TINA/Spice/OPA2189: OPA2189 BPF design

webench_design_4592932_12.tscHi Kai,

The phase of using webench  for BPF 25KHz is 0 degree, but import to TINA-T finds phase has shift, not 0 degree

Hello Tseng,

It is normal for a simple op amp amplifier circuit, or a circuit as complex as an active filter, to incur non-zero phase shift at all but dc. Even at the lowest of frequencies there may be a fraction of a degree of phase shift occurring from input to output. Therefore, it is not practical to expect an active filter to have 0 degrees of phase shift at a specific frequency unless you are dealing with a special class of filters called All-pass filters. They are different in that they are designed for specific phase shift, or time delay characteristics, but they have nearly flat amplitude response over most of their usable frequency range. They can't provide the low-pass, band-pass, etc. responses that filters are most commonly used to provide.

Note in your TINA circuit that you are specifying the BUF634 in the multiple-feedback active filter stages. The BUF634 has a gain of less than 1 V/V, and is intended to be employed with an op amp to provided higher output current than can be obtained from an op amps alone. Thus, operational amplifiers having high open-loop-gain and appropriate bandwidth must be employed in the active filter stages. The OPA2189 titled in this e2e is an example of an op amp usable in an active filter.

Regards, Thomas Precision Amplifiers Applications Engineering

Yes Tseng, this seems like a very aggressive filter Q wise in the stages. You might consider an extremely capable topology shown in this tread - also, why use a precision amp like the OPA2189 when BP filters are by definition AC coupled? What do you think your highest Q stage is - if over 10, you have will production problems. And, as Tom says, MFB filters need to implemented with a VFA op amp, not a buffer like the BUF634 - did the webench tool deliver with that device?

e2e.ti.com/.../788559

Hi sir,

Yes, BUF634 is recommended by webench. I just type BP filter parameters as I mentioned to webench. Anyway, can you teach me how to design 8 order BP filter base on webench?

Hello Tseng,

It is unfortunate that the filter tool directed you to the BUF634; that is an error. If you are using the New TI Filter Design Tool, it is a beta version of the program and the "bugs" are still being discovered. Hopefully, by time the beta testing time is complete the Webench team will have the bugs cleared out.

Often, an 8th-order filter is difficult to realize based on the passive component tolerance issues that Kai pointed out. About the best tolerances components that are available within reason from distributors are 0.1% resistors, and 1 % capacitors. Even with components having those tolerances a practical filter may still have a somewhat distorted response from the ideal. Lower tolerance components may be available, but their costs become too expensive for many applications.

I suggest for now that you consider using an older TI filter tool, FilterPro. It is easy to use and doesn't make any op amp suggestions. However, it does provide a key op amp parameter "Minimum GBW (Gain Bandwidth) reqd" in the filter design results. Having that minimum GBW allows one to select op amps from the TI amplifiers web pages that will have sufficient bandwidth to support the specific filter response. You find the GBW listed on the final FilterPro schematic screen by scrolling the filter's schematic image upwards.

You can access and download TI's FilterPro v3.1 here:

http://www.ti.com/design-tools/signal-chain-design/webench-filters.html

There used to be some training materials for FilterPro on line, but it looks like they have all been removed. That shouldn't be a problem because the application of FilterPro is very straightforward. If you are really serious about synthesizing an 8th-order bandpass response, just check the "Optional - Filter Order" box to "8" and progress through the screens in FilterPro.

Regards, Thomas

Precision 

Hey tom,

I tried that link and I don't think it takes you to the older Filterpro tool. It takes you to the new front end GUI for filter design selection. I do have the old Filterpro which is super useful for common filter stage designs Q and Fo. But it always way overestimated the required GBP.

I also alerted the team the BUF634 should not be in these solutions - all BUF parts will be removed.

Also, for the highly selective BP implied here, why not try that Ackerberg-Mossberg topology in this recent thread - I was amazed how well it worked and how low the spread was in MonteCarlo.

https://e2e.ti.com/support/amplifiers/f/14/t/788559

Can you point to an actual orderable 0.1% low cost R?

Hi Michael,

If you scan down the FilterPro page you find a line that lists:

Download FilterPro v3.1

Following the screens it has you log in and eventually you come to:

I didn't reload FilterPro V3.1 because I already have it on my machine.

Yes, the GBW calculation in FilterPro and the other TI active filter programs is in some cases are very conservative, but that is intentional. Certainly, a highly knowledgeable filter designer can use advanced methods to more precisely determine the minimum op amp GBW required to provide accurate filter responses.  However, for most users providing a conservative GBW assures that they will be able to successfully obtain the desired response while investing minimal solution time.

