Howdy,
In a 1st-order high-pass crossover, is it best to place an L-pad before or after the capacitor?
The tweeter is an Audax TW0A250, if anyone has experience with what works best with it.
Thanks,
David
In a 1st-order high-pass crossover, is it best to place an L-pad before or after the capacitor?
The tweeter is an Audax TW0A250, if anyone has experience with what works best with it.
Thanks,
David
The L pad should be fitted after the capacitor and next in line to the tweeter.Remember the xover point is dependant on the impedance it sees from the tweeter and resistor combination.In most implementations a series first order network and especially the quasi second order version, is preferable to the parallel connected alternative.
thanks VaNarn.
I saw that AllenB, in his excellent crossover-design-without-measurement guide, connects the series resistor before the capacitor, and the parallel resistor after the capacitor. He doesn't state why, but must have a reason for doing so. I'll make the decision easy and do as AllenB does. That's good enough for me. (o:
Regards,
David
I saw that AllenB, in his excellent crossover-design-without-measurement guide, connects the series resistor before the capacitor, and the parallel resistor after the capacitor. He doesn't state why, but must have a reason for doing so. I'll make the decision easy and do as AllenB does. That's good enough for me. (o:
Regards,
David
Placing the full amplifier output across an attenuator, i.e. before the crossover filter,
will severely heat the attenuator components.
If you did this you MUST rate the attenuator components for the worst case power dissipations expected.
In general, we always place tweeter and even mid driver attenuator components after the filter.
Splitting the series attenuator component from the shunt attenuator component and placing the filter between the two, does not expose the attenuator components to the full amplifier output.
Work it out and see for your self.
His arrangement seems to simply be a way to confuse those that read it, or maybe he did that to make his "look" different and thus impress readers with this new topology.
It certainly impressed you to the extent you misunderstood the critical difference that moving the shunt element would have made.
will severely heat the attenuator components.
If you did this you MUST rate the attenuator components for the worst case power dissipations expected.
In general, we always place tweeter and even mid driver attenuator components after the filter.
Splitting the series attenuator component from the shunt attenuator component and placing the filter between the two, does not expose the attenuator components to the full amplifier output.
Work it out and see for your self.
I see no technical reason for swapping the location of the capacitor and resistor.
His arrangement seems to simply be a way to confuse those that read it, or maybe he did that to make his "look" different and thus impress readers with this new topology.
It certainly impressed you to the extent you misunderstood the critical difference that moving the shunt element would have made.
Last edited:
Andrew,
Thanks for the reply.
I had been using a 3-ohm series resistor placed before the capacitor for tweeter attenuation. The crossover board is mounted externally, and during use the 10-watt resistor has remained cool to the touch.
I didn't want the added impedance of the single series resistor interfering with the crossover frequency (I read and was told it does) so that's why the addition of the parallel resistor. But you know that.
Maybe AllenB will chime in and state why he prefers, if there is a preference, to place the series resistor before the capacitor and the parallel resistor after the capacitor.
Regards,
David
Thanks for the reply.
I had been using a 3-ohm series resistor placed before the capacitor for tweeter attenuation. The crossover board is mounted externally, and during use the 10-watt resistor has remained cool to the touch.
I didn't want the added impedance of the single series resistor interfering with the crossover frequency (I read and was told it does) so that's why the addition of the parallel resistor. But you know that.
Maybe AllenB will chime in and state why he prefers, if there is a preference, to place the series resistor before the capacitor and the parallel resistor after the capacitor.
Regards,
David
Andrew is right, it makes no difference. These components are in series and passing the same current at the same time.
The impedance the capacitor sees with regard to working out the cutoff frequency will be the driver impedance in parallel with the shunt resistance, plus the series resistance, the amplifier source impedance (usually zero ohms), and any impedance in the negative connection.
thanks Allen.
Ok, so it doesn't matter what order the capacitor and resistor are in. Originally I was told that the series resistor should go before the capacitor in the 1st-order high pass.
But from Rod Elliott, "Passive Crossover Network Design":"The attenuator must be placed either before the filter (a very bad idea indeed!), or between the crossover filter and the driver--the latter includes impedance compensation. The driver and its associated impedance-correction network should be considered as one, and they should not be separated (unless you feel like re-calculating the entire crossover and compensation networks). I have seen a number of design examples that state that the attenuator should be before the crossover--wrong, wrong, wrong! This practice increases power dissipation needlessly, since the attenuator must work over the entire frequency range. If attenuation is after the crossover, then power requirements are greatly reduced."
