I have achieved good results in the past without DC blocking caps on HF filter outputs nor series resistors. I might increase C1 just on principle, not because it is really necessary.
I think leaving the previous R28 is a good idea. While most commercial op amp circuits don't use it because it doesn't do much, it will help to balance the currents/voltages at the input terminals and thus reduce output offset voltage if it's value is the input and feedback resistor in parallel. However, with most new op amp bias voltages and currents are so low this resistor cannot help much. And with the smaller feedback resistors you have made the output offset will be lower. It's up to you. If you have room, I would include it just for peace of mind.
As for op amps, the NE5532 is rather cheap and has a flat FR but not so much in the way of sparkle. The OP275 is wonderful IMO. The bass is a bit full and perhaps bloated, but it is very clear and sweet in the top end and also sounds great when used just for bass. Only in full range systems have I noticed the slightly heavy sound it has to the bass. I haven't used the BB op amps suggested but I have read that BB op amps are closed over the vocal range. I think this is really a matter of choice to take our suggestions as just that - suggestions. And perhaps you can use a socket and test some different op amps out before you buy?
I think leaving the previous R28 is a good idea. While most commercial op amp circuits don't use it because it doesn't do much, it will help to balance the currents/voltages at the input terminals and thus reduce output offset voltage if it's value is the input and feedback resistor in parallel. However, with most new op amp bias voltages and currents are so low this resistor cannot help much. And with the smaller feedback resistors you have made the output offset will be lower. It's up to you. If you have room, I would include it just for peace of mind.
As for op amps, the NE5532 is rather cheap and has a flat FR but not so much in the way of sparkle. The OP275 is wonderful IMO. The bass is a bit full and perhaps bloated, but it is very clear and sweet in the top end and also sounds great when used just for bass. Only in full range systems have I noticed the slightly heavy sound it has to the bass. I haven't used the BB op amps suggested but I have read that BB op amps are closed over the vocal range. I think this is really a matter of choice to take our suggestions as just that - suggestions. And perhaps you can use a socket and test some different op amps out before you buy?
I'm putting together an order at Digikey... All these little things are adding up. I'm not ordering anything yet. I'm waiting to hear any other suggestions on the design.
Also I have to figure out what I need for a power supply. I have a 12-0-12 48VA transformer I can possibly use. I have +/- 18.3VDC after rectification with no load. This circuit shouldn't draw enough current to make that drop much. Should the supply be regulated? If so, should I go with +/-12V or +/-15V regulators?
Also I have to figure out what I need for a power supply. I have a 12-0-12 48VA transformer I can possibly use. I have +/- 18.3VDC after rectification with no load. This circuit shouldn't draw enough current to make that drop much. Should the supply be regulated? If so, should I go with +/-12V or +/-15V regulators?
Second most comments...
Things to consider further:
- Linkwitz transform from published values or measured ones? SL on his site describes somewhere how to measure Q0 and F0 in a simple way.
- Op-amps: even TL082 might do a very good job. I used BB OPA2134, a good "luxury" type, and after blowing some due to PSU failure I replaced them as a quick fix with TL072 for the LP section. Nothing happened to the sound - no subjective change at all. HP may be different. All these FET input op amps have little bias current so R28 will do little, but hey, technically it is correct to leave it in.
- DC HP: surely can't hurt at the LP ouput.
- series R: yes, in series, recommended to prevent op amp oscillation, at the very end of the chain (last op amp). Anything from 47 to 200 Ohms will do the job. Typical is 47 or 50.
- PSU: on the safe side a 12-0-12 will yield 12 V regulated. 15V may be possible but likely unnecessary. If your amps have a gain of just 10x (20dB), and say you have 40 V rails (a ca. 100 W into 8 Ohms power amp), the power amp will clip at 1.4 V signal voltage. 12 V, 15 V, hey, even 9 V may work for the PSU. Well, technically you may reach higher values interstage but likely not 12 V if final max necessary output is 1.4 V.
- construction: whatever op amp you choose, read the data sheet - it will tell you the necessary decoupling cap values, for once, and more. Proper decoupling and short PSU lines after regulation may be THE key to good performance, and IMO easily beats choice of capacitor brand in the influence on good sound. Also, good layout.
Good luck!
