The usual procedure for realizing active loudspeakers is to add an active crossover network device between the pre-amplifier output and the inputs of each power amplifiers.
However, this is always associated with the disadvantage that additional semiconductors (transistors or in most cases operational amplifiers in the signal pad) add their own sound signature. This disadvantage is audible only in the upper frequency area, so that this approach would be of no advantage for subwoofers but of great advantage for the high frequency range especially with very high resolution tweeters.
What about the approach to implement the associated filter network in the already exist power amplifier (and not in additional introduced gain stages) as describe under the below URLs ?
TDA2030 active speaker audio systems circuit with 60 watt output power
and page 12/23 under
https://www.st.com/resource/en/datasheet/cd00000129.pdf
Are there any experiences ?
I have implemented this approach only once on a "ZEN" - go to
ZEN include active crossover without additional OP AMP for ultimate sounding PHL1230
but have recorded a considerable increase in sound quality compared to the normal procedure of using before (i. e. a separate device for the active crossover network equipped with OPA627 and Sallen-Key topology).
However, this is always associated with the disadvantage that additional semiconductors (transistors or in most cases operational amplifiers in the signal pad) add their own sound signature. This disadvantage is audible only in the upper frequency area, so that this approach would be of no advantage for subwoofers but of great advantage for the high frequency range especially with very high resolution tweeters.
What about the approach to implement the associated filter network in the already exist power amplifier (and not in additional introduced gain stages) as describe under the below URLs ?
TDA2030 active speaker audio systems circuit with 60 watt output power
and page 12/23 under
https://www.st.com/resource/en/datasheet/cd00000129.pdf
Are there any experiences ?
I have implemented this approach only once on a "ZEN" - go to
ZEN include active crossover without additional OP AMP for ultimate sounding PHL1230
but have recorded a considerable increase in sound quality compared to the normal procedure of using before (i. e. a separate device for the active crossover network equipped with OPA627 and Sallen-Key topology).
Last edited:
I believe these are called "Passive Line-Level" filters - a network not using any active devices and placed upstream of the power amp. If the filter uses the power amp itself as the active device, then you just have something like a "power op-amp" based filter.
These things are quite possible to construct and not all that new/novel. The problem is that they are not transferable from amplifier to amplifier because often each amp has a different input impedance. The PLL filter must be designed to operate into that (constant) load impedance and doesn't' work the same into a different load.
To design a PLL filter, just find the input impedance for your amplifier and then use standard design tables/equations or do impedance scaling of an existing design.
These things are quite possible to construct and not all that new/novel. The problem is that they are not transferable from amplifier to amplifier because often each amp has a different input impedance. The PLL filter must be designed to operate into that (constant) load impedance and doesn't' work the same into a different load.
To design a PLL filter, just find the input impedance for your amplifier and then use standard design tables/equations or do impedance scaling of an existing design.
Sometimes folks design an amp to have limited bandwidth to do this. For example, I think Nautibuoy, is making a choke loaded SE Class A SIT amp with a high pass around 150Hz or so by using a small undersized output coupling cap. This will then be used on the full range driver for the top and I assume some other more powerful amp with a low pass filter will be used for the woofer. Very similar in concept to a passive line level filter but taken to the next level where the amp itself is frequency modified.
This has the advantage that a non specific source can be used as the impedance of the filter is not tied to the source. In a way it is sort of like a filter at the speaker level but installed in the amplifier vs the speaker cabinet and multiple amps are used.
You can also design speaker cabinets using drivers that have a natural roll off to make a woofer that is low passed at say 150Hz. You can drive it full range but it will only give output up to a certain point. A subwoofer combined with a TL can achieve this as tortuous walls of the TL can absorb the higher frequency stuff that might escape a direct radiator.
So a very elegant way to do this is a TL based bass unit driven full range by amp 1. And a full range top driven amp 2 fitted with a limited bandwidth (high pass cap coupled SE Class A) for the top. Match the frequency by trimming the cap and adjusting the levels.
This has the advantage that a non specific source can be used as the impedance of the filter is not tied to the source. In a way it is sort of like a filter at the speaker level but installed in the amplifier vs the speaker cabinet and multiple amps are used.
You can also design speaker cabinets using drivers that have a natural roll off to make a woofer that is low passed at say 150Hz. You can drive it full range but it will only give output up to a certain point. A subwoofer combined with a TL can achieve this as tortuous walls of the TL can absorb the higher frequency stuff that might escape a direct radiator.
So a very elegant way to do this is a TL based bass unit driven full range by amp 1. And a full range top driven amp 2 fitted with a limited bandwidth (high pass cap coupled SE Class A) for the top. Match the frequency by trimming the cap and adjusting the levels.
Last edited:
The TDA2030 circuit is not a passive line level filter, but a Sallen and Key filter that uses the main amplifiers. With passive line-level filters, you either have to accept that you can't make any complex poles and in fact not even entirely coincident real poles, or use big inductors. You haven't got such a limitation with Sallen and Key.
In some cases amplifiers and op-amps that rely on gain combined with feedback for low distortion seem to have reduced performance when they are given a capacitive load. You may even have situations where the 'better' the op-amp, the worse it gets, precisely because of its claim to fame for having low distortion.