So Elliott has a strong opinion that the series resistor should be placed after the capacitor.
And I've read other opinions that state there is no hard-and-fast rule, and that differing positions for the series resistor offer advantages/disadvantages like phase matching, various tilts to the tweeter's output, an opinion that "the net effect with a resistor placed before the circuit will attenuate the low and mid treble far more than the upper end, and a resistor placed after the circuit will attenuate the high and mid treble far more than the low treble", and another opinion that a resistor before will not usually apply a tilt to the response, whereas one after in series will tilt it down with increasing frequency.
I guess I'll try differing positions; I've used velcro to attach the crossover components to the poplar xover board, so changing is simple. (o:
Regards,
David
Ok, so it doesn't matter what order the capacitor and resistor are in. Originally I was told that the series resistor should go before the capacitor in the 1st-order high pass.
But from Rod Elliott, "Passive Crossover Network Design":"The attenuator must be placed either before the filter (a very bad idea indeed!), or between the crossover filter and the driver--the latter includes impedance compensation. The driver and its associated impedance-correction network should be considered as one, and they should not be separated (unless you feel like re-calculating the entire crossover and compensation networks). I have seen a number of design examples that state that the attenuator should be before the crossover--wrong, wrong, wrong! This practice increases power dissipation needlessly, since the attenuator must work over the entire frequency range. If attenuation is after the crossover, then power requirements are greatly reduced."
So Elliott has a strong opinion that the series resistor should be placed after the capacitor.
And I've read other opinions that state there is no hard-and-fast rule, and that differing positions for the series resistor offer advantages/disadvantages like phase matching, various tilts to the tweeter's output, an opinion that "the net effect with a resistor placed before the circuit will attenuate the low and mid treble far more than the upper end, and a resistor placed after the circuit will attenuate the high and mid treble far more than the low treble", and another opinion that a resistor before will not usually apply a tilt to the response, whereas one after in series will tilt it down with increasing frequency.
I guess I'll try differing positions; I've used velcro to attach the crossover components to the poplar xover board, so changing is simple. (o:
Regards,
David
Not exactly. The shunt resistor is providing a path for extra current, and is in series with the capacitor if after it, or not if before. The series resistor is after the capacitor in Rod's example for more predictable results from the second order filter.
Two series components with nothing shunting between them can be seen as occupying the same circuit location. As an example, consider an inductor with resistance through the wire. This resistance is in series with the inductance and is the same as using a low resistance inductor with a resistor added, either before or after the inductor, again with only one current path through them.
In other words the two components can in theory be reduced to a single impedance without affecting the circuit.
thanks Allen.
Have to remember the theoretical grain of salt. (o:
Doesn't a capacitor act on the crossover frequency relative to the impedance, and wouldn't a resistor after the capacitor's action affect the impedance and the crossover point to follow? I guess I'm thinking in sequential terms and shouldn't be.
What brought up my original question were statements found on techtalk.parts-express.com; Donradick's : "Not quite a 'can of worms', but definitely a big topic. I'd suggest a lot of modeling in PCD. It seems to vary a lot with different tweeters. Not one is 'best' overall. Sometimes guys will use 2 resistors, one in front and one 'in back' to help with phase matching, or to allow easy tweeks with the tweeter level for a specific room. The Lpad configuration helps if a tweeter has a strong inductive rise from 5K to 20K."
And by Chris Roemer: "This is ONE thing that Vance really covers well in his LDC. By moving the padding resistance around, you can get various 'tilts' to the tweeter's output, and possibly even carve out the mid part of its pass band. These also apply to the attenuation of a mid in a 3-way."
And one more comment but there are others, this one by johnnyrichards: "To reiterate what has already been said, you have to experiment. In my experience, the benefits of putting the series resistor before the crossover provides more benefits. However, I have done the two-series resistor "split" L-pad before, and only once have I put the entire L-pad after the crossover. It is interesting to see how the same component affects things drastically differently. In any event, I have yet to hear a tweeter perform worse with a parallel resistor than without. I keep meaning to experiment with putting the parallel resistor in front of the crossover to see what happens."