MBK
Things to consider further:
- Linkwitz transform from published values or measured ones? SL on his site describes somewhere how to measure Q0 and F0 in a simple way.
- Op-amps: even TL082 might do a very good job. I used BB OPA2134, a good "luxury" type, and after blowing some due to PSU failure I replaced them as a quick fix with TL072 for the LP section. Nothing happened to the sound - no subjective change at all. HP may be different. All these FET input op amps have little bias current so R28 will do little, but hey, technically it is correct to leave it in.
- DC HP: surely can't hurt at the LP ouput.
- series R: yes, in series, recommended to prevent op amp oscillation, at the very end of the chain (last op amp). Anything from 47 to 200 Ohms will do the job. Typical is 47 or 50.
- PSU: on the safe side a 12-0-12 will yield 12 V regulated. 15V may be possible but likely unnecessary. If your amps have a gain of just 10x (20dB), and say you have 40 V rails (a ca. 100 W into 8 Ohms power amp), the power amp will clip at 1.4 V signal voltage. 12 V, 15 V, hey, even 9 V may work for the PSU. Well, technically you may reach higher values interstage but likely not 12 V if final max necessary output is 1.4 V.
- construction: whatever op amp you choose, read the data sheet - it will tell you the necessary decoupling cap values, for once, and more. Proper decoupling and short PSU lines after regulation may be THE key to good performance, and IMO easily beats choice of capacitor brand in the influence on good sound. Also, good layout.
Good luck!
MBK
As for power supply. For the regulators I am looking at LM340T-12 & LM320T-12.
Looking at the spec sheets for these the + reg (LM340) requires a .1 uF ceramic disc on the output and the - reg (LM320) requires a 1uF tantalum or 25uf electrolytic. Do I build the +ve/-ve supply different from each other going entirely by the spec sheet?
Also what size filter caps should I use?
Looking at the spec sheets for these the + reg (LM340) requires a .1 uF ceramic disc on the output and the - reg (LM320) requires a 1uF tantalum or 25uf electrolytic. Do I build the +ve/-ve supply different from each other going entirely by the spec sheet?
Also what size filter caps should I use?
MBK said:
Measuring the impedance as per SL's website...
Judging by the way my ohmmeter fluctuates when trying to measure the DC resistance of the subs, I am thinking my measurements might not be any more accurate than the simulation. It seems to be somewhere between 7.2 ohm and 7.5 ohm with the resistance of the speaker wire... that's about as accurate as I can get. But when I connect the ohmmeter to a 7.2 ohm resistor it is stable.
Is this normal? Should I go on with the measurements like this being as accurate as I can?
My measurements and calculations give me F0 @38 hz and Q0 @ .94.
This cant be right. My box is already double the size of what the manufacturer recommends. Are they recommending a final Q of 1.1 or higher? Software predicted F0 @ 40hz and Q0 @ .62. The Q0 is WAY OFF.
Could it be my subs need breaking in? It's kinda hard for someone to verify my calculations.
I got
Rdc=7.2ohm
F0 @38hz impedance 61.5ohm
F1 @30.6hz impedance 21ohm
F2 @44.4hz impedance 21ohm
This cant be right. My box is already double the size of what the manufacturer recommends. Are they recommending a final Q of 1.1 or higher? Software predicted F0 @ 40hz and Q0 @ .62. The Q0 is WAY OFF.
Could it be my subs need breaking in? It's kinda hard for someone to verify my calculations.
I got
Rdc=7.2ohm
F0 @38hz impedance 61.5ohm
F1 @30.6hz impedance 21ohm
F2 @44.4hz impedance 21ohm
I hope some can follow what I'm saying...
The way this calculation (for the Q0 and F0 of the sub) is done it seems to me that it uses too many assumptions and approximations.
linkwitz' link
If impedance at f1 or f2=sqrt(Rmax*Rdc) and F0=sqrt(f1*f2) then Fo is not at the impedance peak(ie Rmax)? By the graph Linkwitz puts F0 right at Rmax.
Using F0=sqrt(f1*f2) I now get a F0=36.7hz and a Q=.91.
