Yeah, I've just been looking on the Farnell website, and you can get 20nF film caps with a 2.5% tolerance very cheaply. What kind of power supply is usually used to power this kind of circuit? I'm thinking of a transformer with 2 x 0-12 outputs to give me 12 - 0 - -12 output. Then feed this into 2 x bridge rectifiers and use caps on the output for smoothing. Is this the best way or is there something simpler?
There is no simpler way, but you should use a voltage regulator after the smoothing caps.
A transformer with 2x12 VAC will give you ~16-17 VDC, even more, when the load is low. You can also build a split supply with a single rectifier.
It is also possible to add the voltage regulator to the power supply you already have and save the additional transformer, rectifier, smoothing caps, fuses, etc. You need to watch the heat dissipation then, because the voltage difference might be quite high. So you will either need heatsinks for the regulators or add a power resistor in series between the amp power supply and each regulator or cascade several regulators. The resistor is the cheapest and easiest method.
A transformer with 2x12 VAC will give you ~16-17 VDC, even more, when the load is low. You can also build a split supply with a single rectifier.
It is also possible to add the voltage regulator to the power supply you already have and save the additional transformer, rectifier, smoothing caps, fuses, etc. You need to watch the heat dissipation then, because the voltage difference might be quite high. So you will either need heatsinks for the regulators or add a power resistor in series between the amp power supply and each regulator or cascade several regulators. The resistor is the cheapest and easiest method.
Thanks pacificblue, never thought of using resistors and using my existing supply. So I just worked out if I used 120ohm resistors from my 30v rails, thats a max current of .25A and 7.5 watts. 7.5 watts should be easily enough to power the active filter. So 120ohm 10watt resistors and +12 and -12 regulators seems a nice and cheap way to power the filter.
Choose your topology first. Then choose your opamp and if need be adjust the topology.
Then choose your supply voltage to suit the opamp and to suit the maximum signal + dB signal overhead. Finally choose the regulator and transformer/supply.
You may decide to use +-5Vdc to suit a low voltage rail to rail opamp or to use +-21.5Vdc with 44V opamp to maximise the signal overhead.
Then choose your supply voltage to suit the opamp and to suit the maximum signal + dB signal overhead. Finally choose the regulator and transformer/supply.
You may decide to use +-5Vdc to suit a low voltage rail to rail opamp or to use +-21.5Vdc with 44V opamp to maximise the signal overhead.
Last edited:
You will need a few V more before the regulator than you can get out of them. E.g. a 12 V regulator will need more than 15 V at its input to work properly. Assuming your 30 V rails are ideally stable the resistor must drop 15 V which means 120 Ohm will work for up to 125 mA. That leaves you with 1,5 W after the regulators and 1,875 W of power dissipation across the resistor.
That question usually leads to religious wars in the Forum.
It depends on money, how sensitive you react to the quality differences of op amps in the circuit and your taste. There are people who don't mind having any amount of NE4558 in their audio chain, others spend more on a single op amp than on the transformer or build their own op amps from transistors. In a subwoofer filter you may not even notice the difference, because the human ear is less sensitive at very low frequencies.
From a filter point of view you want low offset and high input impedance which leads to op amps with FET inputs. From the list on the schematic that means the TL072 followed by the LF3x3. For better audio quality, look at the OPA2134 which is more expensive, but reasonably so.
Why don't you test the circuit with TL072 first? If you find any issues with the audio quality due to the op amps, upgrade to OPA2134, first the input buffers and the satellite section, then if you are still not satisfied the subwoofer section as well.
It depends on money, how sensitive you react to the quality differences of op amps in the circuit and your taste. There are people who don't mind having any amount of NE4558 in their audio chain, others spend more on a single op amp than on the transformer or build their own op amps from transistors. In a subwoofer filter you may not even notice the difference, because the human ear is less sensitive at very low frequencies.
From a filter point of view you want low offset and high input impedance which leads to op amps with FET inputs. From the list on the schematic that means the TL072 followed by the LF3x3. For better audio quality, look at the OPA2134 which is more expensive, but reasonably so.
Why don't you test the circuit with TL072 first? If you find any issues with the audio quality due to the op amps, upgrade to OPA2134, first the input buffers and the satellite section, then if you are still not satisfied the subwoofer section as well.
Hi pacificblue, thanks again for your input. I've found the OPA2134 chips on the Farnell website for £2.54 TEXAS INSTRUMENTS|OPA2134PA.|OP AMP, DUAL AUDIO FET I/P, 2134 | Farnell United Kingdom
which seems reasonable, and I'll only need 5 of them, so I think my budget can stretch that far.
What is the main advantage of using an active filter like this. Could I not cut down on a lot of components in the signal path by feeding the left channel into 2 x 3886 chips and the right channel into 2 x 3886 chips, and then using one left and one right channel feeding my 3 inch full range speakers and the other left and right channels feeding 2 small sub woofers, with passive filters. Kind of bi-amping I suppose.
If I were to do this, how would I split the original signals to the 2 LM3886 chips. Would I need op amps as buffers? Or should I stick with an active filter?
which seems reasonable, and I'll only need 5 of them, so I think my budget can stretch that far.