Well, one more good one from Wolf: "A resistor before will not usually apply a tilt to the response, whereas one after in series will tilt it down with increasing freq. The resistor across the driver will also not impart much of a tilt, and be a uniform attenuation. If you want a lower phase angle on the impedance, then placing a resistor before the xover will minimize it if when placed after makes it swing capacitive wildly."
Have to remember the theoretical grain of salt. (o:
Doesn't a capacitor act on the crossover frequency relative to the impedance, and wouldn't a resistor after the capacitor's action affect the impedance and the crossover point to follow? I guess I'm thinking in sequential terms and shouldn't be.
What brought up my original question were statements found on techtalk.parts-express.com; Donradick's : "Not quite a 'can of worms', but definitely a big topic. I'd suggest a lot of modeling in PCD. It seems to vary a lot with different tweeters. Not one is 'best' overall. Sometimes guys will use 2 resistors, one in front and one 'in back' to help with phase matching, or to allow easy tweeks with the tweeter level for a specific room. The Lpad configuration helps if a tweeter has a strong inductive rise from 5K to 20K."
And by Chris Roemer: "This is ONE thing that Vance really covers well in his LDC. By moving the padding resistance around, you can get various 'tilts' to the tweeter's output, and possibly even carve out the mid part of its pass band. These also apply to the attenuation of a mid in a 3-way."
And one more comment but there are others, this one by johnnyrichards: "To reiterate what has already been said, you have to experiment. In my experience, the benefits of putting the series resistor before the crossover provides more benefits. However, I have done the two-series resistor "split" L-pad before, and only once have I put the entire L-pad after the crossover. It is interesting to see how the same component affects things drastically differently. In any event, I have yet to hear a tweeter perform worse with a parallel resistor than without. I keep meaning to experiment with putting the parallel resistor in front of the crossover to see what happens."
Well, one more good one from Wolf: "A resistor before will not usually apply a tilt to the response, whereas one after in series will tilt it down with increasing freq. The resistor across the driver will also not impart much of a tilt, and be a uniform attenuation. If you want a lower phase angle on the impedance, then placing a resistor before the xover will minimize it if when placed after makes it swing capacitive wildly."
Ok. A circuit has no beginning nor end. It has only sources of voltage and current, it has impedances and current paths.
A crossover has one apparent source of voltage (the amp) and one apparent load (the speaker). Now, not to confuse (getting to a point) but actually in a working crossover the capacitors and inductors and the speaker will also become sources of voltage and will source and sink current.. any resistors, stray resistances and the speaker will be loads as well.
Simplify for a moment and draw a series of four attached resistors and join the two ends. Amp (0.1 ohms), resistor (10 ohms), capacitor (10 ohms), speaker (10 ohms). Consider the amp resistance insignificant but have a voltage appearing across it (3V).
The voltage is also going to be appearing across the series combination of the other three. Each will see 1V of the total 3V (which draws 100mA). Try rearranging all of these resistances in the circuit and you won't find a combination where this isn't true.
Add a shunt resistor across the driver also 10 ohms. This part will now look like 5 ohms total and can still be drawn as a single resistor representing the driver and shunt resistor combination.
The voltage will drop to 0.6V, which is 5ohms/25ohms x 3V. Finally the point, this will apply even if the series resistor is in the negative leg of the circuit, or half and half around the amp or capacitor.
Although this is true, this is an attitude toward making the best out of a collection of circuit elements (l-pads and so forth) and in order to produce various combinations, the shunt elements need to be intermingled with the series to create changes in the circuit.
The changes that will happen involve the impedances (and hence the voltages) changing at either some or all frequencies.
On the other hand when I make a crossover from scratch, I begin with a response, and a target response. I have an impedance to work with and can use any combination of filtration that turns one into the other. There are going to be many acceptable ways to do it.
[Not important but some ways will be producing the same result, some will produce a similar result with a downside, like a difficult impedance for the amp to drive, and some ways may help or exacerbate minor issues like the parasitic concerns with capacitors and what not.]
As you noted the L-pad is good for balancing the raw driver impedance. None of this is strictly necessary if the filters for example, fit nicely around the impedance peaks and driver response. On the other hand, some peaks might be tricky, steep or whatever and it's easier to fix them separately. This also helps when you want to make a later change.
An L-pad is not capable of making a complete fix, for this you need reactive components so an RLC can be used for a tweeter resonance, after which the shunt resistor has nothing left to fix. The tweeter will look more or less like a resistor and a shunt resistor merely changes its value.
