Hi all,
I'm working on an active crossover project (2 ways, variable freq, variable slope, maybe variable Q too)
Such a project can be done in various ways, one of them is using tons of opamps. And I'm a bit reluctant to that.
For example, the input stage. I'll need something like 6 to 12dB of gain, between the source and the filter stages, in order to prevent the filter stages from working with too low signals.
What should I use for that? A classical non inverting opamp, with gain=4? A discrete opamp (still a "heavy" feedback configuration. That's the point, I'm wondering if I need this high level of feedback and "high" parts count, simply for a x4 gain stage that will (at most) output a few mAs), an open loop gain stage, or even a simple common emitter with emitter degeneration resistor?
The problem is that I have no idea what topology I can use in this situation. But in fact it's a wider topic, not only specific to this application.
Opamps ICs are often criticized here, but are discrete opamps relly better in the same applications? Is there a reason not to go for an open loop stage, or a low feedback one, instead?
I'm all ears
Alex
I'm working on an active crossover project (2 ways, variable freq, variable slope, maybe variable Q too)
Such a project can be done in various ways, one of them is using tons of opamps. And I'm a bit reluctant to that.
For example, the input stage. I'll need something like 6 to 12dB of gain, between the source and the filter stages, in order to prevent the filter stages from working with too low signals.
What should I use for that? A classical non inverting opamp, with gain=4? A discrete opamp (still a "heavy" feedback configuration. That's the point, I'm wondering if I need this high level of feedback and "high" parts count, simply for a x4 gain stage that will (at most) output a few mAs), an open loop gain stage, or even a simple common emitter with emitter degeneration resistor?
The problem is that I have no idea what topology I can use in this situation. But in fact it's a wider topic, not only specific to this application.
Opamps ICs are often criticized here, but are discrete opamps relly better in the same applications? Is there a reason not to go for an open loop stage, or a low feedback one, instead?
I'm all ears
Alex
The answer is........
There is no easy answer.
It depends on personal taste, space and cash limitations, and a whole bunch of other things that usually have little to do with science.
But......having said that:
Followers work well as buffers/filters for active crossovers. Don't take up lots of space, are cheap, and easy to make.
For some obtuse reason, the nature of which I have no idea why, op-amps configured as buffers never seem to "sound right".
So.........you could put your gain stage in front an array of followers (not that you really need that many) as a practical suggestion.
As for open loop, or closed loop...............
A gain of 12 dB is only a gain of 4. Something that you could easily do with a handful of discrete components, and no GFB, at line levels.
Seems that would have some intrinsic value as a learning experience...............
Jocko
There is no easy answer.
It depends on personal taste, space and cash limitations, and a whole bunch of other things that usually have little to do with science.
But......having said that:
Followers work well as buffers/filters for active crossovers. Don't take up lots of space, are cheap, and easy to make.
For some obtuse reason, the nature of which I have no idea why, op-amps configured as buffers never seem to "sound right".
So.........you could put your gain stage in front an array of followers (not that you really need that many) as a practical suggestion.
As for open loop, or closed loop...............
A gain of 12 dB is only a gain of 4. Something that you could easily do with a handful of discrete components, and no GFB, at line levels.
Seems that would have some intrinsic value as a learning experience...............
Jocko
Thank you for your answer, Jocko
I've been re-reading the threads about active XO here, this afternoon (Grey's ones, Fred's one, Moamp's ones...).
And I came to one conclusion (please someone tell me if I'm mistaken): I can't go with a follower in the filter stages (but I'd be pleased to can!) because there's no easy way to set the Q without an opamp.
So, I'm thinking about discrete opamps for the filter stages, a discrete input gainstage, and discrete output buffers.
Jocko, when you wrote that 12dB of gain can be easily done with a handful of components, did you mean a common emitter stage, or something more complicated?
Alex
I've been re-reading the threads about active XO here, this afternoon (Grey's ones, Fred's one, Moamp's ones...).
And I came to one conclusion (please someone tell me if I'm mistaken): I can't go with a follower in the filter stages (but I'd be pleased to can!) because there's no easy way to set the Q without an opamp.
So, I'm thinking about discrete opamps for the filter stages, a discrete input gainstage, and discrete output buffers.
Jocko, when you wrote that 12dB of gain can be easily done with a handful of components, did you mean a common emitter stage, or something more complicated?
Alex
Theory, schmeory..........
Problem is..........
They all sound different. You expect him to go out and a handful of every audio nerd approved op-amp and try them????
The kid is trying to learn something. He will learn a lot more by building a discrete gain cell than throwing a bunch of op-amps together.
Anyone can do that............just look at P**** D******.
Jocko
Problem is..........
They all sound different. You expect him to go out and a handful of every audio nerd approved op-amp and try them????
