Frequently Asked Questions
Probing Systems
Question - Where in the PCB design cycle do we test the differential traces for bandwidth and impedance specifications?
Answer - On the “final prototype PCB product form” where a small number of prototype boards are tested from the PCB manufacturer are verified against design specifications using visual analysis from VNA/TDR instruments. Successful results at this stage significantly increases the PCB manufacturer’s ability to accurately manufacture these boards in high-volume before releasing the design to at-speed functional product testing.
Question - What is the maximum probe bandwidth of your differential probes at a 1 mm pitch?
Answer - The maximum bandwidth of our differential probes is 70 GHz at a 1 mm pitch. We have differential probes with bandwidths from 40 Ghz to 70 Ghz and all are tested with a 1 mm pitch.
Question - Do you have any variable pitch differential probes?
Answer - Yes, the DVT30 has a 20 GHz bandwidth and the DVT40 has a 40 GHz bandwidth, maintained with a variable probe pitch of .35 mm to 1.8 mm.
Question - Do you provide samples of your probes?
Answer - A sample probe, by itself, has little practical value. Probes are generally used in pairs, and need supporting devices to be effective in PCB testing; such as positioners, to hold the probes in place and move them gently onto test pads, magnifying cameras to visually ensure planarization, and other accessories for creating a system that keeps everything at the optimum height and position for repeatable, reliable measurements.
If you have an existing wafer probe system with two probe positioners and a standard three-hole probe interface you can purchase our DVT-FFP70 probes that have a three-hole interface adapter and use them with your existing positioners. You would also have to verify that the test fixtures are large enough to hold your printed circuit boards and replace the stereo look-down microscopes with the USB cameras that we use in our systems in order to view the probe tips from the front at a 40 deg approach, because since our probes are designed to probe vertically, you would not be able to see the probe tips make contact with the test pads with a look-down microscope.
We have a no-fault return policy, yet have never had a probe or probe system returned in over 20 years of operation. We would appreciate the opportunity to provide a quotation for one of our probe systems configured to support the DVT70 70 Ghz differential probe which can be used to characterize all of your PCBs. Note that our positioners and components can also be used with wafer probes and we would be happy to consult with you on converting your current system to support our probes. Reach out to me at sales@gigaprobes.com and I'm sure we can work something out for you.
AtaiTec ISD Probe De-embedding
Question - You mention that you support 2x-Thru, 1x-Open & 1x-Short and 1x-Open / 1x-Short only. How do they compare regarding their accuracy?
Answer - The highest accuracy is 2x-Thru as it is electrically very well defined. Very close is the 1x-Open & 1x-Short, particularly in this case, as the 1x-Open and 1x-Short use the same structure (probe) and have the same lengths. 1x-Open only and 1x-Short only typically have lower accuracies, particularly 1x-Open only due to the undefined nature of an uncharacterized Open.
Question - Can I use the same file created for de-embedding measurements with an oscillosope with a VNA?
Answer - Yes you can. So, since VNAs do not start at DC; to meet the requirements of the harmonic grid, the start frequency and step frequency are typically the same and often set to 10 MHz or even higher. Because of that, it is important to have an accurate extension of the test fixture model down to the DC point. This is provided in the advanced settings of the ZNx-K220 / ISD as DC extrapolation. This is model-based and not just a numerical extension of the S-Parameters of the fixture model down to DC.
Question - Probes typically come with S-Parameter files for de-embedding. How are they measured and why do you not recommend to use these S-Parameter files to de-embed the probe?
Answer - The discontinuity at the contact point of probe tips and PCB pads is specific to the PCB structure under test. For accurate de-embedding, the specific impedance profile of the probe, including this contact point needs to be modeled. This can only be achieved when measuring the actual PCB structure under test and by applying impedance-corrected de-embedding with proper flight time scaling. This cannot be provided by a standard S-Parameter file from the probe maker. Such a file can only be a first approximation, but will not be able to de-embed the discontinuity at the probe contact point. Because of this, our probes are not shipped with de-embedding files. It is strongly recommended that customers characterize the probe on their own board. To accomplish this, we provide the ISBNN-2 (NN= 40, 50 or 70) Probe De-embedding Kit, which includes a 6-month license to Ataitec's In-Situ De-embedding software and the ISBNN ISD In-Situ board.
Question -The de-embedding procedure uses Open and Short. But aren’t these terminations less than ideal at 67 GHz (especially the Open)? Answer - Yes this is true, however, the ISD tool creates an effective 2x Through calibration standard based on these two measurements. The key requirement is that the Open and Short are very close to 180 degrees phase-offset from each other. This means that the local maximums of the ripple for the Open measurement are at the local minimums of the ripple for the Short Measurement and vice versa. This allows a good effective 2x Through to be constructed which can be used as a de-embedding algorithm. There is then no need to implement any extra 2x-Thru reference structures since de-embedding can be accomplished with only one structure and with an accuracy close to using multiple 2x-Throughs.
Question - Is there a way to model the contact impedance without using Flight Time Scaling? Answer - This requires the coupon to provide enough electrical length to include the entire part of the transmission line where the impedance is “disturbed” by the probe contact. One possibility would be to construct a 2-port Through structure with pads for the probe to contact on each side, and then a small amount of transmission line in-between. This could be used for a 2x Through coupon measurement. With this 2x Through measurement, the ISD algorithm would know to model a distance slightly past where the probe touches the board. To implement this 2x-Through, we would need to know how far we have to de-embed past the probe tips into the trace structure. To determine the appropriate 2x-Through length, you can use this general guideline: If shorter flight time scaling, 0.9xx, if longer, flight time scaling 1.xx.
Question -The de-embedding procedure uses Open and Short. But aren’t these terminations less than ideal at 67 GHz (especially the Open)? Answer - Yes this is true, however, the ISD tool creates an effective 2x Through calibration standard based on these two measurements. The key requirement is that the Open and Short are very close to 180 degrees phase-offset from each other. This means that the local maximums of the ripple for the Open measurement are at the local minimums of the ripple for the Short Measurement and vice versa. This allows a good effective 2x Through to be constructed which can be used as a de-embedding algorithm. There is then no need to implement any extra 2x-Thru reference structures since de-embedding can be accomplished with only one structure and with an accuracy close to using multiple 2x-Throughs.
Question - Is there a way to model the contact impedance without using Flight Time Scaling? Answer - This requires the coupon to provide enough electrical length to include the entire part of the transmission line where the impedance is “disturbed” by the probe contact. One possibility would be to construct a 2-port Through structure with pads for the probe to contact on each side, and then a small amount of transmission line in-between. This could be used for a 2x Through coupon measurement. With this 2x Through measurement, the ISD algorithm would know to model a distance slightly past where the probe touches the board. To implement this 2x-Through, we would need to know how far we have to de-embed past the probe tips into the trace structure. To determine the appropriate 2x-Through length, you can use this general guideline: If shorter flight time scaling, 0.9xx, if longer, flight time scaling 1.xx.