Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Software Tools SPICE, PCB CAD, speaker design and measurement software, calculators

FFT mystery?
FFT mystery?
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 23rd January 2019, 03:37 PM   #11
bugbear is offline bugbear  United Kingdom
diyAudio Member
 
Join Date: Jul 2016
Location: Norfolk/Suffolk border
I had no FFT windowing selected (as per my guide page.

Distortion measurements with LTSPICE | Audio Perfection

)

This appears to make a big difference, although I'm not sure what the difference is, or what windowing is "good".


BugBear
  Reply With Quote
Old 23rd January 2019, 04:03 PM   #12
bugbear is offline bugbear  United Kingdom
diyAudio Member
 
Join Date: Jul 2016
Location: Norfolk/Suffolk border
Hmm. I think I need to do some general reading up about FFT.

It's becoming obvious I have no clue what the parameters I'm bashing at actually mean.


BugBear
  Reply With Quote
Old 24th January 2019, 09:51 AM   #13
bugbear is offline bugbear  United Kingdom
diyAudio Member
 
Join Date: Jul 2016
Location: Norfolk/Suffolk border
OK; I have managed to get a FFT I consider plausibly meaningful.


It looks like there's some kind of quite slow (around a second) settling in process somewhere in the circuit.


By running the sim for a whole second, but only sampling in the last 15 ms, even the DC coupled output gives a plausible result.


Since I'm running with a small (100n) max time interval, this takes a long time to simulate.

There is still much I don't understand, but this seems to be progress, albeit halting.



FFT and model file attached.


BugBear
Attached Images
File Type: jpg fft.jpg (185.7 KB, 78 views)
Attached Files
File Type: asc rotel850.asc (10.1 KB, 4 views)
  Reply With Quote
Old 24th January 2019, 12:18 PM   #14
jan.didden is offline jan.didden  Europe
diyAudio Member
 
jan.didden's Avatar
 
Join Date: May 2002
Location: Westende Resort, BE coast
Quote:
Originally Posted by scott wurcer View Post
You need to force the time steps to be the samples for the FFT otherwise the transient analysis can interpolate causing this effect.


.param num_cycles 20
.param num_fft_pts 2048
.param freq 1000
.param timestep {1/((num_fft_pts-1)*freq)}
.tran 0 {(num_cycles+1)/freq} {1/freq} {timestep}
You also have to set the FFT points in the FFT trace selection window to the same number of points as you specified, and the timestep to the correct value.

I am using the attached which is equivalent to your stuff Scott, and then set the FFT data points to 0.5M see attached.

Also set the AC source freq to {freq}!

dlycyc and numcyc are the number of cycles to wait to start logging data and the number of cycles over which the FFT is done.

Jan

Jan
Attached Images
File Type: png fftparams.PNG (8.8 KB, 76 views)
File Type: png fftdatapoints.PNG (18.8 KB, 78 views)
__________________
Cable: a potential source of trouble interconnecting two other potential sources of trouble - Erik Margan
Linear Audio pubs and articles . The SilentSwitcher now at diyaudio store SilentSwitcher. Keeping in touch with SY.

Last edited by jan.didden; 24th January 2019 at 12:20 PM.
  Reply With Quote
Old 24th January 2019, 01:56 PM   #15
bugbear is offline bugbear  United Kingdom
diyAudio Member
 
Join Date: Jul 2016
Location: Norfolk/Suffolk border
Quote:
Originally Posted by jan.didden View Post
You also have to set the FFT points in the FFT trace selection window to the same number of points as you specified, and the timestep to the correct value.

I am using the attached which is equivalent to your stuff Scott, and then set the FFT data points to 0.5M see attached.

Also set the AC source freq to {freq}!

dlycyc and numcyc are the number of cycles to wait to start logging data and the number of cycles over which the FFT is done.

Jan

Jan
I modified my model/simulation to use this approach, but the FFT reverted to that of my initial post.


BugBear
Attached Images
File Type: png fft.png (31.3 KB, 64 views)
Attached Files
File Type: asc rotel850.asc (10.0 KB, 5 views)

Last edited by bugbear; 24th January 2019 at 02:03 PM. Reason: add images
  Reply With Quote
Old 27th January 2019, 03:55 AM   #16
cogsncogs is offline cogsncogs  United States
diyAudio Member
 
cogsncogs's Avatar
 
Join Date: Jul 2003
Location: Wayne, West Virginia
Here's what I used:
Code:
.options plotwinsize=0
.options numdgt=7
Code:
.param input=1
.param freq=1k
.param numcyc=40
.param dlycyc=20
.param FFT=2**18
.param simtime=(dlycyc+numcyc)/Freq
.param dlytime=dlycyc/Freq
.param numsampl=simtime/Freq/((simtime/numcyc)*FFT)
.four {Freq} V(op)
.tran 0 {simtime} {dlytime} {numsampl}
Attached Images
File Type: png rotel850-fft.png (36.9 KB, 43 views)
File Type: png rotel850-fftdialog.png (18.5 KB, 16 views)
__________________
Sent from my Linux desktop computer using a keyboard and a mouse.
  Reply With Quote
Old 27th January 2019, 03:58 AM   #17
cogsncogs is offline cogsncogs  United States
diyAudio Member
 
cogsncogs's Avatar
 
Join Date: Jul 2003
Location: Wayne, West Virginia
Also changed
C601 to 10000uF
C607 to 1F
to get a flatter line.
__________________
Sent from my Linux desktop computer using a keyboard and a mouse.
  Reply With Quote
Old 27th January 2019, 08:08 AM   #18
steveu is offline steveu  United States
diyAudio Member
 
Join Date: May 2005
Location: Colorado Springs
Interesting problem. It seems to be a windowing issue and a very slight phase lead at the pure_ac_op node makes it worse (I added a loading resistor). A ~5uS delay in the input source or 5uS start time for the FFT makes an improvement but better is to use a "Blackman" window and compare V(op) to V(ip)*22.

This simulation file has the models included:
Attached Files
File Type: asc rotel850.asc (12.9 KB, 2 views)
  Reply With Quote
Old 27th January 2019, 08:13 AM   #19
jan.didden is offline jan.didden  Europe
diyAudio Member
 
jan.didden's Avatar
 
Join Date: May 2002
Location: Westende Resort, BE coast
Quote:
Originally Posted by bugbear View Post
I modified my model/simulation to use this approach, but the FFT reverted to that of my initial post.


BugBear
Well the blue curve (input signal?) seems quite alright, so the method looks fine.
It may be that the FFT accurately reports the actual output signal.

Jan
__________________
Cable: a potential source of trouble interconnecting two other potential sources of trouble - Erik Margan
Linear Audio pubs and articles . The SilentSwitcher now at diyaudio store SilentSwitcher. Keeping in touch with SY.
  Reply With Quote

Reply


FFT mystery?Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Mystery i.c. ErikR Parts 6 14th January 2016 12:15 PM
Mystery PCB GringoAudio Parts 10 11th December 2011 07:16 PM
There is some mystery in the air... smartx21 Everything Else 5 24th July 2010 12:54 AM
little 6.5 mystery ... oface Full Range 6 5th November 2007 12:20 AM
mystery... mista2 Multi-Way 1 14th January 2005 05:20 AM


New To Site? Need Help?

All times are GMT. The time now is 12:53 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki