Hi John,
To start with voids, they only impact reliability of they are 'champagne voids' [lots of small voids concentrated at one interface] or very large [50%+].
SJ Rel prediction contains 3 individual efforts [4+ if you have multiple loading scenarios]: 1) the creep-fatigue relationship = Engelmaier-Wild model; 2) quantification of cyclic damage [in order of increasing complexity and increasing need to make assumptions] Engelmaier delta-D, Clech hysteresis loop modeling, CALCE strain-range partitioning, FEA analysis; 3) estimation of number of cycles to the desired failure probability based on the mean cyclic life from 1), 4) cumulative damage ratio modified from Palmgren-Miner.
1) was determined from symmetrical load conditions---in your case you have to improvise based on insight, but I would suggest working on tD not Tsj as you propose [given your T(high) = 70C and T(low) = 50C I would average 6+54 to tD=30; for larger differences in T tD should tend more towards the hiT tD]; 2) will depend whether or not you ave truley compliant leads; 3) be careful not to choose a non-conservative beta [I would suggest 3.0]; & 4) may include many environments [we needed to consider the combined impact of 59 different cyclic environments fro the F-22].
Your question on accuracy makes no sense as posed, since given a sample size of at least 32 you get differences between the first and last failure by an order of magnitude. In terms of mean cyclic life, actual comparisons have shown agreements between predicted and measured20fatigue lives better than a factor of 2.
Werner
-----Original Message-----
From: Joyce Koo <[log in to unmask]>
To: [log in to unmask]; [log in to unmask]; [log in to unmask]
Sent: Tue, 31 Mar 2009 8:12 pm
Subject: Re: [TN] solder fatigue predictions for highly asymmetrical duty cycles
Does your estimate containing 25 percent allowable voids in bga? I'll keep my mouth shut and let werner go for big money :-). A lot more!
--------------------------
Sent using BlackBerry
From: Nieznanski, John A - SSD <[log in to unmask]>
To: [log in to unmask] <[log in to unmask]>; [log in to unmask] <[log in to unmask]>; Joyce Koo
Sent: Tue Mar 31 16:35:13 2009
Subject: RE: [TN] solder fatigue predictions for highly asymmetrical duty cycles
Werner and Joyce,
Thanks for the prompt feedback. My responses
are below your questions.
John N.
From:
[log in to unmask] [mailto:[log in to unmask]]
Sent: Monday, March 30, 2009 5:03
PM
To: [log in to unmask];
Nieznanski, John A - SSD;
[log in to unmask]
Subject: Re: [TN] solder fatigue
predictions for highly asymmetrical duty cycles
Hi Joyce & John,
Let's not be too hasty with a categorical NO--the real answer is: [drum roll]
'It depends.'
First, what does 'valid' mean. It needs to be understood
that any reliability prediction is just an estimate of th
e outcome. This varies
by the accuracy of quantifying the cyclic load conditions [geometry, physical
properties, temperatures, time, etc.] which in most cases are approximations
and median values.
“Valid”
in this general context means “useful for predicting or comparing outcomes,
either analytical or experimental”. This leaves the door open for either
a qualitative / comparative analysis or a quantitative analysis between
alternative implementations. The alternatives can be different hardware
configurations or different usage scenarios (thermal environments, duty cycles).
I agree and understand that the standard disclaimers apply, my mileage may
vary, and that any TechNet offer is not “valid” where prohibited by
law! ;-) However in this case, all the primary analysis parameters are known
with reasonably good accuracy. The uncertainty is in the accuracy and
applicability of the analysis method. Obviously, empirical data obtained by
testing would be best for correlating predictions with results.
Second, what is the purpose of the exercise. Are you comparing the projected
reliability relative to some other set of inputs? That can be done validly with
some insights of what is going on and what is important.
More
specifically and to the point for this discussion, the purpose of this exercise
is to attempt to quantify the “estimated”
solder joint reliability (probability
of failure) for each SMT leaded and leadless component after a defined number
of thermal cycles in specific thermal environments. Is it 10%, 1%, 0.1% or 0.01%?
The thermal environments and thermal cycles are not uniform, but have known
thermal limits, known durations and known duty cycles. All the components, geometries,
component temperatures, circuit board temperatures, and material properties are
also known.
Third, the dwell at the higher temperature is more important
because of the much higher creep rates and therefore a more complete creep
process. Thus, in an asymmetric cycle, the inputed dwell should not be just an
average of the high and low-T dwells, but needs to be judged according to the
T's involved. There have even been arguments advanced by some that because of
the slow creep rates at the lower T, it can be ignored altogether. [Not that I
agree with this argument]
For
the example with 6 minute power-on (assume system reaches thermal equilibrium after
1 minute) followed by 54 minute power off, the proposed tD = 6 minutes and proposed
Tsj = 10% T(high) + 90%T(low). So if T(high) = 70C and T(low) = 50C, then average
Tsj = 10% (70C) + 90% (50C) = 52C. Is there a better way to make a first order approximation
for the values tD and Tsj for fatigue analysis? Is there a way to
quantify the error with this approach compared to a symmetrical duty cycle? =0
A
Fourth, your 6/54 minute argument would give you somewhat of an overprediction
of the actual cycles, because the hysteresis loop will be larger because of the
54-minute dwell at T(low)--the actuals involved would determine the level of
inaccuracy.
