Your SlideShare is downloading. ×
  • Like
Electrical Performance Effects on Wafer Level Test
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×

Now you can save presentations on your phone or tablet

Available for both IPhone and Android

Text the download link to your phone

Standard text messaging rates apply

Electrical Performance Effects on Wafer Level Test

  • 1,546 views
Published

 

Published in Technology , Business
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads

Views

Total Views
1,546
On SlideShare
0
From Embeds
0
Number of Embeds
1

Actions

Shares
Downloads
14
Comments
0
Likes
0

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide
  • Introduce yourself Name, TitleThank the chairs for setting up workshop.I was recently asked to generate data sheet values for a spring pin contact engine. This created a lot of questions and I hope answers. This presentation is some was what was discovered from that activity.I am going to discuss how the spring pin, the housing and material physical constants determine the curcuit performance.
  • My goal today is to discuss the how the physical featues of a WLCSP contact engine effect test performance.I will focus on how the equivalent model lumped element values are defined and extracted from the mechanical features of a contact engine.
  • I’m going to COVER:CLICK -What is a WLCSP contact engine?CLICK - What electrcial parameters are of interest.CLICK - How these parameters are reported on datasheetsCLICK - Some of the methods that are used to characterize contact enginesCLICK - Propose a method of characterization that utilizes analysis, equivalent circuit lumped element models and measurement (data fitting).CLICK - Wrap it up with how the contacts behave in use and a summaryCLICK
  • For the purposes of this presentation I’ll limit the use of WLCSP to spring pin contacts contained in a dielectric housing with pitches that range from 300 – 500 um.Here shown, is an isometric view of a frame populated with a X8 solution and a x-section of a contact engine.CLICK
  • There circuit needs can be broken into two areas.CLICKPower delivery - This is most affected by inductance. This shows up as ground bounce or power rail collapse.I will focus on inductance later in this presentation.CLICKAnd signal integrity High speed data edges for digital signals and controlled impedance for RF.CLICK
  • CLICKAfter reviewing a selection of data sheets and white papers no consistency was discovered in the reporting of fundemental parameters.CLICKAt a minimum Length, Pitch, Radius, Conductivity and Permiativity need to be reported.CLICKAll of the electrical performance of a contact engine is a function of these basic factors.CLICK The inductances CLICK capacitances CLICK and the other basic signal measurements.CLICK
  • CLICKWhen comparing contacts or contact engines know how the values were derived.CLICKBeaware that lumped element models are only valid when Length < wavelength/10.CLICKUse microwave models above this cutoff.CLICK
  • CLICKJust because a pin claims to have low inductance this doesn’t mean that the contact has highbandwidth.CLICKHigh Bandwidth is obtained by controlled impedance.CLICK
  • CLICKWhen measuring a contact engine loop inductance is the only inductance that is meaured.This measurement will also include stray capaciatance and resistance.CLICKThe self and mutual inductance adjust the effective final contact inductance. This adds two new variables that need to be fit to the data or simulation.CLICKTheory can guide us here.
  • CLICKCapacitance cou be calculated but haven’t done this and rely on extraction from simulation and data.CLICKResistance on short pins is small and is a small part of the overall solution.CLICKThere are 5 unique variables to specify on a datasheet or use in electrical simulations.To uniquely define these variables 5 boundary conditions are needed to fit this data.