I'd want to see what he wrote before and after this one. It seems reasonable but dependent on a few things. For example, if there was a slope, and it changed in frequency, would you call that a change in slope? Although it appears it has just moved, it could also be the same as a reverse slope plus a new slope.
Thanks again Allen. You should write a book. (o:
Wolf's complete comment was: "A resistor before will not usually apply a tilt to the response, whereas one after in series will tilt it down with increasing freq. The resistor across the driver will also not impart much of a tilt, and be a uniform attenuation. If you want a lower phase angle on the impedance, then placing a resistor before the xover will minimize it if when placed after makes it swing capacitive wildly."
Another interesting techtalk thread ("placement of L-pad") has Chris Roemer's graph of differing response curves based on differing L-pad placements in the circuit.
See here: Placement of L-pad
His graph's yellow and green curves give the flattest response; yellow is with the L-pad after the crossover, green in front of the crossover, additionally flattening the Z-profile.
It's all very interesting. I need to get a crossover-modeling program. Which do you recommend that's not too expensive, or is free?
Regards,
David
Wolf's complete comment was: "A resistor before will not usually apply a tilt to the response, whereas one after in series will tilt it down with increasing freq. The resistor across the driver will also not impart much of a tilt, and be a uniform attenuation. If you want a lower phase angle on the impedance, then placing a resistor before the xover will minimize it if when placed after makes it swing capacitive wildly."
Another interesting techtalk thread ("placement of L-pad") has Chris Roemer's graph of differing response curves based on differing L-pad placements in the circuit.
See here: Placement of L-pad
His graph's yellow and green curves give the flattest response; yellow is with the L-pad after the crossover, green in front of the crossover, additionally flattening the Z-profile.
It's all very interesting. I need to get a crossover-modeling program. Which do you recommend that's not too expensive, or is free?
Regards,
David
Using the A, B and C example in the first post of the Parts Express thread, I'd point out three things. Firstly the response could likely be made the same in all three cases. Second, it would then primarily sound the same. [I'd add though that if there is a different draw on the amp, then the amp may sound different, not so much if because the amp has a different output impedance as this could also be adjusted for being just another series impedance, but perhaps due to other loading aspects that you'll only notice with some amps.]
And thirdly, that each has practical benefits that will likely make the decision. These come down to such things as convention stemming from the intended purpose of each circuit element.
The practical issues as I see them: The RC impedance compensation and any similar compensation functions as intended in its simplest form when it is placed directly across the driver. Although it can be elsewhere, since its function doesn't change with other crossover adjustments it makes sense not to put any adjustable elements between it and the driver.
An L-pad minimises remaining impedance variations and can be useful closer to the driver as well. Placing this at the beginning like with the third example also can be wasteful of amp power. On the other hand, it will create an easier load for the amp with regards to the impedance phase (for what little effect this might have if any), but there are other ways to achieve this.
And thirdly, that each has practical benefits that will likely make the decision. These come down to such things as convention stemming from the intended purpose of each circuit element.
The practical issues as I see them: The RC impedance compensation and any similar compensation functions as intended in its simplest form when it is placed directly across the driver. Although it can be elsewhere, since its function doesn't change with other crossover adjustments it makes sense not to put any adjustable elements between it and the driver.
An L-pad minimises remaining impedance variations and can be useful closer to the driver as well. Placing this at the beginning like with the third example also can be wasteful of amp power. On the other hand, it will create an easier load for the amp with regards to the impedance phase (for what little effect this might have if any), but there are other ways to achieve this.
Last edited:
There's a bit of a gap between the professional offerings and the freeware. The freeware is good, your knowledge and experience can become the limiting factor. Do you have any computer programming experience? I'm only talking about dealing with data files and scripting. I notice someone from the PE thread uses xoversim. This is a simple one, nothing wrong with that. The crossover is entered using text columns and circuit nodes, ie: mark the circuit connections with numbers and say, there is a capacitor between 6 and 7....
Speaker workshop is much more comprehensive. A little clunky, but ways to manipulate data files, you know, combine them, adjust plots and bring in measurements as you go.
Anything else I need I'd write myself so I'd be the wrong person to ask.
LEAP would be nice but I'm not a professional, just a novice tinkerer so spending that much isn't justified.