What I'm getting at is now @ f0=36.7hz the impedance is only about 53ohm as opposed to 61.5ohm which is my measured Rmax. So do I use 53ohm as Rmax now and find new F1 and F2 with impedance of sqrt(Rmax*Rdc)? I could keep going on forever because now no doubt my f0 will move again because I have a new F1 and F2.
This post deserves 3 of these things
The way this calculation (for the Q0 and F0 of the sub) is done it seems to me that it uses too many assumptions and approximations.
linkwitz' link
If impedance at f1 or f2=sqrt(Rmax*Rdc) and F0=sqrt(f1*f2) then Fo is not at the impedance peak(ie Rmax)? By the graph Linkwitz puts F0 right at Rmax.
Using F0=sqrt(f1*f2) I now get a F0=36.7hz and a Q=.91.
What I'm getting at is now @ f0=36.7hz the impedance is only about 53ohm as opposed to 61.5ohm which is my measured Rmax. So do I use 53ohm as Rmax now and find new F1 and F2 with impedance of sqrt(Rmax*Rdc)? I could keep going on forever because now no doubt my f0 will move again because I have a new F1 and F2.
This post deserves 3 of these things
The idea behind this whole exercise is to get to know where resonance occurs (f0) and how broad the peak is (Q0). The f0 can be measured at Rmax. The measure of the peak breadth, the Q, gets calculated from f1 and f2. Those in turn are defined by a standard measure of impedance falloff. The formulas assume one resonance only, symmetry , etc.
I think you should use as many real, measured data as you can. The formulas use idealizations. Better use real data than ones recalculated from formulas. So, get f1 and f2 from measurement (RDC, Rmax). Then calculate Q0 from measured RDC, Rmax, f1 and f2, but use the measured, not the calculated f0.
Of course, getting large deviations may mean that your data may have a problem - maybe the driving amp, or a box resonance peak, etc, that make the formula disgress from reality. (just my opinion!)
Now I must admit I have not carried out a Linkwitz transform myself yet - was planning to these weeks with my woofers and tweeter - so I have no experience of the practical problems one is getting... I made my remark because I sort of suspected things wouldn't behave like in theory...
MBK
I think you should use as many real, measured data as you can. The formulas use idealizations. Better use real data than ones recalculated from formulas. So, get f1 and f2 from measurement (RDC, Rmax). Then calculate Q0 from measured RDC, Rmax, f1 and f2, but use the measured, not the calculated f0.
Of course, getting large deviations may mean that your data may have a problem - maybe the driving amp, or a box resonance peak, etc, that make the formula disgress from reality. (just my opinion!)
Now I must admit I have not carried out a Linkwitz transform myself yet - was planning to these weeks with my woofers and tweeter - so I have no experience of the practical problems one is getting... I made my remark because I sort of suspected things wouldn't behave like in theory...
MBK
Actually, expanding on the issue:
assuming no measurement issues, you face a trade-off. If you take the calculated f0 and the measured f1 and f2, your filter will correct for the real Q but centered slightly wrongly (bump at real f0 and dip at calculated f0, but good correction at f1 and f2). If you take the measured f0 (from f at Rmax) the correction will be somehow off at f1 and f2. It may be better to also use the measured f0 in this case, rather than the calculated one.
Honestly, I wouldn't worry about some deviations from the ideal - you will already correct out the most severe aberrations with this method and the filter you're building from it, and a few dB peaks and dips in those frequencies will hardly matter.
Box construction btw may be the reason for the deviation in Q0 - stuffing, sealing etc. But, also , note that Q0 apparently is defined differently from Thiele-Small Qts. The manufacturer likely talks about T/S Qts. Sadly, my understanding of the matter stops here
assuming no measurement issues, you face a trade-off. If you take the calculated f0 and the measured f1 and f2, your filter will correct for the real Q but centered slightly wrongly (bump at real f0 and dip at calculated f0, but good correction at f1 and f2). If you take the measured f0 (from f at Rmax) the correction will be somehow off at f1 and f2. It may be better to also use the measured f0 in this case, rather than the calculated one.
Honestly, I wouldn't worry about some deviations from the ideal - you will already correct out the most severe aberrations with this method and the filter you're building from it, and a few dB peaks and dips in those frequencies will hardly matter.
