Resend:
Hi Hakim,
a bad launch can (and most propably will) ruin your deembedding, or at least
limit the upper frequency.
Just as an example imagine that your launch creates a 30dB notch at 15GHz (a
long via stub will do that for you easily). If your Measurement is done quickly
without much averaging, then your Dynamic range will be in the 50dB area (-35dB
for Time Domain Equipment) This would already mean that you DUT Loss at 15GHz
must be less than 20dB to be deembeddable. If you have two Launches of that
kind, then deembedding is not possible at all because the Loss of the Launches
is already below the dynamic range.
In addition to that a bad launch generates multiple reflections and these
aren’t deembedded because they were not present at the time you took the
measurement of the launch.
BR
Gert
From: Hakim Sellaoui [mailto:hakim.sellaoui@xxxxxxxxxxxxxxx]
Sent: Thursday, December 22, 2016 11:56 AM
To: mailto:tom@xxxxxxxxxxxxxxxxx; Havermann, Gert ;
<mailto:Gert.Havermann@xxxxxxxxxxx>
Cc: 'si-list' <mailto:si-list@xxxxxxxxxxxxx>; 'sunil bharadwaz'
<mailto:sunil_bharadwaz@xxxxxxxxx>
Subject: RE: [SI-LIST] Re: 50 ohm - RF Impedance Traces
Hi Tom,
Say you do have a bad launch, but you’re able to de-embed the fixture/SMAs
through whatever de-embedding technique, would the bad launch still have an
effect?
Can I get a copy of that pdf too?
Thanks,
-
Hakim Sellaoui
Associate Hardware Engineer
DisplayLink (UK) Limited
Tel: +44 (0)1223 443920
Fax: +44 (0)1223 443921
E: mailto:hakim.sellaoui@xxxxxxxxxxxxxxx
W: http://www.displaylink.com/
140 Science Park, Milton Road,
Cambridge, CB4 0GF,
United Kingdom.
Registered in England No. 04811048
DisplayLink Corp.
A US incorporated company.
480 S. California Avenue Ste 305, Palo Alto, CA 94306, USA.
From: mailto:si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] ;
On Behalf Of Tom Dagostino
Sent: 20 December 2016 22:48
To: mailto:Gert.Havermann@xxxxxxxxxxx
Cc: 'si-list' <mailto:si-list@xxxxxxxxxxxxx>; 'sunil bharadwaz'
<mailto:sunil_bharadwaz@xxxxxxxxx>
Subject: [SI-LIST] Re: 50 ohm - RF Impedance Traces
Sunil
It does not take much of a discontinuity in a launch to make a good trace have
poor return loss. See the enclosed pdf "tale of three launches". I compare
three different launches, a very good Ohmish discontinuity launch; one of
several Ohms and a hand soldered poor launch. The discontinuity at each end
affects both the return loss and the insertion loss. The three different
launches have widely different unable bandwidths on the identical trace, from
less than 2 GHz to greater than 20 GHz.
The two poor launches looked inductive. I've also looked at other SMA launches
that were capacitive. The same results apply.
The best launch was one designed by Scott McMorrow. The second best was an
early attempt by Molex for their connector. The worst example was a SMA
connector soldered to Scott's design. The large loop formed a large inductor.
Some say a rule of thumb for evaluating measurement systems is to look at the
difference between the insertion loss and return loss. If you have at least
10-15 dB difference you can use the system out to that limit.
I doubt that the attachment will get through the list. If you are interested in
this drop me a note.
Regards,
Tom Dagostino
971-279-5325
mailto:tom@xxxxxxxxxxxxxxxxx
Teraspeed Labs
9999 SW Wilshire Street
Suite 102
Portland, OR 97225
-----Original Message-----
From: mailto:si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] ;
On Behalf Of Havermann, Gert
Sent: Tuesday, December 20, 2016 2:01 AM
To: mailto:dmarc-noreply@xxxxxxxxxxxxx; mailto:tom@xxxxxxxxxxxxxxxxx
Cc: si-list
Subject: [SI-LIST] Re: 50 ohm - RF Impedance Traces
I think that the topology is the problem. Coplanar waveguides are not easy to
control, especially when the clearance to GND is too tight or too large, and 18
mil is really large. Just imagine that the electrical field will be active in
an area at least 3 times the GND- distance all around your trace. When there
are other traces/circuits underneath, or some mechanical parts (table, skrews,
housings....) then they all influence the Impedance.