  • CLICKPartial inductance may be uniquely calcualted. This particular calculation I’m about to walk through is due to an ideal wire with uniformly distributed current. So it should set a theoretically lowest value if the geometery is set to be the largest radius of a pin and nominal test length.How do you get to segment inductance from Loop inductance?CLICKStart with the total magnetic flux per unit current in the area between the pins that make the loop.Convert the area calculation to a line integral of the vector magnetic potential. As an aside the vector magnetic potential is the magnetis equivent to the scalar voltage field that we are all familar with.CLICKBreak the line integral into individual segments.CLICK, CLICKThe partial mutual and partial inductance are one of these calculated segaments.CLICK
  • The calculation yields these two equations:CLICKL partialCLICKM partialNotice that L is a function of length/radius and M is a function of Length/(Pitch + Radius)CLICK
  • The sign of the mutual inductance is important to get right. It comes from the integration path (shown in the blue dotted line) and the direction of the magnetic vector potential. The direction of the magnetic vector potential comes from the direction of the current creating it.For example the mutual inductance on 2 from 1 is negative (Top line 2).CLICK
  • CLICKAfter looking at a few datasheets it was obvious that most all models attributed the same physical characteristcs to the same electrical effects.L to lengthC to separation of chargeResistivity to material effectsCLICKUnfortunately there are many models used to describe a contact engine.CLICK
  • CLICKSome models are simple.CLICKSome are complex.CLICKSome are about right.CLICK
  • CLICKDepending on the equivalent circuit model chosen there are at least 4 parameters to be determined (L, C, M, R). With four unknowns four boundary conditions are needed to uniquely determine the parameters.The lumped element equivalent circuit model is best fit to the impedance profile measured or simulated.CLICKThis can be done by eye or better by optimization in a schematic simulator such as ADS.CLICKRemember lumped element models can be used up frequencies bounded by the condition that the length of what is being modeled < tenth of the wavelength.CLICK
  • CLICKFor typical wireless frequencies Contacts less than 7.0mm can be modeled with a lumped element model up to 2.5GHz. Contacts less than 3.0mm can be modeled with a simple lumped element model up to 6.0 GHz.CLICK
  • CLICKHere is an example of data from a physical simulator (HFSS), a measured set of pins and two different lumped element values in the same equivalent circuit model.Notice that the two equivalent circuit modeled lines (High light with pointer) show very similar fits to the data.In one case Lself is 0.48nH and M is 0.32nH In the other solution Lself is 1.5nH and M is 1.33nH. In this example if I chose the Lself and M to be a constant value the curve will always be the same.A good way to evaluate the mutual component of the equivalent circuit is to measure more than 1 pitch. Since pitch will effect the mutual inductance independantly from the self inductance.What I’d like you to take away from this graph is that there is not enough information in this data set alone to extract the self inductance from the mutual inductance.CLICK
  • CLICK, CLICKThe mutual inductance becomes important when parallel current paths are considered such as a ground return.In this case the inductance of contacts in parallel is not the reciprical sum of the reciprcals. This effect is stronger at higher frequencies.CLICKThis leads to the proximity effect.CLICK
  • CLICK, CLICKMost die have a single ground return for a supply path.To reduce the inductance effct from ground an equally spaced set of ground returns will share current equallyand reduce loop inductance.CLICKThis case will reduce loop inductance the most but the current will not be shared equally in all return contacts due to the reduced current in the diagonal contacts.CLICKFinally, current will be shared equally amonst the supply and return contacts in this case.CLICK
  • CLICK
  • Partial inductance Is the inductance from a segment of a loop. This must be calculated.Loop inductance This is the inductance caused by the magnetic flux through a loop of current.