Box construction btw may be the reason for the deviation in Q0 - stuffing, sealing etc. But, also , note that Q0 apparently is defined differently from Thiele-Small Qts. The manufacturer likely talks about T/S Qts. Sadly, my understanding of the matter stops here
Too many op-amps?
Your design is probably being driven from a pre-amp or other low impedance source. The input buffers on the HP seem redundant. In my book fewer op-amps is a good thing. 2nd The summing resistors at 100K seem higher than needed lower values typically sound better.
Most pre-amps have output impedances in the 100 Ohm range many solid state pre-amps could be as low as 5 Ohms often lower than most op-amps.
Your design is probably being driven from a pre-amp or other low impedance source. The input buffers on the HP seem redundant. In my book fewer op-amps is a good thing. 2nd The summing resistors at 100K seem higher than needed lower values typically sound better.
Most pre-amps have output impedances in the 100 Ohm range many solid state pre-amps could be as low as 5 Ohms often lower than most op-amps.
Re: Too many op-amps?
The summing resistors were reduced if you look at the updates of the schematic I have made as the thread progressed.
The problem is my signal could be coming from my soundcard, PS2, VCR ,etc... So I would be assuming they all have a low output impedance. Is this safe to assume?
Without the first buffers my circuit would look like this:
Will this work?
edit: i notice R8 will have to be changed to from 3k3 to 10k now, not that it makes much difference
hermanv said:
The summing resistors were reduced if you look at the updates of the schematic I have made as the thread progressed.
The problem is my signal could be coming from my soundcard, PS2, VCR ,etc... So I would be assuming they all have a low output impedance. Is this safe to assume?
Without the first buffers my circuit would look like this:
Will this work?
edit: i notice R8 will have to be changed to from 3k3 to 10k now, not that it makes much difference
I'm gonna need a preamp for a volume control if I want to hook up anything but the computer directly. My apologies,I was getting ahead of myself.
So I can build the circuit without the input buffers to hook up to my computer and if that doesn't work I will use a preamp with a volume control as a buffer because I'm gonna have to build this simple preamp for other inputs anyway.
I really wish I had a receiver with preouts...
So I can build the circuit without the input buffers to hook up to my computer and if that doesn't work I will use a preamp with a volume control as a buffer because I'm gonna have to build this simple preamp for other inputs anyway.
I really wish I had a receiver with preouts...
Banned
Joined 2002
My intention was for this to be a temporary fix until I get a good reciever with a better built-in crossover. (Then I'll have the problem of a 4th order acoustical HP with only a 2nd order LP but I still wont be able to use this circuit because I'll need the front channels to be powered by the receiver and the DAC too.)
I am going to put a preamp circuit in with the crossover.
If you mean make a DIY HT amp, I would like to but it's just cheaper to buy and guaranteed to work as a bonus!
(I don't think I am up to the task.)
I haven't even looked at how to build a DAC as the whole idea sounds scary to me.
jleaman said:
I am going to put a preamp circuit in with the crossover.
If you mean make a DIY HT amp, I would like to but it's just cheaper to buy and guaranteed to work as a bonus!
(I don't think I am up to the task.)I haven't even looked at how to build a DAC as the whole idea sounds scary to me.
I couldn't wait so I built a Linkwitz transform on a breadboard. And it actually worked right away! I get a small 60hz buzz but not much.
It seems like the subs are gonna reach Xmax before the mains... I may not even need a HP for the mains.
I am glad I didn't get a 250W plate amp like I had considered because I got the clip indicator to flash on my 370W amp and the subs could still take more. I may have to change the wiring for 2 ohm rather than 8 ohm.
Its just damned hard to get under 25hz with any amount of SPL.
edit: I gonna have to try better positioning for the subs....Walking around the room reveals how much I can achieve. Every little knicknack throughout the house vibrates even way at the other end of the house.
It seems like the subs are gonna reach Xmax before the mains... I may not even need a HP for the mains.
I am glad I didn't get a 250W plate amp like I had considered because I got the clip indicator to flash on my 370W amp and the subs could still take more. I may have to change the wiring for 2 ohm rather than 8 ohm.
Its just damned hard to get under 25hz with any amount of SPL.
edit: I gonna have to try better positioning for the subs....Walking around the room reveals how much I can achieve. Every little knicknack throughout the house vibrates even way at the other end of the house.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.