Ray Alden in one of his books doesn't seem to think that Fh /Fl strictly must be equal to 8 or greater. So but he suggests narrowing the bandwidth of the mid-range driver,
Fh' / Fl' = (Fh / Fl) -1
where
Fh' = the upper cut-off frequency of the band-pass filter
Fl' = the lower cut-off frequency of the band-pass filter
Then finally,
Fh' * Fl' = Fh * Fl
So you could possibly keep your Fl = 500 Hz and Fh = 1.7 kHz and just narrow the band-width of the bandpass filter.
Don't know if you might be interested in doing this, but there it is.
cT
Fh' / Fl' = (Fh / Fl) -1
where
Fh' = the upper cut-off frequency of the band-pass filter
Fl' = the lower cut-off frequency of the band-pass filter
Then finally,
Fh' * Fl' = Fh * Fl
So you could possibly keep your Fl = 500 Hz and Fh = 1.7 kHz and just narrow the band-width of the bandpass filter.
Don't know if you might be interested in doing this, but there it is.
cT
I just printed out your crossover schematic and I think I see the problem with the tweeter! I thought that the tweeter pad resistor (shunt) was 15 ohms, but when printed it looks like 1.5 ohms!! No wonder the impedance is dropping as frequency is rising. I assume this is a typo? If it is 1.5 ohms in the sim then that will definitely be causing the droping impedance issue, and if it is 1.5 ohms in your implementation then the amp definitely won't like it!
edit: reason I was doing it was I was curious to see how the circuit modeled with my drivers (impedance only) Attached is impedance plot of just the mid and tweeter part of the crossover.
black is simmed impedance. Blue is midbass impedance green is tweeter impedance... Note that this is with 15 ohms in the shunt position on the tweeter not 1.5 ohms.
The dip below 1000 Hz appears to be mainly due to the bandpass circuit.
Tony.
edit: reason I was doing it was I was curious to see how the circuit modeled with my drivers (impedance only) Attached is impedance plot of just the mid and tweeter part of the crossover.
black is simmed impedance. Blue is midbass impedance green is tweeter impedance... Note that this is with 15 ohms in the shunt position on the tweeter not 1.5 ohms.
The dip below 1000 Hz appears to be mainly due to the bandpass circuit.
Tony.
Attachments
Last edited:
Regarding the impedance graph in post #11
If the culprit are some components in the tweeter section needed to attenuate the tweeter at frequencies > 5KHz, you can also put a series RC in parallel to the tweeter. See here for an example - read the crossover section: Zaph|Audio - ZA-SR71
Ralf
I do think it's bad, instead. What worries me is the trend at high frequencies which, if continues after 20KHz, can bring down the impedance to near zero. I wouldn't connect such a speaker to any amp, solid state or valve.
If the culprit are some components in the tweeter section needed to attenuate the tweeter at frequencies > 5KHz, you can also put a series RC in parallel to the tweeter. See here for an example - read the crossover section: Zaph|Audio - ZA-SR71
Ralf
As I said in the OP I am concerned about the tweeter impedance curve for exactly the reason you state. The need for a solution to that problem was one of the reasons I posted. When I said "that wasn't so bad..." I meant grabbing the chart out of PCD and converting it to post.
I know for a fact that the 800 Hz impedance dip comes from the mid because in PCD you can temporarily remove the woofer and tweeter drivers from the sim, and when I do that impedance dip is still there.
Thanks for you replies - I feel closer to a solution!
I know for a fact that the 800 Hz impedance dip comes from the mid because in PCD you can temporarily remove the woofer and tweeter drivers from the sim, and when I do that impedance dip is still there.
Thanks for you replies - I feel closer to a solution!
Basically what you are seeing is just the 1.5 ohms in parallel with the tweeter at high frequencies (and it is exacerbated by the addition of the 4.7uF accross the series resistor.
I think what you really need to do is re-design your Lpad. You probably shouldn't have anything less than about 8 ohms in parallel with the tweeter which means a bigger series resistor.
I don't know if you have ever used speakerworkshop, but the optimizer is very good for situations like this. You give it some reasonable starting values for the pad resistors, select a frequency range you are interested in, and tell it to optimize for your target response. You can even throw in your bypass cap (I use a small one of 1uF on my tweeter pad to boost stuff above about 10K).
