• State Resolved
  • Locked Locked
  • Replies 15 replies
  • Subscribers 49 subscribers
  • Views 664 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

THS4551: 8th active filter quality factor calculation

Aykut ULUSAN
Intellectual 365 points
Part Number: THS4551
Other Parts Discussed in Thread: OPA659, THS4531, THS4531A

Hello, 

I have a differential 8th order active filter (It includes 4 x 2nd order active filters) and four quality factor. (Q1,Q2,Q3,Q4).

How can I find or calculate totally quality factor of 8th order active filter? 

Thanks.

  • 0
    Guru 47535 points

    So the old Filterpro program would allow you to specify 8th order, what type of filter and frequencies. 

    There are also some vendor programs that will do this for you if you don't have filterpro. 

  • 0 in reply to Michael Steffes
    Intellectual 365 points
    Michael, thank you for your reply.

    I designed the filter and it is working good but just
    I wonder are there simple calculation method for quality factor of all filter?

    For example, Q1=1,5, Q2=Q3=Q4=3.5

    how can calculate quality factor of all filter as simple?
  • 0 in reply to Aykut ULUSAN
    Guru 47535 points
    actually no, these Q by stage come from lookup tables for the different filter types. So I wrote the THS4551 data sheet and did a lot of MFB filter designs with that. Put a lot into the model to head off issues, what type of filter are you doing? There a tons of app notes that list out the Q by order and type, do you need a reference.
  • 0 in reply to Michael Steffes
    Guru 47535 points
    And actually, I might add, the Q by stage in academia is a bit off - they are all assuming a gain of 1 overall where ascending Q limits clipping - however, with any gain at all in the design a descending Q design reduces integrated noise.
  • 0 in reply to Michael Steffes
    Intellectual 365 points
    Dear Michael,

    Sorry, Maybe my question is very simple, I read a lot of documents but I don't find total Q or BW practical calculations of cascaded filters.
    if you can share me any document, I would be glad.

    I'm using MFB filter, Frequency:70khz, Gain:1, four stage and
    Q of first stage:1.5 others :about 3.5
    BW of Stage 1: 48Khz (+/- 24Khz)
    BW of Stage 2-4: 20Khz (+/- 10Khz)

    Actually, I wonder change of BW. (because of Q is related to BW )
    stage by stage, how is change of BW for 70Khz?

    First stage : should be from 46Khz to 96Khz (70Khz-24Khz =46Khz 70Khz+24Khz=94Khz)
    how can I calculate for other stages( for 2,3,4 stage)?

    thanks in advance
  • 0 in reply to Aykut ULUSAN
    Guru 47535 points
    Sounds like you are doing a bandpass?
  • 0 in reply to Michael Steffes
    Intellectual 365 points
    Yes, bandpass filter.
  • 0 in reply to Aykut ULUSAN
    Guru 47535 points

    8th order bandpass options in filterpro.docxI am not as familiar with BP designs, but I think the Q of any stage is Fo/Fbandwidth

    Doing an 8th order in Filterpro with a couple of different shape targets centered on 1kHz, gives the Q's in the attached file -

  • 0 in reply to Michael Steffes
    Intellectual 365 points

    Dear Michael, Thanks a lot,

    I used chebyshev 0.5dB for narrow BW.

    I think BWn = Qn / BW(n-1) 

    For example,

    BW1 = 48Khz (first stage BW or second stage input BW)

    Q2=3,5  (second stage Q value)

    BW2= BW1/Q2 = 48/3,5=14Khz (second stage output BW)

    BW3=14/3.5=4Khz

    BW4=4Khz/3.5=1,2Khz

  • 0 in reply to Aykut ULUSAN
    Guru 47535 points

    Morning Aykut,

    Again, I have very little active BP design experience - but a lot of THS4551 familiarity - if I work through what you are saying you are trying to do and step through the Filterpro solution it gives me something else entirely in stage targets - that is in the attached word file where I also attached the filterpro file - not sure what the current TI online active filter designer will deliver. The final RC circuits with the FDA fromfilterpro look pretty problematic - way too high of R values for a 150Mhz part like the THS4551 - I believe you are trying to hit

    70kHz center frequency

    48kHz BW centered on that with 0.5dB Chebychev

    Gain of 1 midband.

    This is pretty aggressive - I am interested to see how close the THS4551 could hit these - but I don't think I could use the RC out of Filterpro, might try a different tool and see if does any better. I have 3 other tools I could try to use from different sources. Mostly they just give Fo and Q targets, but one of them will try to give RC as well.

    Apparently it won't let me attach the Filterpro design report, sorry,

    4382.70kHz center 8th order BP filter design in Filterpro.docx

  • 0 in reply to Michael Steffes
    Guru 47535 points

    Incidentally, I did get the current TI online filter design tool to generate this
    1. It is the same solution as filterpro, so using that code still
    2. It does say it needs much higher GBP product than Filterpro and initially proposes the OPA659
    3. Apparently has a download toTINA feature but then says it will use an ideal op amp as there is no model for the OPA659 - there is actually, but not sure what happened there,
    4. Tried to download and open in TINA, not successful yet there.

  • 0 in reply to Michael Steffes
    Guru 47535 points

    So I did get far enough into one of these other tools and got a first stage that confirms what I thought (in TINA),

    Q = Fo/(delta F-3dB span) for that stage.

    Topologically though, what I am seeing at low freqeuncies is that big attenuator with high R values at the input -running an input noise sweep showed pretty high input noise. 
    Again, I have little experience with active BP filters - but just curious, do you have any RC values for a THS4551 stage yet, and what it is supposed to be doing -I would be interested in running some sims including LG phase margin. The stated shape for the stage I was simulating with just an op amp was met surprisingly well.

  • 0 in reply to Michael Steffes
    Guru 47535 points
    Hey Aykut,

    This little discussion is pretty good where it eventually says the overall Q of a bandpass filter (your original question) is again the center frequency/F-3dB span.

    Now I think they define the center as the geometric mean of the two F-3dB points. in your your case that is 46k to 94kHz. so the overall Filter Q according the definition in here would be

    67k/48kHz = 1.4. So this is a pretty wide filter made up of higher Q stages. Folks also do this as a high pass low pass combination.

    www.electronics-tutorials.ws/.../filter_7.html
  • 0 in reply to Michael Steffes
    Guru 47535 points
    Morning Aykut,

    I did run your requirements through some other tools - one thing that came out was one of the tools thinks you can do this design with a 15Mhz GBP part? If so, you might be able to use the THS4531 in this and save power. In another thread, we are trying to figure out if that model, or the THS4531A updated model, is more accurate - not sure yet.

    I am noticing the designs all seem to produce a very peaked output noise - that is normal, but I do wonder if some RC solutions are better than others in that respect.
  • 0 in reply to Michael Steffes
    Intellectual 365 points

    Dear Michael, 

    Many thanks for your help.

    Final BW is not 48Khz, BW of first 2nd filter is 48Khz.

    I didn't use filterpro because of very big resistors. I calculated MFB filter in excel reading many MFB filter docs and I checked it with signal generator, It's working very well.

    Final BW is a few khz. Biggest resistor is 16.2K and smallest is 332R. C=1nf.

    Thanks again.