THS4551: BW of Butterworth Filter

Vincent Chen64988

Part Number: THS4551

Hi Team,

My customer is using THS4551 as a attenuate differential amplifier to driver SAR ADC. He has some questions below and need team's comment.

1. Below is a 100KHz filter, is there a document or speadsheet to guide customer how to calculate out the R, C value to achieve the bandwidth that customer want?

2. Below is customer's circuit, R&D would like to form a 25KHz butterworth filter, so is the R, C value of black line area correct? if not, please provide the recommendation.

3. Why need a differential and common mode filter in blue line area?

4. Can customer large the R, C value in blue line area to form a main filter(like to build a 25KHz filter)? If it is not recommended, why?

5. Customer try to change C210 from 3.3nF to 470nF, the -3dB should be at 1/(2*3.14*20*470n)=16.9KHz but his test result is about 8.9KHz, why? And he also found the total power consumption increase about 1W, does it make sense?

Thanks very much.

Vincent Chen

over 5 years ago

0 Michael Steffes over 5 years ago

Guru 48395 points

Boy you sure got a lot going on there Vincent, kind of the stuff I have the tools for - that datasheet circuit does not look like the one I did originally, slight changes in RC values that might have hurt the noise. No, actually I guess this is the one I did, being an attenuator it looked a little odd to me, - will take a while to assemble a response - but where the RC values come from is from my MFB filter design tools iterated over the last 10years - The starting point (long ago) for that was this app note, but enhanced considerably since,

http://www.ti.com/lit/an/sboa114/sboa114.pdf

0 Michael Steffes over 5 years ago in reply to Michael Steffes

Guru 48395 points

Here you go Vincent, some answers and a sim file,

THS4551 25kHz filter discussion.docx

Gain of 0.2 25kHz 3rd order filter with OPA192.TSC

0 Michael Steffes over 5 years ago in reply to Michael Steffes

Guru 48395 points

One of the important changes here is you will notice the feedback caps go outside that imbedded 20ohm R value - I later found that is much more effective isolating the reactive open loop Zout from the feedback C.

0 Vincent Chen64988 over 5 years ago in reply to Michael Steffes

TI__Expert 4480 points

Hi Michael,

Thanks a lot for the descriptions and simulation result. I found you pasted a excel calculation form in the TINA simulation file, could you please share it with customer? My customer needs the file to calculate out R,C values since that attenuate raion is 0.357 and maybe he will need to change the cutoff frequency.

Thanks very much.

Vincent Chen

0 Michael Steffes over 5 years ago in reply to Vincent Chen64988

Guru 48395 points

Sorry Vincent, that file is no more available than if you were to ask ADI to provide their calculation file (which I also think is pretty good).

0 Vincent Chen64988 over 5 years ago in reply to Michael Steffes

TI__Expert 4480 points

Hi Michael,

Due to the calculation equation of filter is pretty complex that customer doesn't have ability to design a butterworth filter. Do you have any advice for the question? I think we cannot always leverage your effort for different filter design.

Thanks a lot.

Vincent Chen

0 Michael Steffes over 5 years ago in reply to Vincent Chen64988

Guru 48395 points

Indeed it is complicated, well

1. The TI online filter designer uses the the Filterpro design code I think - circa 1989, plausible but primitive

2.The ADI tool delivers pretty good solutions I think - one of the input variables I have is approximate min input R to avoid loading the prior stage - that also plays into final noise profile - not sure they let the designer input that- but they do appear to be tuning the noise peaking down and adjusting for GBP - probably the best public tool currently.

0 Vincent Chen64988 over 5 years ago in reply to Michael Steffes

TI__Expert 4480 points

Hi Michael,

Thanks very much for the advise. There is another question need your comment.

Customer's circuit is similar to below diagram, let input floating and no supply voltage on Vocm pin. Then the voltage of below red circle should be about 2.5V, right?

It is strange that customer feedback he measured that point but the voltage is about 1.5V, do you have any idea what might cause this problem? What experiments that we do to verify this issue?

Thanks a lot.

Vincent Chen

0 Michael Steffes over 5 years ago in reply to Vincent Chen64988

Guru 48395 points

So the Vocm input pin defaults to mid supply if floated, that puts 2.5V on the output pins if single 5V supply is used

That output CM voltage divides back to the input by the DC resistor divider path there - so normally lower depending on input source DC voltage. Never probe on the summing junctions, hopefully you are talking sim numbers.

Amplifiers

Amplifiers forum

THS4551: BW of Butterworth Filter