I don't do much with computers, but can download and install if my old PC is up to it; it runs XP Pro, 2002. (o:
I've come across Boxsim, PXover, PSD_Lite, Jeff Bagby's and David Ralph's programs; didn't know which one to try considering my PC's limitations, and not having experience with them.
I started my 1st-order crossover fiddling by using values from V-Cap's crossover calculator. Then I read/applied your detailed guide and found that the value (tweeter cap) thus determined didn't match V-Cap's calculator's, so I am kinda stuck.
However, my monitors with Audax HT170K0 mid-bass and Audax TW025A0 tweets, using the V-Cap-calculator values of 8.2uF and ~.50mH (selected .55mH to perhaps help attenuate the mid-bass' rising upper-midrange response) for the xovers (I added a Zobel for the mid-bass and L-pad for the tweets) sound really nice, with surprising dynamics and presence, with good imaging and smoothness. The blending between drivers is good, with no hugely discernible peaks/nulls. Still, I'd like to know what they're doing objectively.
Regards,
David
I don't do much with computers, but can download and install if my old PC is up to it; it runs XP Pro, 2002. (o:
I've come across Boxsim, PXover, PSD_Lite, Jeff Bagby's and David Ralph's programs; didn't know which one to try considering my PC's limitations, and not having experience with them.
I started my 1st-order crossover fiddling by using values from V-Cap's crossover calculator. Then I read/applied your detailed guide and found that the value (tweeter cap) thus determined didn't match V-Cap's calculator's, so I am kinda stuck.
However, my monitors with Audax HT170K0 mid-bass and Audax TW025A0 tweets, using the V-Cap-calculator values of 8.2uF and ~.50mH (selected .55mH to perhaps help attenuate the mid-bass' rising upper-midrange response) for the xovers (I added a Zobel for the mid-bass and L-pad for the tweets) sound really nice, with surprising dynamics and presence, with good imaging and smoothness. The blending between drivers is good, with no hugely discernible peaks/nulls. Still, I'd like to know what they're doing objectively.
Regards,
David
For me, other peoples computers can be a handy source of bread and butter. I don't go for telling people they need more than they want. I'm not even all that keen on the latest offerings myself from the ms farm.
One thing I can say is it's not the power of your crossover simulator, you might have the best and still not be happy with the results. You need to apply it well. Experience can get good results with a simple simulator, it may just take a little longer to process the data.
The tutorial thread is good for what it claims, if you feel up to it you should begin to measure. The typical intermediate level is to have a response/phase, impedance/phase measurement for each driver, and if necessary to apply baffle effects then find a flat response.
Baffle step compensation is a complex and often misunderstood thing. At this level it is best seen as an 'averaging' type of fix and be combined with good room placement. Singular measurements like this may produce a good crossover but tweaking will often still be necessary IMO.
A truly good crossover begins in the acoustic domain, and considers what radiation is desirable for a given room and listening situation. Not to put too fine a point on it, but beginning with the electrical filters can be like getting someone else to design you a house and making it yours by choosing the paint and curtains.
Last edited:
To expand on that: I could not find a way to integrate a dome tweeter with a woofer, there was always a harshness and discontinuity. Back then I crossed near 5kHz using a second order filter because of what I had read in books and spec sheets.
Compensating the impedance peak helped but not as I'd hoped so I would set up one of these filters whenever I bought new tweeters and leave it.
Changing the tweeter level helped, and this was a clue, there wasn't one that worked properly. Same if I tilted the response, it came close. I chased down the individual responses to within a fraction of a dB, and phase to within a few degrees.
The change that helped was lowering the crossover frequency. Distortion rose, but that wasn't really a problem. A first order filter helped as well but only for the same reason.. I think now, was finding a frequency that gave better matching of the dispersion of each driver, and easing the crossover away from the most sensitive region of hearing.
YMMV
Compensating the impedance peak helped but not as I'd hoped so I would set up one of these filters whenever I bought new tweeters and leave it.
Changing the tweeter level helped, and this was a clue, there wasn't one that worked properly. Same if I tilted the response, it came close. I chased down the individual responses to within a fraction of a dB, and phase to within a few degrees.
The change that helped was lowering the crossover frequency. Distortion rose, but that wasn't really a problem. A first order filter helped as well but only for the same reason.. I think now, was finding a frequency that gave better matching of the dispersion of each driver, and easing the crossover away from the most sensitive region of hearing.
YMMV
- Status
- Not open for further replies.