 dB

-50 dB

100k 10M 1G

1K

2 mA 447 uA

150 KHz

(9)

Transient simulation

I_VDD

Vout

Vctrl

C_offchip=100 nF, R_ndl=30 Ohm, RR_eq_eq=1K Ohm, C=1K Ohm _tank=100 pF.

Pulse width = 10 ns, Time interval = 500 ns

(10)

Transient simulation – cont.

I_VDD

Vout

Vctrl

C_offchip=100 nF, R_ndl=30 Ohm, RR_eq_eq=10K Ohm, C=10K Ohm _tank=100 pF.

Pulse width = 10 ns, Time interval = 500 ns

too low

Can not recover to 8V

(11)

Conclusion on the FIB work



Basically, this new structure ready for FIB works fine both in Basically, this new structure ready for FIB works fine both in frequency and time domain,

frequency and time domain,



The smaller R The smaller R

_ndl_ndl

, the better EMI suppression and the lower the , the better EMI suppression and the lower the -3 dB frequency. It won’t hurt in real situation.

-3 dB frequency. It won’t hurt in real situation.



Trade-off on R Trade-off on R

_eq_eq

: :



The smaller R The smaller R

_eq_eq

, the lower 3-dB frequency and the higher , the lower 3-dB frequency and the higher current peak the EMI-Suppressing Regulator can sustain;

current peak the EMI-Suppressing Regulator can sustain;



However, the smaller R However, the smaller R

_eq_eq

, the more DC currents the circuit , the more DC currents the circuit burns.

burns.

(12)

Future plan



FIB the EMI-Suppressing regulator & measurements, FIB the EMI-Suppressing regulator & measurements,



Characterization & quantification of EME from AMIS digital test Characterization & quantification of EME from AMIS digital test structures,

structures,

 The effectiveness of direct supply, LDO or serial regulator on EMC The effectiveness of direct supply, LDO or serial regulator on EMC performance,

performance,

 The impact of lowering the digital supply voltage on EMC performance,The impact of lowering the digital supply voltage on EMC performance,

 The impact of the number of gates on EMC performance, The impact of the number of gates on EMC performance,

 The impact of using D_FF or MS_FF on EMC performance,The impact of using D_FF or MS_FF on EMC performance,

 The effectiveness of distributed NMOS, PMOS and MIM decoupling The effectiveness of distributed NMOS, PMOS and MIM decoupling capacitors,

capacitors,

 ……



Prediction of EME of digital circuits, Prediction of EME of digital circuits,

(13)

Update of the “Digital EMC project”