Circuits / Info on Tracking Notch Filters

[JohnW]

Circuits / Info on Tracking Notch Filters

I would like to add a Tracking Notch filter say 20Hz to 20KHz on the front end of a PC based ADC to improve the measurement dynamic range – something like the notch filters in the Audio Precision / HP 8903 etc.

The notch “Depth” need only be say 30dB to 60dB as its function is only to extend measurement dynamic range – but should be as narrow as possible to allow accurate measurement results.

The intention is to develop a cheap USB based PC system for audio measurement – once the board is designed I would post on the site for others – If there is significant interest I could also offer PCB’s both bare and pre – assembled / Tested as it will be heavily SMD and undoubtedly use difficult to source components.

I’m thinking of designing a system where one channel of the ADC is used “Pre filter” to allow measurement of input RMS - setting the correct input reference level, and the second channel is used “post Notch Filter” to display the increased dynamic range FFT / THD measurements.

As the unit would also generate the original test tone frequency, this could be used as the reference frequency to set the “tracking” notch.

A second mode, which would be less accurate due to noise and harmonics, is to use a “Zero” crossing detector - this would allow tracking of any external Source.

Does anyone have or know where I could find such circuits?

Cheers,

John

[ghost_au]

Originally posted by JohnW
Circuits / Info on Tracking Notch Filters

I would like to add a Tracking Notch filter say 20Hz to 20KHz on the front end of a PC based ADC to improve the measurement dynamic range – something like the notch filters in the Audio Precision / HP 8903 etc.

The notch “Depth” need only be say 30dB to 60dB as its function is only to extend measurement dynamic range – but should be as narrow as possible to allow accurate measurement results.

The intention is to develop a cheap USB based PC system for audio measurement – once the board is designed I would post on the site for others – If there is significant interest I could also offer PCB’s both bare and pre – assembled / Tested as it will be heavily SMD and undoubtedly use difficult to source components.

I’m thinking of designing a system where one channel of the ADC is used “Pre filter” to allow measurement of input RMS - setting the correct input reference level, and the second channel is used “post Notch Filter” to display the increased dynamic range FFT / THD measurements.

As the unit would also generate the original test tone frequency, this could be used as the reference frequency to set the “tracking” notch.

A second mode, which would be less accurate due to noise and harmonics, is to use a “Zero” crossing detector - this would allow tracking of any external Source.

Does anyone have or know where I could find such circuits?

Cheers,

John

Hi John, you could try using a state variable notch filter, the centre frequency of these can be adjusted by varying the loop gain, so if you sample the quadrature outputs, which are not seriously attenuated, with a phase detector (this will produce a +/- dc output as an error signal) you should be able to derive a control voltage which can then control the state variable filter loop gain, possibly via an led/ldr cell. it should be possible to lock the centre frequency to the test signal quite accurately. I think I have a circuit of a distortion meter designed by Robert Cordell around 20 years ago that used this approach, though from memory the quadrature detector he used is no longer produced, however there are many other better IC's available now. By the way, it is possible to design a state variable notch filter with an asymmetric ( elliptic ) response so that the attenuation in the passband is around 0.1dB for the second harmonic, without having to maintain a very high Q. Have a look at Ian Hickman's article on p. 627, "Electronics World" August 1999.

regards from downunder

[JohnW]

Hi Ghost_au,

Thanks for the lead – since I posted the requested I also came across Roberts 3 part article on building a THD analyser, this and downloading the service manual for the HP8903 (about 40Mb in total) has given me a better idea of how to proceed.

Looking at the high value resistors (typically 10K) used in the state Variable filter in both designs, suggest a high noise floor – which will limit the dynamic range of any FFT measurements.

I’m hoping to extend the –140dB noise floor of the AKM5394A (24Bit ADC) to about say –170dB (or the limits of thermal noise). Also notching the fundamental test tone should / will reduce ADC distortion components.

I use the R&S UPD05, but its noise floor ref. to 0dB is limited to about –130dB without notch and –155dB with notch. The units ADC and front end Alias very badly with any high frequency noise – such as from Noise shaped DAC’s and switching amplifiers. The AP AUX0025 switching amplifier filter does not help as it has less then 10dB of attenuation at 100KHz.

Anybody have the Circuit for AP’s BP / notch filter on the SANL board?

Cheers,

John

[JohnW]

I've been doing some more digging and found the following AP patents: -

5,818,240
Method and system for measuring signal transfer characteristics in the presence of WOW, flutter and intermodulation distortion

5,748,001
Method and apparatus for fast response and distortion measurement

5,649,304
Method and apparatus for communicating auxiliary information in a measurement signal

5,475,315
Method and apparatus for fast response and distortion measurement

5,420,516
Method and apparatus for fast response and distortion measurement

5,265,201
Master-slave processor human interface system

5,247,458
Method and apparatus for testing a digital system for the occurrence of errors

5,136,267
Tuneable bandpass filter system and filtering method

5,089,981
Hybrid form digital filter

4,631,522
Method and circuit for compensation of a multiplying digital-to-analog converter

4,614,914
Low-distortion transformer-coupled circuit

4,563,652
Dual-stage filter with feedback

As typical with US company’s, they seem to Patent the obvious - in an attempt I guess to increase the IP value of the company, see 5,265,201 Master-slave processor human interface system for a good example.

Anyway, Patent 4,563,652 Dual-stage filter with feedback seems to describe the AP Notch filter - and expires in less than 1 year