Indeed there are more sophisticated active topologies requiring more then one op amp per stage, that easily achieve the response needed for high-Q band-pass filters. It looks like the Ackerberg-Mossberg topology requires 3 op amps per stage, or 12 total for the 8th-order response. One of my favorites is the Dual-amplifier band-pass (DABP) discussed in the Williams "Electronic Filter Design Handbook." I have recommended that topology a couple of times here on the our amplifiers e2e forum. It is a high-performance band-pass filter topology that requires two op amps per filter stage.

I suppose "low-cost" is a relative term, and I do tend to look at the cost in relation to what one paid for a similar component just a decade ago. Here's an example of a 1 kilohm, 0.1%, 25 ppm thin-film resistor from Mouser:

https://www.mouser.com/ProductDetail/IRC-TT-Electronics/PCF-W0603LF-03-1001-B-P-LT?qs=sGAEpiMZZMu61qfTUdNhG1Flup5Au%252BywTu6ghkpJc0Y%3D

There may be lower cost options out there.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Sir,

Use webench to design the 4th BP filter, which recommends using INA1620, but it does not have library in TINA-TI.

Can you suggest the right product according to the attached file? Thank you.

TINA-TI's Version 9.3.200.277

filter-design-report-20 (1).pdf

Hi Tseng,

you will find the TINA-TI reference design on the product page:

I have run the simulation for you:

tseng.TSC

Kai

Yes Kai, that was a really interesting thread - I saw where Paul was suggesting a +/-30% GBP spread. For parts that have trimmed supply currents (like all of the recent higher speed parts) I use +/-15% on the GBP spread.

In any case, on this thread that INA suggestion out of Webench is a odd - normally you would just use a dual op amp.

I cranked out a design jumping between several different tools - I actually have had very little BP active filter experience, so a lot of this is new to me - a lot of LP however, - just for curiosity, I will see what a single stage Ackerberg-Mossberg design will give me here. My Filterpro seems to have stopped working, so I could not use it for the design - and the webench results seem a 1/2 bubble off.

25kHz 4th order BP suggestions initial MFB.docx

TLC2272 25kHz BP 4th order.TSC

Hi Michael,

Here is the link to the recent e2e thread where I provided information about the Dual amplifier band-pass (DABP) filter:

The Electronic Filter Design Handbook is where I originally found the topology. Arthur B. Williams, the handbook author, provided a reference to the original article, Sedra, A. S., and Espinoza, J. L. "Sensitivity and Frequency Limitations of Biquadratic Active Filters," IEEE Transactions on Circuits and Systems, Vol. CAS-22, No. 2, February 1975.

Regards, Thomas

Precision Amplifiers Applications Engineering

Some results with the AckerbergMossberg approach.

the app note was aimed at other things, some of which did not seem to matter on just a basic filter - so the circuit simplifies a little.

Briefly, the 1/RC with all those equal R and C sets F0

The R across the first stage C set the Q

The input C is an independent tune for peak gain - just ratioing from where you start to where you want works well

Added all of to this word file, and the very low cost RC4558 seemed to work pretty well - that is 3MHz, probably could use a 5MHz part in this design if you can find one - one key to BP, is it is probably a mistake to spend anything on DC precision. Also, I would say it seems this topology holds a tighter spread than MFB 2 stage.

3441.25kHz 4th order BP suggestions initial MFB.docx

AcerbergMossbert BP with RC4558 25kHz redesign 4th order.TSC

Hi sir,

I run your design(OPA1678) in TINA-TI, AC transfer is okay, but transient result have some problem as attached file. 

Can you help to check it?

I also have a lot of problems running transient, get kind of random results depending on time length, not sure - the V7 version of the RC4558 is attached 4213.AcerbergMossbert BP with RC4558 25kHz redesign 4th order.TSC

Morning all,

Well, as I said at one point I normally only get LP filter questions and virtually no BP questions - so it was time to explore this DABP approach.Wow, how cool and simple is this? It can't generate more than a 2V/V peak gain, but otherwise seems pretty simple and powerful. One thing to keep in mind is these designs generate exact RC at first, and then you need to step them off to E96 R's and E24 C's - that always give a shift in the nominal response from your design goal - then you apply tolerances on those for spread.

Here I assembled the targets 2 stages for your design Tsing and looked at it in detail - OPA1678 implementation attached as well - So only 2 duals now and looks as good as the Akerberg Mossberg - that one can generate easily tuned wide range of peak gains though.

4th order BP using the DABP topology.docx

4th order DABP with OPA1678.TSC

Hi Michael,

here are my results of your last simulation. (I replaced the current source by a voltage source, for clarity, and added a measuring point to the output):

tseng1.TSC

Kai

Hello Tseng,

I hope the collective responses from Kai, Michael and myself provide sufficient information for you to apply the DAPB filter effectively in your application. If so, please go ahead and close this e2e inquiry.

Regards, Thomas

Precision Amplifiers Applications Engineering