Transcript

  • 1. The Electromechanical Design of a Spring Pin Based WLCSP Contact Engine and it’s Effect on Signal Fidelity.
    Mike Fredd
    Cascade Microtech
    June 12 to 15, 2011
    San Diego, CA
  • 2. IntroductionGoal
    To better understand the electrical implications of the mechanical features in WLCSP spring pin based contact engines.
    June 12 to 15, 2011
    IEEE SW Test Workshop
    2
  • 3. Agenda
    What is a Spring Pin Based WLCSP Contact Engine?
    Spring Pin Circuit Performance Requirements.
    Electrical Performance - Datasheets.
    Contact Circuit Equivalent Models.
    Analytical Analysis.
    Techniques For Model Fitting.
    Pin Placement in Contact Engine.
    3
    April, 2010
  • 4. IntroductionWhat is a Spring Pin Based WLCSP Contact Engine?
    4
    April, 2010
    For the purposes of this presentation a WLCSP contact engine is a spring pin type of contact in a dielectric housing for wafer level probing using pitches ranging from 300 m to 500 m.
  • 5. Spring Pin Circuit Performance Needs
    Power Delivery.
    Low Inductance to avoid ground bounce and power rail collapse.
    Signal Integrity.
    Digital
    High Speed Edges
    RF
    Controlled Impedance
    5
    April, 2010
  • 6. Data Sheet Comparisons
    A survey of datasheets shows an inconsistant reporting of electromechanical parameters.
    At a minimum these mechanical factors need to be known for electrical characterization.
    Length, Pitch, Radius, Conductivity and Dielectric Constant.
    From these parameters all of the following can be modeled.
    Loop, Partial and Mutual Inductance.
    Shunt capacitance or Mutual capacitance.
    S-Parameters, Phase Delay, Band Width, Rise Time and Isolation.
    6
    April, 2010
  • 7. Data sheet values
    The methods used to extract, calculate or measure the lumped element electrical parameters are important in understanding how to use them in practice or when comparing pins.
    Lumped element equivalent model parameters are only valid when the physical Length << /10.
    When the Length is not << /10 microwave models should be used.
    7
    April, 2010
  • 8. Inductance
    8
    April, 2010
    The lowest self inductance does not mean highest bandwidth.
    High bandwidth is obtained by controlled impedance.
  • 9. Inductance
    Loop inductance is the only inductance parameter that can be measured directly.
    Self and mutual Inductance are only mathematical concepts that cannot be measured directly.
    The inductance of a single pin can be calculated and uniquely assigned to a pin.
    9
    April, 2010
  • 10. Capacitance and Resistance
    Capacitance can be modeled from measurements or simulation.
    This can be bounded by calculation as well.
    Resistance
    Is the smallest factor to model and the easiest to calculate.
    10
    April, 2010
  • 11. Spring Pin Partial Inductance
    Partial Inductance may be uniquely calculated in piecewise segments using the magnetic vector potential.
    11
    April, 2010
    A2
    A1
    M12
  • 12. 12
    April, 2010
    Spring Pin Partial Inductance and Partial Mutual Inductance Between Two Straight Parallel Segments
    A2
    A1
    M12
  • 13. Mutual Inductance and the Integration Path for Two Pins Carrying a Shared Current.
    13
    April, 2010
    I2
    I1
    The sign of the mutal inductance coefficients comes from the dot product of
  • 14. Model Topologies
    A survey of the electrical models shows an attempt to attribute inductance to length, capacitance to separation and resistivity to material effects.
    There are many lumped element models used to extract the lumped parameters of spring pins.
    14
    April, 2010
  • 15. Models
    15
    April, 2010
  • 16. Techniques For Model Fitting
    The model is best fit to the measured s-parameters.
    S11 and S21 are used to fit the impedance profile.
    The model can be used up to the condition of L<</10.
    16
    April, 2010
  • 17. Length << /10
    17
    April, 2010
  • 18. Model Output
    June 12 to 15, 2011
    IEEE SW Test Workshop
    18
    HFSS
    DATA
  • 19. Pin Placement
    At frequency the inductance of pins in parallel are not the reciprocal sum of the recipricals due to the mutal inductances.
    Due to the proximity effect current is not shared equally between all pins.
    19
    April, 2010
  • 20. Inductances from a Square Array
    20
    April, 2010
  • 21. Summary
    21
    April, 2010
    • For optimal power and ground performance. partial Inductance should be used as the main metric to compare contact engines
    • 22. Mechanically compare the length to radius ratio.
    • 23. Beware of stated inductance values in datasheets.
    • 24. When high bandwidth is desired a matched impedance is required.
    • 25. The dielectric constant of the housing becomes important.
    • 26. Z is a function of L and C.
  • Summary
    22
    April, 2010
    • Inductance should be extracted from an analytical, simulation and measurement.
    • 27. Lumped element extractions are accurate when L << /10
    • 28. The industry should develop a more standarized way to specify and determine WLCSP spring pin lumped element circuit parametrs.
  • 23
    April, 2010
    References: Clayton R. Paul, Inductance Loop and Partial
    Thank You and Have a Nice Day