The kid is trying to learn something. He will learn a lot more by building a discrete gain cell than throwing a bunch of op-amps together.
Anyone can do that............just look at P**** D******.
Jocko
Well, it depends what you're trying to do. The best discrete designs will outperform op-amps, with lower THD, wider bandwidth etc (yes P.Lacombe, you need to go back and look at those theories, as they don't say there will be no difference!). Of course good op-amps already offer extremely good performance, so don't expect huge benefits, or even truly audible ones at all.
On the other hand, they are much more complex, requiring many times more components. You can build much simpler discrete amps, but they will not come close to op-amp performance. However, they would still be good to build for learning purposes, as Jocko Homo suggested. Having said that, apparently some people prefer the distorted sound offered by some of the various simple designs.
As for feedback. In my opinion there is never a reason to go without overall negative feedback. You will get better perfomance by doing so. However, as above, apparently some people prefer the more distorted sound offered by some designs so you'll have to experiment for yourself.
On the other hand, they are much more complex, requiring many times more components. You can build much simpler discrete amps, but they will not come close to op-amp performance. However, they would still be good to build for learning purposes, as Jocko Homo suggested. Having said that, apparently some people prefer the distorted sound offered by some of the various simple designs.
As for feedback. In my opinion there is never a reason to go without overall negative feedback. You will get better perfomance by doing so. However, as above, apparently some people prefer the more distorted sound offered by some designs so you'll have to experiment for yourself.
Here are a few suggestions from recent experience designing the Virgin Electronics boomtube ex(tm). These apply to both integrated and discrete op-amp-based crossovers. For the buffer stage, you will probably be stuck with using a non-inverting buffer configuration set to the gain you require. The reason is that you will need to avoid loading or interaction with the previous stage. For an inverting buffer to achieve this, it would require large value resistors that can introduce excessive noise. For the crossover filters, you should try to use the multiple feedback (MFB) configuration. This configuration avoids the common-mode distortion of the Sallen-Key configuration. However, it uses more components and is somewhat harder to design. There are other differences between the two configurations and you might take some time to look into them. Among the many other well-known references, I recommend TI's filter ap notes. TI also has a good filter design program, FilterPro. Try the following TI docs:
SBFA001A - FilterPro MFB and Sallen-Key Low-Pass Filter Design Program
SLOA024B - Analysis of the Sallen-Key Architecture
SLOA049B - Active Low-Pass Filter Design
SLOA062 - Using the Texas Instruments Filter Design Database
SLOA065 - Filter Design on a Budget
SLOA093 - Filter Design in Thirty Seconds
SLOA096 - More Filter Design on a Budget
SBFA001A - FilterPro MFB and Sallen-Key Low-Pass Filter Design Program
SLOA024B - Analysis of the Sallen-Key Architecture
SLOA049B - Active Low-Pass Filter Design
SLOA062 - Using the Texas Instruments Filter Design Database
SLOA065 - Filter Design on a Budget
SLOA093 - Filter Design in Thirty Seconds
SLOA096 - More Filter Design on a Budget
Mr Evil said:
Thanks, instructive link.
Nelson's discrete opamp tutorial is also very nice
http://www.passdiy.com/pdf/diyopamp.pdf
gembler said:
You forgot this one
SLOA088 - Active Filter Design Techniques
But please, keep this thread on the feedback discuttion
I have seen this one -- it is actually Chapter 16 by Thomas Kugelstadt, from Ron Mancini's Op-Amps for Everyone. It contains good background info but the other references may be better starting points.
Sorry if I went OT but in my defense I was focusing on your objective of designing a crossover and not the bigger issue of whether to use feedback or not. The MOR advice is of course "a moderate amount of feedback is OK", and I tend to agree with this in general. Unfortunately I do not have references as readily available as the previous ones for feedback design, although Op-Amps for Everyone probably has some good info. However, the feedback question contains so many variables that it is difficult to make any valid general recommendations. Also, crossover design is interlinked with feedback issues which are in turn interlinked with implementation (discrete vs. discrete op-amp vs. integrated op-amp). OT or not OT, that is the question...
do you want to use some active filters, which need opamps, or something like buffer isolated passive filters + gainstage
http://home.tu-clausthal.de/~tpa/space/axover.gif ?
http://home.tu-clausthal.de/~tpa/space/axover.gif ?
till said:
First, my active XO consists of a 2 way, variable frequency, slope and Q sallen key filter
I'd like to use followers (simple and clean
) but this is giving me headaches.Look at the equations for K (gain) =1. I haven't found a way to easilly change Q without altering the other parameters (easilly means: without having to change all other part's value)
So I think I'll go for discrete opamps, unless I find an easy way to have the variable Q with the followers. (and I need it, simply for the variable order since for 2nd order LR Q=0.5, 3rd order butterworth Q=1, and 4th order LR Q=0.707 (for both))
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.