You also need to take into account that in reality you will not see an perfect
open in the smith chart just because the trace has an open end somewhere. You
need to take the trace inductance and capacitance into account, and electrical
length also changes the "dot" smith chart. The Laminate itself is also
frequency dependent so it will be almost impossible to create an perfect open
with a long trace.
Please define " the launch does not look bad either". What is the TDR Risetime
and what kind of impedance deviations do you measure? For 6GHz you should test
with at least 100ps Risetime, and a decent launch should then vary by less than
5 Ohm. A good launch will vary in the 1ohm range. If you plan on even higher
frequencies, then the same applies for 30ps risetime. And that’s just some rule
of thumb without knowing all the details it is hard to really give a
recommendation right on the spot.
BR
Gert
-----Original Message-----
From: mailto:si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] ;
On Behalf Of sunil bharadwaz
Sent: Tuesday, December 20, 2016 8:38 AM
To: mailto:tom@xxxxxxxxxxxxxxxxx; mailto:dmarc-noreply@xxxxxxxxxxxxx
Cc: si-list <mailto:si-list@xxxxxxxxxxxxx>
Subject: [SI-LIST] Re: 50 ohm - RF Impedance Traces
Hello Tom ,
Thanks much for your response .Thanks much to all ,who have replied to this
thread.
The launch is a proper SMA connector . The final board is going to carry this
RF connector . On the TDR , the launch does not look bad either .However , one
observation is that , the larger the 50 ohm traces , worse is thereturn loss @
5 Ghz & above .
I'am checking @ Test coupon as well as the actual board .On the actual board
,the RF trace has a SMA connector on one side & the other end is open .On the
VNA , when we check the S11 @ Smith chart , ideal Transmission line should have
a dot to the right extreme indicating that the line is open .However , i see
all kinds of shapes @ frequency.
Can we get the solder mask done in normal way or do we have to take any special
care in getting the solder mask done on the RF traces ?
We generally use Co planar Waveguide traces (with ground ) , for 50 ohms lines.
Typically , theRF trace widths are 15 mil @ 18 mil distance from ground running
on either sides . We did extractthe S-parameters before sending the Layout for
Fab . Fab of course did send us the impedance report (measured on TDR ).
RegardsSunil.B
From: Tom Dagostino <mailto:tom@xxxxxxxxxxxxxxxxx>
To: mailto:dmarc-noreply@xxxxxxxxxxxxx
Cc: si-list <mailto:si-list@xxxxxxxxxxxxx>
Sent: Sunday, December 11, 2016 11:47 PM
Subject: [SI-LIST] Re: 50 ohm - RF Impedance Traces
How good is your launch into your test trace? If it has a large discontinuity
it may be masking the true impedance. Try looking at the trace with a TDR
display and see what the trace impedance is.
Tom Dagostino
Teraspeed LabsOn Dec 11, 2016 8:17 AM, sunil bharadwaz
<mailto:dmarc-noreply@xxxxxxxxxxxxx> wrote:
------------------------------------------------------------------
Hello Group ,
We seem to be running into issues while getting PCB's fabricated with 50 ohm
lines.Our Freq band of interest is 2 Ghz to 6 Ghz.Tried checking these traces
on a VNA.The return loss looks decent @ 2 Ghz , but seldom looks good at 5Ghz
& above .
The fabricator does send test coupons & impedance reports @ TDR .This does
notcover the freq response .
This is not issue with any single fab . This seems to be a general problem ,
where return loss isnot flat across freq band .
Wanted to check if there are ways to tackle this issue .
Thanks much.
RegardsSunil.B
------------------------------------------------------------------
To unsubscribe from si-list:
mailto:si-list-request@xxxxxxxxxxxxx with ;'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
mailto:si-list-request@xxxxxxxxxxxxx with ;'help' in the Subject field
List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu