high gain 2nd order high pass active filter topology

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Yes, in fact it is quite easy to do.

Think of a regular MFB filter with amplifer block with gain = 1. You have a feedback path from output to the input network in form of a capacitor. In order to provide 30dB gain, you need to move the feedback pick-off point from the output of the amp (Av=1) to a point where Av=1 regardless of amp gain, and this in fact will be the junction of the resistor network that sets the 30dB gain, i.e. the inverting input of the 'OPamp'. In order for this to work with reasonable accuracy, the feedback network needs to be low impedance, or at least the impedance as seen from the non-inverting OPamp input should be much lower than that of the filter network at the input of the amp. This is the less of a problem, the higher the gain required, and the better the drive capability of the OPamp.

For instance, assuming a NE5532 OPamp, and assuming twice the minimum load is used, the same load can be presented in the form of the feedback network. Example: if you use 600ohms as the minimum total load for the 5532, and 1.2k load of the next stage, then your feedback network must have 1.2k end-to end. For a gain of 30dB = 31.62, you come up with 1162 ohms from output to inverting input, and 38 ohms from inverting input to ground. The impedance seen at the non-inverting input is roughly equal 38 ohms, so this is perfectly acceptable for all normal values of your filter components. The remaining problem is DC offset, whch can be catered for in a number of ways, but this is outside of the scope of this topic.

I've been using this trick for, amongst other things, making simple two-way active speakers on the cheap, you use a chip power amp instead of the OPamp. Since the chip amp normally requires at least 24dB of gain to insure stability, and by definition drives low impedances, it is quite easy to use in this manner.
 
Just pretend the feedback node of the power amp is the output node of a unity gain opamp. If the lower arm feedback resistor is low enough then you won't have any problems.

If the lower arm feedback resistor is a high value, then just add a voltage divider at the output of the amp that has the same ratio as the feedback resistor pair. Sony did this around a 30dB preamp line stage, works fine.

The NAD 3020 integrated amp uses this on the main power amp input too, the lower arm feedback resistor is 27R.
 
If you think you need 30db of bass boost something is really wrong.
example: Running your mids and HF driver with 10 Watts
about 100 dB SPL. The power amp and woofer would be running 30 db higher or 10,000 Watts to get the low freq extention.
Not too realistic
 
why not just use a lowpass filter with a gain? i'm not sure i see what benefit you get with a highpass function in a feedback network.

in general, there is a balence between component values and gain. for a sallen key filter, i have the analysis done in generic terms, R, C, k, Fc, Q for a few cases. for equal value components, you can't get a gain of 4, but for unequal components you can choose almost anything. but note that in a sallen-key the component variations will cause the filter to be further from the design on paper!

3 opamp designs can often have better tolerances, but are a bit excessive.


for learning, look for nuhertz filter free (or pro or lite). these programs are excessively expensive, made more for real engineering firms over the casual user. the pro version is an excellent trial and error learning tool. you can see a passive version of a design, move it to an active implentation, or make a digital implementation. view component tolerences, ect... filter free lets you make up to 3rd order filters, making it a great tool.
 
theChris said:
why not just use a lowpass filter with a gain? i'm not sure i see
what benefit you get with a highpass function in a feedback network.

Hi,

Low pass ? at bass frequencies ?

What I was wondering was if you wanted to build a speaker with
overdamped bass and then add some bass boost and very low
frequency rejection, i.e. a 2nd order highpass with Q=2 would
give 6dB boost and low frequency rejection, could this be built
into the power amplifiers feedback network avoiding an op-amp.

Seems that you can do this, and it is a very useful technique.

:)/sreten.
 
active filters often require fixed drive impedance (usually assumed to be ~=0 )

not a great idea for a general purpose amp which may see anything from near 0 to over 1 K source Z from even desktop cd players - depending on manufacturer

throw in boutique tube preamps, transformer or resistor stepped attenuator/passive preamps and the input may go beyond 5 K and may vary with volume setting

ps: avoid Sallen-Key positive feedback filters at high gains - for typical power amplifier gains Q sensitivity is impractically high, trimming required even with 1 % components (and when did you last try to buy a 1% cap?)
 
Hi,

from what I understand you take a unity gain sallen and key filter,
but instead of taking the feedback from the output (which is the
same as the - input) you add the gain setting resistors of the
power amplifier and take the filter feedback from the - input.

As it a moderate Q 2nd order filter (Q = 2 to 3 typically) it could
be regarded as a high ripple Chebyshev, but not really the point.

You would need to ensure you don't compromise the input
impedance, i.e. make the filter sensitive to source impedance.

:)/sreten.
 
he is looking to combine the following:
1.) bass boost, boost audible bass.
2.) subsonic (aka rumble) filter, attenuate inaudible bass in hopes to reduce excursion.
3.) amplification.

all into a single design. a high Q highpass filter with gain will give a boost to some bass frequencies while rejecting lower bass frequencies and the gain allows for the amplification.


If you don't need a _good_ 2nd order function, you should be able to accomplish the other two goals easily. you could get away with a highpass on the input, followed by a highpass on the inv to gnd network, and a resistor in parallel with a (resistor - capacitor series). should be able to get a Q = 0.5 highpass, and a variable bass boost.

other variations to concider might be placing a LOW QUALITY twin-t network in the feedback, or on the input. these should be fairly component tolerant.

there are many possibilites that acheive the 3 goals stated thus far. the MFB thing might work as well.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.