No threadjacking intended, but may I take the liberty to ask almost the same question?
Can we safely insert the filtersection in the feedback path of a poweramp only (just like active the RIAA topology/ or any frequency dependant fb network) , instead of having a dedicated active filter section at line level before the power amp?
Suppose I would like to use e.g a LM 3886 for bass duties only (say 20-500 Hz) , could I then safely wrap an electric 3th order filter section in the feedback path of that LM3886, or would I run into stability issues?
Can we safely insert the filtersection in the feedback path of a poweramp only (just like active the RIAA topology/ or any frequency dependant fb network) , instead of having a dedicated active filter section at line level before the power amp?
Suppose I would like to use e.g a LM 3886 for bass duties only (say 20-500 Hz) , could I then safely wrap an electric 3th order filter section in the feedback path of that LM3886, or would I run into stability issues?
It is possible, but not ideal. Whether the amplifier remains stable, depends on what you place in the feedback loop. Modifying the feedback loop of a power amplifier, however, might reduce negative feedback. Negative feedback [in class B solid state amplifiers] should be maximized to suppress harmonic distortion.
Last edited:
Then I get only 6db/oct but easy to realize by replace already present parts.
This is only a good approach for a mixed topology, i. e. active and passive crossover network units, where as the active part is only to be regarded as pre-filtering for reducing the voltage swing a little at power amps' input in the not used frequency areas.
At that time Plessey's SL403 was one of the first integrated circuits for small/medium power amps:
SL 403, Tube SL403; Rohre SL 403 ID53072, IC - Integrated Ci
Last edited:
I suppose you meant keeping feedback within the design parameters?
Not to bring up a tired old debate but in simulations I often found that there is such a thing as "too much" negative feedback, as it seems to 'flatten' the harmonic response, disproportionately adding high harmonics. THD really needs to be modified with something like a Fletcher Munsen curve to make more sense. It also doesn't take into account peak vs average power, which can vary significantly based on changes in phases. Eg. Harmonics can be aligned to produce sharp high amplitude peaks despite having low power.
But yeah, in class b amplifiers. And adding capacitance to a feedback loop can make things messy quickly.
I kind of have to toss in the "get real" flag. Do you realize how many stages of electronics your music has been through before you get it? Older recordings likely dozens of 5558 Op amps or worse. Newer ones, better. Newest ones more keeping in the digital domain. But still. One can build vanishingly low distortion active filters (Or buy from X-Kitx cheap). Yes, every stage degrades, but I would spend my money on speakers and better power amps as that is what matters.
Please read Self and others on why just using a smaller input blocking cap may not be wise unless you go to a high quality film. There is a reason that unusually large value electrolytics are used. It is a free and cheep subsonic HP if done correctly. I believe all power amps should be AC coupled with subsonic filters to prevent excess cone excursion. I am also a fan of ultrasonic filters, though not of VAS dominant pole filtering.
Filtering in the feedback is possible, but do simulate in Spice first as it is too easy to make a small mistake in compensation and blow up a lot of outputs. A lot of amps do not have as much margin as we may hope for. MOSFET very tricky.
I am sure everyone has their take on this, but that is mine. Results matter more than opinions.
Please read Self and others on why just using a smaller input blocking cap may not be wise unless you go to a high quality film. There is a reason that unusually large value electrolytics are used. It is a free and cheep subsonic HP if done correctly. I believe all power amps should be AC coupled with subsonic filters to prevent excess cone excursion. I am also a fan of ultrasonic filters, though not of VAS dominant pole filtering.
Filtering in the feedback is possible, but do simulate in Spice first as it is too easy to make a small mistake in compensation and blow up a lot of outputs. A lot of amps do not have as much margin as we may hope for. MOSFET very tricky.
I am sure everyone has their take on this, but that is mine. Results matter more than opinions.
Many years ago I tried this idea with a LM3886.
Results was an amplifier that was oscillating very nicely.
I would expect that any other kind of amplifier would have the same issues.
I kind of missing the point what is against an actual proper active filter?
Since it has more advantages than disadvantages.
As long as you make sure the filter stage performs one or two orders of magnitude better than the power amplifier, it can be seen as non-existing from a signal point of view.
Which isn't so difficult to achieve anymore these days.
Results was an amplifier that was oscillating very nicely.
I would expect that any other kind of amplifier would have the same issues.
I kind of missing the point what is against an actual proper active filter?
Since it has more advantages than disadvantages.
As long as you make sure the filter stage performs one or two orders of magnitude better than the power amplifier, it can be seen as non-existing from a signal point of view.
Which isn't so difficult to achieve anymore these days.
Last edited:
Putting the reciprocal of the desired response in the feedback network doesn't make much sense because the response of the feedback network would have to increase at an nth order rate to get an nth order roll-off, but making a Sallen and Key stage or a multiple feedback filter stage using the main amp is possible, as shown by the examples linked to in this thread.
My questions were only asked from the perspective of both curiosity (and simplicity) and have nothing to do with signal quality. And yes, designing even the most complex active filter circuitry with non-standard transfer functions is a breeze in VituixCad. But that is not at all the point here.