Without the 4.7uF cap (and the lpad values as they are) you will probably see a minimum impedance in the high frequencies of around 3.4 ohms, and it should rise as frequency goes up.
How much attenuation do you actually need on the tweeter?
Tony.
I think what you really need to do is re-design your Lpad. You probably shouldn't have anything less than about 8 ohms in parallel with the tweeter which means a bigger series resistor.
I don't know if you have ever used speakerworkshop, but the optimizer is very good for situations like this. You give it some reasonable starting values for the pad resistors, select a frequency range you are interested in, and tell it to optimize for your target response. You can even throw in your bypass cap (I use a small one of 1uF on my tweeter pad to boost stuff above about 10K).
Without the 4.7uF cap (and the lpad values as they are) you will probably see a minimum impedance in the high frequencies of around 3.4 ohms, and it should rise as frequency goes up.
How much attenuation do you actually need on the tweeter?
Tony.
Last edited:
Wintermute is correct, your circuit has to lead to 1.5 ohms (or a little less) at high frequencies. Since you are padding down a fair amount you can probably rescale the network to give the same curve with higher average impedance. If you don't have access to optimizer programs then just change the 1.5 ohm to something sensible (say 6 ohms) and adjust the other values to see if you can get back to the curve you want.
A Zobel won't help. You have already normalized impedance via the low 1.5 ohm. Unfortunately you have normalized it too low.
You can generally re-optimize a circuit for higher impedance, especially if the circuit has loss at all frequencies. The only time you are stuck with a low impedance is when you want full (zero dB) output at some frequencies, or worse, if you are trying to get a little gain out of the circuit.
Regards,
David
A Zobel won't help. You have already normalized impedance via the low 1.5 ohm. Unfortunately you have normalized it too low.
You can generally re-optimize a circuit for higher impedance, especially if the circuit has loss at all frequencies. The only time you are stuck with a low impedance is when you want full (zero dB) output at some frequencies, or worse, if you are trying to get a little gain out of the circuit.
Regards,
David
Dave! So good to see you - been a while. Hope all is well with you and yours. 
I was kinda hoping for something that would raise the impedance above cutoff, but 'normalizing at 1.5 ohms' does sound pretty bad!
I started bread-boarding the crossover last night - will definitely make some mods to the tweet circuit before starting my tests. Thanks!
I was kinda hoping for something that would raise the impedance above cutoff, but 'normalizing at 1.5 ohms' does sound pretty bad!
I started bread-boarding the crossover last night - will definitely make some mods to the tweet circuit before starting my tests. Thanks!
Dave! So good to see you - been a while. Hope all is well with you and yours.
Things are great, thanks for asking. I've started a new job in the Boston area, so the last few months have been busy, but things are starting to settle down now.
Regards!
David
Just be aware that because of the position in the circuit (ie between the driver and the crossover) that changing the value of the series resistor in the pad is going to effect the transfer function of the rest of the network.
If you want something "safe" for a starting point, get rid of the 4.7 uF resistor that is in parallel with the 4.7 ohm series resistor. That way the minimum your amp should see is around 6 ohms at higher frequencies. (I'm not sure what I was thinking last night when I said 3.4 ohms)
But I think it likely the entire tweeter network will need to be re-done unfortunately. Consider putting the lpad before the rest of the tweeter network as that way tweaking it will not effect the functioning of the rest of the network.
Tony.
If you want something "safe" for a starting point, get rid of the 4.7 uF resistor that is in parallel with the 4.7 ohm series resistor. That way the minimum your amp should see is around 6 ohms at higher frequencies. (I'm not sure what I was thinking last night when I said 3.4 ohms)
But I think it likely the entire tweeter network will need to be re-done unfortunately. Consider putting the lpad before the rest of the tweeter network as that way tweaking it will not effect the functioning of the rest of the network.
Tony.
A new job, in this economy? Lucky! - and congrats!
Hope to see you around here more too.
This does put me in a bit of a quandary, because PCD has very limited options to sim those kind of changes. I figured it was a minimal tool, but I didn't think I'd outgrow it with my very first speaker!Just be aware that because of the position in the circuit (ie between the driver and the crossover) that changing the value of the series resistor in the pad is going to effect the transfer function of the rest of the network.
