IREAP

MURI 2001 Review

Experimental Study of EMP Upset Mechanisms in Analogand Digital Circuits

John Rodgers, T. M. Firestone,V. L. Granatstein, M. Walter

Institute for Research in Electronics and Applied PhysicsUniversity of Maryland

College Park, MD 20742

jrodgers@glue.umd.edu

IREAP

Outline and Motivation• Out-of-band frequency response in communications circuits

– Effect of parasitic elements on network performance– Degradation in filter rejection ratios– EMP propagation on signal path– Need for wideband circuit characterization and verification throughout the

communications network (RF and IF path, mixer, A/D, power vias, etc.)• Experimental study of device upset using direct RF injection

– Identify RF characteristics that produce bit errors, latch-up– What are the EMP effects at the device level?– Modulation and nonlinear circuit response

• Directions to pursue– Experiment– Modeling

IREAP

Schematic of a “loop-back” test circuit for investigating RFeffects in digital communications systems and components

ADC

SAW Filter BP FilterLNA

LO

Mixer

In

Logic

Out

DAC

RAM

LO

Mixer

SAW Filter BP FilterLNA

Probe Probe

Probe

ProbeProbeProbeProbe

Find possible RF entry points, pathways and circuit effectsthat may upset the system or corrupt data.

IREAP

Example: 2 GHz RF LNA

-20

0

20

0 5 10 15

Frequency [GHz]

Gai

n [d

B]

IREAP

Example: 1 GHz low pass filter

-100

-80

-60

-40

-20

0

0 5 10 15 20 25 30

Frequency [GHz]

Forw

ard

Tran

smis

sion

[dB

]

IREAP

140 MHz IF surface acoustic wave (SAW) filter

-100

-80

-60

-40

-20

0

0 5 10 15 20Frequency [GHz]

Forw

ard

Tran

smis

sion

[dB

]

IREAP

Schematic of direct injection experiment

MicrowaveSynthesizer

RF CouplerAmplifier

Computer

10 dB

Power Meter

A

H

U/D

Reset

B1

B8

Load

Carry out

ENB

DRAM

FET Probe

Digitizer

IREAP

Direct injection test facility

IREAP

View of injection coupler and memorymodules inside computer

IREAP

Memory checking code displaying bit errors

IREAP

RAS logic waveform with andwithout RF injection

• Device no longerlatches to Vdd andVss

• RF changesoperating bias point

• Susceptibility mayinvolve synergisticeffects where RFincreases likelihoodof interference frominternal signals.

Row Addressing Pin on DRAM Panasonic 424100RF applied (1.965 GHz at 26 dBm)

-1

0

1

2

3

4

5

6

0 100 200 300 400 500

Time [ns]

RA

S Vo

ltage

[V]

RAS no RF RAS with RF

IREAP

Frequency spectrum of RAS waveform

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

10 20 30 40 50 60 70 80 90Frequency [M Hz]

Am

plitu

de

RF onRF off

Clock Frequency= 33 MHz

IREAP

Results with CW injectionThreshold Power to cause Bit Error at RAS pin

Signal Generator Power

-15

-10

-5

0

5

10

1 1.2 1.4 1.6 1.8 2

Frequency (GHz)

Sign

al G

ener

ator

Pow

er (d

Bm

)

CW

IREAP

RAS Voltage vs. time with PulsedRF Injection (f~2 GHz)

RAS Pin with injected RF before interupt1.965 GHz (PW=150 ns, PRI=300 ns, Pin=29.4 dBm)

-1

0

1

2

3

4

5

6

0 100 200 300 400 500

Time (ns)

Volta

ge (V

)

RF Pulse

RAS Logic Pulse

IREAP

Comparison of results with CW and pulsed injection

Threshold Power to cause Bit Error at RAS pin

0

5

10

15

20

25

30

35

40

1 1.2 1.4 1.6 1.8 2 2.2

Frequency [GHz]

Inje

cted

Pow

er [d

Bm

]

Pulse Mod 50% DF CW

IREAP

Amplitude of demodulated RF signal on RAS vs. frequency

0

0.5

1

1.5

2

2.5

3

0 5 10 15 20

Frequency [GHz]

AM

Lev

el [V

]

Frequency range where

upset was observed

IREAP

Transients induced on RAS by RF pulses atfrequencies up to 20 GHz

-0.4

-0.2

0

0.2

0.4

0 50 100 150 200

Time [ns]

RAS

Volta

ge [V

]

IREAP

What mechanisms may be responsible for theobserved effects?

• Thermal: localized RF energy deposition and rapid heating ofactive MOS regions

• Hot-carriers• Nonlinear circuit elements

– MOS diodes acting as RF detectors– Demodulation of RF by parametric capacitances

IREAP

Upset threshold power vs. duty factor

-70

-60

-50

-40

-30

-20

-10

0

0.001 0.01 0.1 1 10 100

Duty Factor [%]

Thre

shol

d Up

set P

ower

[dBm

]

Average Injected PowerPeak Injected Power

Not a thermal effect

IREAP

Physical Cross-section of CMOS showing equivalentcircuit elements with nonlinear electrical

characteristics

n well

p- type substrate layer

p well

n+n+ p+p+

SourceContact

DrainContact

DrainContact

SourceContact

Polysilicongate

Polysilicongate

GateOxide

GateOxide

FieldOxide

FieldOxide

FieldOxide

n- epi

R

PMOSNMOS

R

IREAP

Drive characteristic of demodulated 4.12 GHz pulse

0

0.5

1

1.5

2

0 0.2 0.4 0.6 0.8 1Injection Power [W]

Puls

e A

mpl

itude

[V]

IREAP

Drive characteristic of 6.0 GHz transient pulse

0.01

0.06

0.11

0.16

0.21

0.26

0.31

0.36

0 0.2 0.4 0.6 0.8 1

Injection Power [W]

Puls

e A

mpl

itude

[V]

IREAP

Conclusions

• High frequency response of communications circuits mustbe considered when analyzing susceptibility to determineprobable entry and propagation paths for EMP.

• The RF shifts the operating bias with respect to Vdd andVss into a nonlinear amplification regime, which could leadto instability, oscillation and chaotic behavior.

• RF pulses are demodulated by nonlinear MOS elements.The envelop voltage constitutes the interrupting signal.

• EMP rise time is a key parameter for inducing interruptsignals over wide bandwidths.

IREAP

Future Work

• The experimental results give basis for modeling highfrequency effects in devices

• Continue to characterize device-level upset mechanismsand seek to develop generalized formalisms

• Study the effects of complex modulation• Look at smaller, faster structures (CPU, RDRAM, DDR,

etc.) and investigate how scaling laws may be applied• Investigated RF effects in mixed signal systems (A/D,

demodulators, etc.)