What is the main advantage of using an active filter like this. Could I not cut down on a lot of components in the signal path by feeding the left channel into 2 x 3886 chips and the right channel into 2 x 3886 chips, and then using one left and one right channel feeding my 3 inch full range speakers and the other left and right channels feeding 2 small sub woofers, with passive filters. Kind of bi-amping I suppose.
If I were to do this, how would I split the original signals to the 2 LM3886 chips. Would I need op amps as buffers? Or should I stick with an active filter?
Subwoofers need low cross-over frequencies. That means for a passive filter you need big component values which makes them expensive. The resonant frequencies of the drivers will be located near the cross-over frequency, so the speaker impedances will have peaks there and make a theoretic calculation of a passive filter difficult. Flattening the impedance with passive components to counteract that is again expensive due to the necessary big component values. It will take some Trial-&-Error which may be fun, but with expensive components not necessarily desirable.
On top of that you need two subwoofers instead of one, if you don't build a passive mixer, which is more complex.
An active filter works into a nearly resistive load and is therefore easy to calculate. You will still need to do some Trial-&-Error, but the components are small and even good quality components are relatively inexpensive.
On top of that you need two subwoofers instead of one, if you don't build a passive mixer, which is more complex.
An active filter works into a nearly resistive load and is therefore easy to calculate. You will still need to do some Trial-&-Error, but the components are small and even good quality components are relatively inexpensive.
Thanks again pacificblue.
I've designed the active crossover to be put onto Veroboard, but I'm pretty busy with other stuff at the moment, so I probably won't get chance to build the crossover and my new amp till later in the year. I want to use my knowledge of PIC programming to build a 2.1 amp with IR remote volume pot, selector and use laser cut aluminium for the front and back.
I'll keep you posted on the progress.... when I have time to start it
I've designed the active crossover to be put onto Veroboard, but I'm pretty busy with other stuff at the moment, so I probably won't get chance to build the crossover and my new amp till later in the year. I want to use my knowledge of PIC programming to build a 2.1 amp with IR remote volume pot, selector and use laser cut aluminium for the front and back.
I'll keep you posted on the progress.... when I have time to start it
I've just been looking at the active subwoofer circuit that pacificblue posted a link too and I wanted to know if I am working this out right.....
1. The L and R channels have a gain of 1 (what goes in comes out)
2. The mixed channels have a gain of 1 up until the last op amp
3. Now if gain = 1 + Rf / Ri, if the pot if turned to one extreme you have 1K for Rf and 6K for Ri. So gain = 1.16. If the pot is turned to the other extreme you have 6K for Rf and 1K for Ri, the gain is now 7
If I have worked out part 3 correctly and a 1 volt signal is applied to L and R, the output from the sub channel would be 14 volts. Is this correct? and if so, to connect this to a LM3886 with a supply voltage of 35, the maximum gain I can use is 5? I tried this on the overture design guide and it says that the circuit may have stability trouble if the gain < 10V/V.
Thanks again for everyone's input
1. The L and R channels have a gain of 1 (what goes in comes out)
2. The mixed channels have a gain of 1 up until the last op amp
3. Now if gain = 1 + Rf / Ri, if the pot if turned to one extreme you have 1K for Rf and 6K for Ri. So gain = 1.16. If the pot is turned to the other extreme you have 6K for Rf and 1K for Ri, the gain is now 7
If I have worked out part 3 correctly and a 1 volt signal is applied to L and R, the output from the sub channel would be 14 volts. Is this correct? and if so, to connect this to a LM3886 with a supply voltage of 35, the maximum gain I can use is 5? I tried this on the overture design guide and it says that the circuit may have stability trouble if the gain < 10V/V.
Thanks again for everyone's input
Another thought....
If the LM3886 chip needs to have at least 10 V/V gain, would it be better to set the gain of the last op amp in the subwoofer circuit to a fixed value (lets say 2) and then use a pot on the output as a volume control before it enters the LM3886? I know this would still clip in certain scenarios of inputs, but it would at least be a stable circuit.
If the LM3886 chip needs to have at least 10 V/V gain, would it be better to set the gain of the last op amp in the subwoofer circuit to a fixed value (lets say 2) and then use a pot on the output as a volume control before it enters the LM3886? I know this would still clip in certain scenarios of inputs, but it would at least be a stable circuit.
The last gain stage of the subwoofer section is inverting, so the gain is Rf/Ri which means you can actually get gains below 1. In this case the gain can be set between 0,17 and 6 times. The output for 1 V on both inputs would be between 0,34 and 12 V.
The stability criteria for the LM3886 are only valid for its own part of the gain. The gain before it must not be added in the Design Guide. The LM gain must still be within the range of 10-50 for stability. The preamp or crossover do not affect that.
Ideally you set the gains so that the resulting noise is as low as possible. Multiply the input noise of the OPA2134 by its own gain, then by the LM3886's gain and add the LM3886's input noise by its own gain. Test which combination of OPA2134 gain and LM3886 gain that results in the desired overall gain leads to the lowest output noise or to the highest acceptable noise. It will probably not be much of an issue for a subwoofer either way.
