I have been doing a lot of experimenting lately and wondering if a zobel is ideal on all drivers or if a line array of all the same driver perhaps on the binding posts so the amps see the same impedance is beneficial with active passive and no crossing? Ok the tone may be altered but on a flat response some full range drivers can have 2 large impedance spikes I believe one was as high as 40 if I'm not mistaken. I can't see a reason to not have them in all situations unless the active crossover has some kind of compensation? Can some one please add color? Also if I'm doing an array without a crossover of the same drivers can I put a larger zobel on the binding posts? Quality caps are expensive as well as resistors and I rather not spend a few hundred to roll the dice. Thanks in advance.
The benefit of compensating the impedance will be that the amp sees an easier load. The effect depends on the amp. Where the output impedance is high, the response will no longer be affected by the nature of the varying speaker impedance. In any case the loadline will be straightened. The effect will probably not be significant or noticeable on some or most amps.
To achieve this it should only be necessary to use one network.
To achieve this it should only be necessary to use one network.
The Zobel has a couple of beneficial effects in multi-way speaker design. It makes your low pass filters more reliable, and possibly more simple, and it can be tweaked to change the phase matching with the high pass filter.
If you want to experiment, I suggest downloading XSim. There are a couple of threads here which include complete speaker designs, one for the A26 and another for the LM-1. The LM-1 includes a Zobel, so take a lookat how removing it affects phase and amplitude for the low pass filter.
It is also true that having a high impedance, as well as one with a narrow amount of change will be easier on amplifiers. The larger the output impedance of an amp, the more the frequency response changes to match the impedance of the speaker. That is, impedance peaks become output peaks.
Best,
Erik
If you want to experiment, I suggest downloading XSim. There are a couple of threads here which include complete speaker designs, one for the A26 and another for the LM-1. The LM-1 includes a Zobel, so take a lookat how removing it affects phase and amplitude for the low pass filter.
It is also true that having a high impedance, as well as one with a narrow amount of change will be easier on amplifiers. The larger the output impedance of an amp, the more the frequency response changes to match the impedance of the speaker. That is, impedance peaks become output peaks.
Best,
Erik
So it always makes sense to implement in all cases active or passive - I don't want to have to account for different amps. If I am running speakers in parallel or series can I use 1 zobel? I'm sure voltage ratings need to be increased but what about the C and R? Hypothetically 16 drivers?
Hi Jared,
Depends. With solid state amplifiers, and active crossovers it's not very important.
Again, grab XSim and model this yourself.
Best,
Erik
Depends. With solid state amplifiers, and active crossovers it's not very important.
Again, grab XSim and model this yourself.
Best,
Erik
Sorry, I think I wrote something that could be read differently than I meant.
When I said that Zobel networks make them more reliable I did NOT mean to say it will make your speakers or electronics last longer.
What I meant to say was that a low pass filter can now operate without having to adjust for the driver's rising impedance.
Again, I encourage you to look at the LM-1 simulatin, and try removing the Zobel in the woofer section. You'll see the subtle change in the woofer output as a result. Also note the system impedance.
As the woofer's impedance goes up, the crossover's effectiveness goes down.
When I said that Zobel networks make them more reliable I did NOT mean to say it will make your speakers or electronics last longer.
What I meant to say was that a low pass filter can now operate without having to adjust for the driver's rising impedance.
Again, I encourage you to look at the LM-1 simulatin, and try removing the Zobel in the woofer section. You'll see the subtle change in the woofer output as a result. Also note the system impedance.
As the woofer's impedance goes up, the crossover's effectiveness goes down.
I'm still missing the reason of why you want to implement a zobel network. A zobel flattens the impedance so IMHO it was a "must" in the past with passive crossovers in order to calculate a crossover without worrying about the change of impedance vs frequency. With modern simulation tools it is no longer necessary, I understand there are cases when it is useful but I don't remember.
Ralf
I don't see the point of your concept. When you use two drivers you get a crossover. Most of the time you need to alter the signal to make an improvement, so eliminating necessary components only removes the components. The deficiencies of components are relatively superficial.
Much of this can't be done electronically.
IMHO you're confusing two different things. Can you post the impedance graph that you would like to flatten?
Ralf
Edit: directed to the OP, not AllenB
Ralf
Edit: directed to the OP, not AllenB
giralfino, I share your confusion. Maybe the OP thinks that fixing the impedance when there is no crossover will make a difference on a low output impedance amp.
The thing is, it might make a difference on some amps. I've attached a sample reactive loadline to show where I'm coming from.
The thing is, it might make a difference on some amps. I've attached a sample reactive loadline to show where I'm coming from.
Attachments
I have build a set of full range line arrays (no crossover) and experimented with such a Zobel network. I started out with a network that made the impedance plot completely flat from bottom to top. After listening sessions I have removed the top end correction.
The circuit that removes the impedance bump at Fs is still in there today. I've done measurements with and without such a conjugation network. As I'm using FIR correction the differences in the measurements were very small. Purely based on subjective listening I removed the upper end correction.
There was something in the impedance bump correction that made me keep it. I wasn't able to come up with any proof though. The one thing that struck me as odd was looking at the group delay difference between the corrected and uncorrected speaker's impedance measurement.
This is the group delay tab on an impedance measurement without the Zobel:
This is the group delay with the Zobel connected:
The impedance measurement was done in REW, with a known resistor and an old laptop using it's soundcard.
A comparison between corrected and uncorrected impedance:
The curve I ended up keeping was this:
Lot's more in my thread: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-132.html#post4423888
The circuit that removes the impedance bump at Fs is still in there today. I've done measurements with and without such a conjugation network. As I'm using FIR correction the differences in the measurements were very small. Purely based on subjective listening I removed the upper end correction.
