June 7-10, 2009 San Diego, CA Issues in Power Delivery System Performance Verification Gert Hohenwarter GateWave Northern, Inc. Objective • Review power and ground path integrity • Examine model and measurement approaches • Show some limitations and workarounds June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 22 Approach • Examine power path in a probe card • Illuminate power delivery system integrity requirements (aka power delivery network) • Use models and measurements to evaluate performance • Provide a workable verification approach June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 33 Physical environment Typical test system Power travels from tester to DUT over a considerable distances compared to switching times of the devices under test, thus……. June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 44 … when …current draw changes… … transients result. June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 55 Noise and impedance example • Sudden current draw results in a voltage according to V=L*di/dt • Impedance at a particular frequency is an alternate way of expressing noise voltage: V=Z*i Z PDS = R + jX June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 66 Remedy: Additional energy storage Bypass capacitors (C) provide additional energy during times of sudden high demand June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 77 Tester to die path model Components of the power delivery system (PDS) { P/G planes act as low Z transmission lines ‘T’ } June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 88 Power distribution inside DUT ? • Problem – probe card manufacturer often doesn’t know precisely what’s in the DUT June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 99 Design • PDS impedance specification provided by IC manufacturer • Z PDS = R + jX • Power delivery system performance depends on path and position • Both delivery and return paths are important • Disruptions/detours add inductance • FEA and CAE tools must be set up to take paths into account • Verification June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 10 10 Typical model setup Planes are established at the DUT level - the impedance Z(f) is then determined from the DUT side June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 11 11 Modeled impedance Resonances are due to multiple transmission line sections between bypass capacitors June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 12 12 But does the PDS really work ? • Instruments – DMM / Impedance meter – Time Domain Reflectometer (TDR) – Vector Network Analyzer (VNA) • Probes – Individual – Whole die (surrogate DUT) June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 13 13 Probing – surrogate DUT Probe card site Surrogate DUT Develop thin film based surrogate DUT circuit, touch down on a probe card die site and measure impedance at relevant locations on surrogate DUT June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 14 14 Surrogate DUT in contact with probes Z can be measured/modeled at various locations of die and represents the Z experienced by a circuit under test June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 15 15 Alternative: Discrete probe test Measure returned signal for each accessible pad June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 16 16 TDR measurement Model plot for inductance values from 200 pH to 5 nH TDR does allow for extraction of some info about PDS performance June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 17 17 VNA measurement (Z) Model plot for inductance values from 200 pH to 5 nH VNA shows whether there are frequencies that are “off limits” June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 18 18 VNA measurement (Phase) Model plot for inductance values from 200 pH to 5 nH VNA measurement of phase allows determination of complex impedance e.g. inductance June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 19 19 Example configuration Hypothetical die layout with bypass capacitor placed at one end June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 20 20 Inductance as a function of pad location both 10 9 8 7 6 5 both 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Inductance L (nH) along the die (capacitor location is at right) June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 21 21 Inductance L (nH) along the die 10 9 8 7 6 same 5 across 4 (capacitor is at right) 3 2 Same side - across 1 0 1 2 3 4 5 6 7 8 Measurement taken with probe ground on the same side as signal contact (pink) or across to opposite ground area (blue) June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 22 22 DUT area with multiple bypass caps 7 6 5 4 1 4 3 2 1 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Impedance Z () at 250 MHz as a function of position across die (2 nearly identical die) June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 23 23 Still don’t know overall Z… • Solution: Combine measurements with simulated results for interconnect inductance R12 R7 L19 R6 L17 R5 L15 L18 R3 L16 L12 R2 L10 R1 L4 L13 L11 Ceramic and PCB area with bypass caps L6 L14 T2 L2 L7 DUT / probe area with lateral interconnect inductance plus inductance of contacts June June 77 to to 10, 10, 2009 2009 L20 L9 R10 C2 C6 C5 C7 L1 L5 L3 L8 R8 R11 R4 R9 IEEE IEEE SW SW Test Test Workshop Workshop 24 24 Z combined Ohms 10 1 0.1 0.01 0.001 0.01 f [GHz] 0.1 1 Individual Z measurements combined into an overall performance assessment June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 25 25 Summary • Power delivery system performance depends on path and position • Grounding during measurement as well as model development must be carefully planned • The overall performance is NOT the result of assuming that all measurements effectively form a parallel circuit of the individual Z measurements • A combination of measurements and models allows for assessment of overall performance • Local performance should be verified against specifications June June 77 to to 10, 10, 2009 2009 IEEE IEEE SW SW Test Test Workshop Workshop 26 26