The stability criteria for the LM3886 are only valid for its own part of the gain. The gain before it must not be added in the Design Guide. The LM gain must still be within the range of 10-50 for stability. The preamp or crossover do not affect that.
Ideally you set the gains so that the resulting noise is as low as possible. Multiply the input noise of the OPA2134 by its own gain, then by the LM3886's gain and add the LM3886's input noise by its own gain. Test which combination of OPA2134 gain and LM3886 gain that results in the desired overall gain leads to the lowest output noise or to the highest acceptable noise. It will probably not be much of an issue for a subwoofer either way.
Thanks again pacificblue for your help. I re-read this article...Op-Amps and found the Rf/Ri equation.
Here are a few specs on my speakers I will be using...
2 x TangBand W3-871 (8ohm, 1W/1m - 87dB)
1 x Monacor SPH-200CTC (2 x 8ohm, 1W/1m - 88dB)
The sub will be connected parallel to give 4 ohms. I am going to drive the sub with 2 LM3886 in parallel as it will be a 4 ohm load and the chips will share the current.
I want my 2.1 amp to have a main volume pot and also a pot to adjust the sub (I listen to a lot of different music and want to be able to adjust it when I need). I think a gain of 6 is very high for the last part of the sub woofer circuit, so what if I have a main volume pot at the start of the whole active filter circuit, set the gain of the last part of the subwoofer circuit to 2 (Rf=2K, Ri=1K) and then used a pot on the output of the subwoofer part of the circuit before it goes into the power amp.
I know I'll need to tweak a lot of things, but I am trying to get a rough idea of what I need to do. Does what I have suggested seem reasonable, or are my calculations wrong?
Here are a few specs on my speakers I will be using...
2 x TangBand W3-871 (8ohm, 1W/1m - 87dB)
1 x Monacor SPH-200CTC (2 x 8ohm, 1W/1m - 88dB)
The sub will be connected parallel to give 4 ohms. I am going to drive the sub with 2 LM3886 in parallel as it will be a 4 ohm load and the chips will share the current.
I want my 2.1 amp to have a main volume pot and also a pot to adjust the sub (I listen to a lot of different music and want to be able to adjust it when I need). I think a gain of 6 is very high for the last part of the sub woofer circuit, so what if I have a main volume pot at the start of the whole active filter circuit, set the gain of the last part of the subwoofer circuit to 2 (Rf=2K, Ri=1K) and then used a pot on the output of the subwoofer part of the circuit before it goes into the power amp.
I know I'll need to tweak a lot of things, but I am trying to get a rough idea of what I need to do. Does what I have suggested seem reasonable, or are my calculations wrong?
Why don't you use a single chip amp per woofer coil? You save on the load sharing resistors and the tight resistor tolerances in the feedback loop you would need in a parallel configuration, but the overall result will be the same.
Once the woofer is set up correctly, there is usually no need to adjust its volume depending on different music. That is only necessary when the woofer does not integrate well with the satellites.
If a gain of six times appears too high to you, the easiest solution is to not set the potentiometer to that value. Another easy solution is to use different resistor/potentiometer values in the feedback loop of the last gain stage and adapt the setting range to your likings. Your idea to set a fixed gain of two times and add a potentiometer between cross-over and power amp is also viable, but means one more component in the signal chain that does not give any obvious advantage.
Once the woofer is set up correctly, there is usually no need to adjust its volume depending on different music. That is only necessary when the woofer does not integrate well with the satellites.
If a gain of six times appears too high to you, the easiest solution is to not set the potentiometer to that value. Another easy solution is to use different resistor/potentiometer values in the feedback loop of the last gain stage and adapt the setting range to your likings. Your idea to set a fixed gain of two times and add a potentiometer between cross-over and power amp is also viable, but means one more component in the signal chain that does not give any obvious advantage.
Thanks again for your input.
So instead of a fixed gain of 2 and a pot at the end of the circuit, I could use 2 x 10K resistors and a 10K pot (Alps motorized pots are available in 10K). This would give me a gain of 0.5 up to 2. I could always change the 10K resistors for different values if it needed tweaking.
The reason I wanted to parallel the subwoofer and not use each coil separately is incase I use the amp with a different subwoofer in the future, which hasn't got the option of dual coil.
You mention 'load sharing resistors' and 'tight resistor tolerances in the feedback loop', I understand that the feedback loops would have to be matched in a parallel config, but what do you mean by 'load sharing resistors'?
So instead of a fixed gain of 2 and a pot at the end of the circuit, I could use 2 x 10K resistors and a 10K pot (Alps motorized pots are available in 10K). This would give me a gain of 0.5 up to 2. I could always change the 10K resistors for different values if it needed tweaking.
The reason I wanted to parallel the subwoofer and not use each coil separately is incase I use the amp with a different subwoofer in the future, which hasn't got the option of dual coil.
You mention 'load sharing resistors' and 'tight resistor tolerances in the feedback loop', I understand that the feedback loops would have to be matched in a parallel config, but what do you mean by 'load sharing resistors'?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.