For
the purposes of discussing analysis methods and accuracy, let’s define a “hot
case” where the maximum T(high) could be as high as 85C with the T(low)
at 50C. Similarly, there will be a “cold case”, which is just an
offset from the hot case. In the cold case, let’s say the T(high) could
be as high as 0C with the T(low) at -35C. What is the expected level of “inaccuracy”
for the “hot case” and the “cold case” assuming the
same 10% duty cycle applies in either case?
Fifth, it is these kind of judegements by which we consultants earn our keep.
Thank
you for both your charitable and paid support.
Werner
-----Original
Message-----
From: Nieznanski, John A - SSD
<[log in to unmask]>
To: [log in to unmask]
Sent: Mon, 30 Mar 2009 4:05 pm
Subject: [TN] solder fatigue predictions for highly asymmetrical duty cycles
Hello Tech-Net Gurus,
The solder fatigue parameters called out in IPC-D-279, IPC-SM-785 and IPC-9701 all assume symmetrical (i.e. 50%) duty cycles and
temperature swings. This raises the question as to how far these calculations can be pushed to predict solder fatigue in low duty cycle, variable duty cycle or high duty cycle applications.
In particular, Tsj ( = mean cyclic solder joint temperature), and tD ( = half-cycle dwell time in minutes) are both used and are directly affected by duty cycle. It is simple enough to recalculate these values based on asymmetrical duty cycles, but how do we know that the predicted solder fatigue results are still valid? If we know, do we know what assumptions are required to
make the results valid? This seems like a fairly common situation. Has anyone correlated measured results to predictions in this regard?
Specifically, there are numerous caveats in the specs including for example, high-frequency (e.g, vibration) conditions where tD < 1 second, but none of these caveats apply here.
For example, with a power on dwell time of 6 minutes (system reaches thermal equilibrium in < 1 minute) and a power off dwell time of 54 minutes (10% duty cycle), I assume I can simply use tD = 6 minute dwells and Tsj = average solder joint temperature over 60 minute cycle. Do you agree with this approach?
Another potential case of interest is if the duty cycle is reduced to something less than 1 minute (but more than 1 second; not a vibration situation), such that the stabilized operating temperature is not reached. Can any valid fatigue calculations be performed in these cases
if I can determine a valid Tsj for a 1 minute tD?
Thanks in advance for your insights and assistance.
Best regards,
John N.
________________________________
This e-mail and any files transmitted with it may be proprietary and are intended solely for the use of the individual or entity to whom they are addressed. If you have received this e-mail in error please notify the sender. Please note that any views or opinions presented in this e-mail are solely those of the author and do not necessarily represent those of ITT Corporation. The recipient should check this e-mail and any attachments for the presence of viruses. ITT accepts no liability for any damage caused by any virus transmitted by this e-mail.
---------------------------------------------------
Technet Mail List provided as a service by IPC using LISTSERV 15.0
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Technet
To temporarily halt or (re-start) delivery of Technet send e-mail to
[log in to unmask]: SET Technet NOMAIL or (MAIL)
To receive ONE mailing per day of all the posts: send e-mail to
[log in to unmask]: SET Technet Digest
Search the archives of previous posts at: http://listserv.ipc.org/archives
Please visit IPC web site http://www.ipc.org/contentpage.asp?Pageid=4.3.16 for
additional information, or contact Keach Sasamori at [log in to unmask] or
847-615-7100
ext.2815
-----------------------------------------------------
---------------------------------------------------------------------
This transmission (including any attachments) may contain confidential information, privileged material (including material protected by the solicitor-client or other applicable privileges), or constitute non-public information. Any use of this information by anyone other than the intended recipient is prohibited. If you have received this transmission in error, please immediately reply to the sender and delete this information from your system. Use, dissemination, distribution, or reproduction of this transmission by unintended recipients is not authorized and may be unlawful.
---------------------------------------------------
Technet Mail List provided as a service by IPC using LISTSERV 15.0
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Technet
To temporarily halt or (re-start) delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL or (MAIL)
To receive ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest
Search the archives of previous posts at: http://listserv.ipc.org/archives
Please visit IPC web site http://www.ipc.org/contentpage.asp?Pageid=4.3.16 for additional information, or contact Keach Sasamori at [log in to unmask] or 847-615-7100 ext.2815
-----------------------------------------------------
|