If you want something "safe" for a starting point, get rid of the 4.7 uF resistor that is in parallel with the 4.7 ohm series resistor. That way the minimum your amp should see is around 6 ohms at higher frequencies. (I'm not sure what I was thinking last night when I said 3.4 ohms)
But I think it likely the entire tweeter network will need to be re-done unfortunately. Consider putting the lpad before the rest of the tweeter network as that way tweaking it will not effect the functioning of the rest of the network.
Tony.
My primary choice would probably have been Akabak, but it has more of a learning curve than I really wanted to put up with right now.
How does speaker workshop compare, as far as learning curve?
I thought PCD had the option for before and after the other components?
Speaker workshop does have a bit of a steep learning curve, but if you are only going to do crossover simulation it is pretty straight forward.
import frd and zma files
create drivers
assign frd and zma's to said drivers.
create a network
Add components and drivers and wire them up.
tick box to simulate network impedance.
Calculate response.
That's the bear minimum.
I've been thinking about doing some docco on how to do this, but haven't gotten around to. The Unofficial Speaker workshop manual is available here --> http://home.exetel.com.au/wintermute/diyAudio/sw_manual.zip
Section six is the relevant section... Note that is much more complicated than what I do, but I have measurements taken in the final enclosures which simplifies things.
I have in the past used PCD to get to a reasonable point quickly, and then started to optimize in speaker workshop.
The optimizer takes a while to get the hang of. I found that you should limit the range of frequencies, and only optimize some of the components at any one time. Note that to use the optimizer you need to have a separate network for each driver. I tend to have separate nets and also a combined one (for getting the overall impedance).
Tony.
Speaker workshop does have a bit of a steep learning curve, but if you are only going to do crossover simulation it is pretty straight forward.
import frd and zma files
create drivers
assign frd and zma's to said drivers.
create a network
Add components and drivers and wire them up.
tick box to simulate network impedance.
Calculate response.
That's the bear minimum.
I've been thinking about doing some docco on how to do this, but haven't gotten around to. The Unofficial Speaker workshop manual is available here --> http://home.exetel.com.au/wintermute/diyAudio/sw_manual.zip
Section six is the relevant section... Note that is much more complicated than what I do, but I have measurements taken in the final enclosures which simplifies things.
I have in the past used PCD to get to a reasonable point quickly, and then started to optimize in speaker workshop.
The optimizer takes a while to get the hang of. I found that you should limit the range of frequencies, and only optimize some of the components at any one time. Note that to use the optimizer you need to have a separate network for each driver. I tend to have separate nets and also a combined one (for getting the overall impedance).
Tony.
I think your 800 Hz dip is due to too much overlap between drivers. That is what I suspected when I suggested you post circuit and simulation.
What happens to the simulation frequency response if you invert the mid polarity?
Without seeing driver (raw) component responses it would be hard to suggest improvements. Flying blind, I would probably shift the tweeter XO up to ~2k, and putting zobels on mid and woofer would not hurt. Change the tweeter crossover a bit, remove the 4.7 ohm resistor and the 1.5 ohm resistor, then put in a standard L-pad between crossover and driver. Put a capacitor in parallel with the series leg of the l-pad to get some horn compensation, if needed.
What happens to the simulation frequency response if you invert the mid polarity?
Without seeing driver (raw) component responses it would be hard to suggest improvements. Flying blind, I would probably shift the tweeter XO up to ~2k, and putting zobels on mid and woofer would not hurt. Change the tweeter crossover a bit, remove the 4.7 ohm resistor and the 1.5 ohm resistor, then put in a standard L-pad between crossover and driver. Put a capacitor in parallel with the series leg of the l-pad to get some horn compensation, if needed.
Last edited:
Say the woofer is around 7 ohms at 5k with no filter and the tweeter around 4 at 5K plus. So there would be a dip that goes maybe sub 3 ohms.
Does this matter at higher frequencies since there is less power up there. Obviously a subwoofer that runs at sub 3 ohm could cook an amp.
Heres what I was looking at. THanks https://www.parts-express.com/pedoc...rs150p-4a-reference-series-specifications.pdf
https://www.parts-express.com/pedocs/specs/264-1006--ne19vts-04-spec-sheet.pdf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.