There was something in the impedance bump correction that made me keep it. I wasn't able to come up with any proof though. The one thing that struck me as odd was looking at the group delay difference between the corrected and uncorrected speaker's impedance measurement.
This is the group delay tab on an impedance measurement without the Zobel:
This is the group delay with the Zobel connected:
The impedance measurement was done in REW, with a known resistor and an old laptop using it's soundcard.
A comparison between corrected and uncorrected impedance:
The curve I ended up keeping was this:
Lot's more in my thread: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-132.html#post4423888
I think before I start asking stupid questions, I am going to look at Xsim. Thank you for that.
I have a similar situation- I am using 6 drivers in series/parallel for the lower band, and I'm trying to decide if I can use just one Zobel. I don't know if the readily available formulas will work with the increased impedance complexity of such a network.
If anyone out there has the "math", I would appreciate it if you posted it here.
As well, thank you Wesayso- I am using your "stacked" construction as a model. VERY nicely done. You are fortunate that you have a supply of premium plywood. The stuff we get here blows.
I have a similar situation- I am using 6 drivers in series/parallel for the lower band, and I'm trying to decide if I can use just one Zobel. I don't know if the readily available formulas will work with the increased impedance complexity of such a network.
If anyone out there has the "math", I would appreciate it if you posted it here.
As well, thank you Wesayso- I am using your "stacked" construction as a model. VERY nicely done. You are fortunate that you have a supply of premium plywood. The stuff we get here blows.
If the only reason to use a zobel is to flatten the impedance curve for the amp I wouldn't do it. There are advantages to decreasing the output current. Amplifier distortion normally increases with output current, and increased impedance at higher frequencies might help offset the typical increase in distortion at higher frequencies. This may not be audible, but I don't see a downside unless there is a passive crossover network or significant resistance between the amp and speaker(s).
If you do want a zobel you only need one across the entire array. You can use Xsim (the easiest route if you have the .zma file) or use one of the online calculators. Since Xsim allows you to easily change component values it is fairly simple to tune the zobel for flattest impedance, or even to "shelve" the impedance curve at a particular value.
If you do want a zobel you only need one across the entire array. You can use Xsim (the easiest route if you have the .zma file) or use one of the online calculators. Since Xsim allows you to easily change component values it is fairly simple to tune the zobel for flattest impedance, or even to "shelve" the impedance curve at a particular value.
Steve- I appreciate your valuable input. I do need a Zobel, as I am forced to use a passive xover network. When I retired, I promised the Frau that she could have her living room back, so all of the tube amps, xovers, rat's nest of cables and hundreds of speaker experiments have sadly gone away.
I am now using a Yamaha HTR, and I need to keep the footprint of the loudspeakers very small. I'm using six 3" HiVi MB drivers and a single planar HF unit.
Anyway, I contacted Parts Express, where I bought the drivers, and explained the problem. Within 24 hrs, Chris sent me the parts values I need.
Now it's time to get started on the enclosures.
I took the easy way out- I am not very computer literate, and none of the recommended software here runs on a Mac laptop.
I am now using a Yamaha HTR, and I need to keep the footprint of the loudspeakers very small. I'm using six 3" HiVi MB drivers and a single planar HF unit.
Anyway, I contacted Parts Express, where I bought the drivers, and explained the problem. Within 24 hrs, Chris sent me the parts values I need.
Now it's time to get started on the enclosures.
I took the easy way out- I am not very computer literate, and none of the recommended software here runs on a Mac laptop.
It's not clear whether you are trying to manage the impedance on the driver side or the amp side. Either way though, the network shouldn't strictly be necessary. It would be helpful to see what problem you are trying to fix.
AllenB
I intend putting the Zobel between the MB driver and the low pass filter of the crossover. I'm using an array of HiVi B3N 3" drivers, and since I have heard they work well past the rise in the impedance curve, I decided the Zobel would be needed. I'm crossing rather high (4.5 KHz into a laminar tweeter that presents a purely resistive load.
Another rule I'm breaking, one of the arrays (center channel), will be slightly concave curved.
I'm using an MMMTMMM driver configuration. I have not yet decided if I want to curve the left and right speakers. I'll make this decision once I get the center "array" built.
I intend putting the Zobel between the MB driver and the low pass filter of the crossover. I'm using an array of HiVi B3N 3" drivers, and since I have heard they work well past the rise in the impedance curve, I decided the Zobel would be needed. I'm crossing rather high (4.5 KHz into a laminar tweeter that presents a purely resistive load.
Another rule I'm breaking, one of the arrays (center channel), will be slightly concave curved.
I'm using an MMMTMMM driver configuration. I have not yet decided if I want to curve the left and right speakers. I'll make this decision once I get the center "array" built.
If you were for example using the driver alone connected to an amp with the typical near zero source impedance, adding this RC network across the driver will not change the drive voltage seen at the driver terminals. Primarily, there will be no difference. If the source impedance is finite, and that includes by way of a low pass filter, the RC network will reduce the voltage seen at the driver at these higher frequencies. Of course, you'd adjust it until it is correct either way after all you have a goal to achieve, the point is the network won't increase the highs or assist the driver function beyond some point. Further to this there is more than one way to achieve a response. Wherever the RC network may be used, there is an alternative circuit that can achieve the same driver response without that branch. In other words the purpose of Zobel networks is to conjugate an impedance to make it resistive. If you use this to make filter calculations easier and proceed to add said filter around it then that ease is the only benefit you'll get. This is another story if you are talking about transmission lines or making a load easier to drive, these are not primarily significant to speaker design, sometimes used with a special amp in conjunction it might be.
- Status
- Not open for further replies.