AnsweredAssumed Answered

Issue with Mixer IP3 In Spectrasys

Question asked by bgoyet on Oct 25, 2006
Latest reply on Oct 31, 2006 by rulonv

Like • Show 0 Likes0
Comment • 3

The standard mixer model in Spectrasys is giving erroneous results for IP3. The attached file has a mixer with specified IIP3 of 27 dbm. The simulation, both cascade / table and and manual calculation from the spectral output gives 29.4 dbm.

This is with version 2006.10.

I have attached the file.

[file]mixer ip3 error.wsx[/file]

Attachments

mixer ip3 error.wsx187.5 KB

No one else has this question

Outcomes

3 Replies

rulonv Oct 27, 2006 5:21 AM
Mark Correct
Correct Answer
Thank you for the post. We did find an issue with only the 'Basic' mixer model when intermods were created in the mixer due to a multicarrier input. Current intercept calculations are correct through mixer in the single carrier case or even a multicarrier case where intermods created in the mixer are not the limiting case.

This issue has already been fixed internally and will ship in the next release. For those that can't wait for the next release please contact support.

The work around is to change the 3rd and 2nd order intercept points of the mixer. The 3rd order intercept point of the basic mixer (ONLY) can be lowered by 2.385 dB and the 2nd order intercept can be lowered by 3.01 dB. The intermod power levels on the graph and the OIP and IIP measurements will be correct. However, the SOIP and SIIP measurements (user entered values) will obviously not correspond to the calculated values.

The 'Double Balanced Mixer' and 'Intermod Table' mixers do not exhibit this problem.
Like • Show 0 Likes0
Actions
bgoyet Oct 30, 2006 3:19 PM
Mark Correct
Correct Answer
Thanks Rulon.
The model response seems to be variable with load, so the correction factor may not be a simple offset.
I have attached a file with a 27 dbm / 7 db loss mixer: 20 dbm OIP3.
The circuit sample with a pad give 1.5 dB higher OIP3 than input, but the mixer driving a duplexer (gives just about the right answer).
Looking forward to the corrected model. When is the next release?
Regards,
Bill

[file]mixer_pad_and_duplexer_load.wsx[/file]
Attachments
- mixer_pad_and_duplexer_load.wsx294.6 KB
Like • Show 0 Likes0
Actions
rulonv Oct 31, 2006 4:07 AM
Mark Correct
Correct Answer
That is correct. The answer will change based on the load impedance since the intermod voltage at that node is dependent on the load impedance. We all know that spreadsheets assume a constant impedance for any frequency going through a filter (or any other device) ... which is really not true.

You can enable the check box "Add Powers, Voltages, and Impedances to the Path Dataset' on the 'Path' tab of the system analysis to see the exact impedance seen by the intermod by looking for the variable 'Z10_Total' in the path dataset ... it is not close to 50 ohms like the attenuator case.

The inband impedance of the filter is dependent on the ripple. If you change the ripple to say 0.0001 dB you will be much closer to 50 ohms and the OIP3 answer will be much closer to the 50 ohm case. When you get the duplexer to be 50 ohms ... the answer is exact between both cases.

As can be seen this is the real power of SPECTRASYS ... because the real world isn't a constant impedance everywhere.
Like • Show 0 Likes0
Actions

Issue with Mixer IP3 In Spectrasys

Attachments

Outcomes

Attachments

Related Content

Recommended Content