Thanks for pointing out the high dissipation of R3. 400mw is a bit much. If the value of R3 is raised to 33k (47K) the dissipation through the resister is reduced to ~119mw(83mw) allowing the use of a smaller 1/4 (1/8 th) watt resistors. These resisters will still pass 1.9ma (1.33ma) through the voltage setting Q4. I checked the data sheets for both the MPSA18 (“This device is designed for low noise, high gain, applications at collector currents from 1uA to 50 mA”) and the also the data sheet for the MPSA42 which seems to have a turn on current of under 0.1ma. I think staying over 1ma will still give adequate voltage stabilization.
When I simulated the circuit changing R3 to 33k(47k) also changes the value of R4 to 1.13k (1.15k) in this case.
I am unsure if a pot is needed to correct for variations between the zvp3310’s although I imagine it would help the current sharing through the 2sk389. I imagine that a pot across R4 would accomplish this or possibly a small drain resistor, neither of which will work well on the opamp boards This worries me a bit because I think small changes in VGS of the zvp3310 might greatly changing the current sharing through the differential pair. At the same time Nelson didn’t use any drain resistors in his DIY opamp article when he used the zvp3310a, maybe that is a good sign?
dave
When I simulated the circuit changing R3 to 33k(47k) also changes the value of R4 to 1.13k (1.15k) in this case.
I am unsure if a pot is needed to correct for variations between the zvp3310’s although I imagine it would help the current sharing through the 2sk389. I imagine that a pot across R4 would accomplish this or possibly a small drain resistor, neither of which will work well on the opamp boards This worries me a bit because I think small changes in VGS of the zvp3310 might greatly changing the current sharing through the differential pair. At the same time Nelson didn’t use any drain resistors in his DIY opamp article when he used the zvp3310a, maybe that is a good sign?
dave
I read through this thread (and post 57) awhile ago and just got a little confused on the ‘drain resistor.’ Now I’m pretty sure that I got it as R3 & R4 refer to the drain of the 2sk389 and not the ZVP3310.
The opamp on the towel looks good.
I am curious as to how you added a 100 ohm gate resistor to the opamp board. Did you just cut the trace on the front of the board between Q1 and the ZVP3310 and then bridge in the 100 ohm resistor?
Do you have any idea what kind of dc offset variation you would get with different zvp3310’s using the same R3 & R4? Or do you plan on using a 10k pot in all of the opamps you build up this way. That would make for a lot of pots when making the crossover. I need to read some more about the importance of the offset. All I can recall is that I think at unity gain the importance is diminished.
Another day or two and I should be getting a little gift from mouser, then the real fun begins.
regards,
dave
The opamp on the towel looks good.
I am curious as to how you added a 100 ohm gate resistor to the opamp board. Did you just cut the trace on the front of the board between Q1 and the ZVP3310 and then bridge in the 100 ohm resistor?
Do you have any idea what kind of dc offset variation you would get with different zvp3310’s using the same R3 & R4? Or do you plan on using a 10k pot in all of the opamps you build up this way. That would make for a lot of pots when making the crossover. I need to read some more about the importance of the offset. All I can recall is that I think at unity gain the importance is diminished.
Another day or two and I should be getting a little gift from mouser, then the real fun begins.
regards,
dave
I installed the 100 ohm resistor in the gate/base hole and bent the gate and resistor leads and joined them together with solder. No cutting or drilling.
I have built three of these and only measured the first one for the actual drain resistance. I dont recall what the actual voltage was gate to source of the zvp3310, but I think it was around 3.3 volts. Since you are using a different CCS resistor the 1850 ohm I measured would be different. You could setup and measure the Vgs of a few zvp3310's for 10mA and design from there. A pot was easy and fit nicely. I plan on using these discrete op-amps for other projects with more gain so being able to adjust for zero offset is a plus.
BDP
I have built three of these and only measured the first one for the actual drain resistance. I dont recall what the actual voltage was gate to source of the zvp3310, but I think it was around 3.3 volts. Since you are using a different CCS resistor the 1850 ohm I measured would be different. You could setup and measure the Vgs of a few zvp3310's for 10mA and design from there. A pot was easy and fit nicely. I plan on using these discrete op-amps for other projects with more gain so being able to adjust for zero offset is a plus.
BDP
And how does it sound?
I´m considering the MOX opamp as a buffer and in a 2way active XO.
Other than moamps remark that it outperforms several integrated opamps, I have not found any remarks on the MOX sound quality in the threads I´ve read.
Could any of you who´ve built it please post some comments on the sound of these opamps?
Thanks in advance,
Johan
I´m considering the MOX opamp as a buffer and in a 2way active XO.
Other than moamps remark that it outperforms several integrated opamps, I have not found any remarks on the MOX sound quality in the threads I´ve read.
Could any of you who´ve built it please post some comments on the sound of these opamps?
Thanks in advance,
Johan
Hello,
Many people think you need a LR response of the active filter to achieve the” perfect” system response. In fact IMHO the electrical response of the filter is not that interesting, because it is the total system response that you hear.
Changing the Q to achieve a “bump” in a HP filter in a SUB is one way of making the subwoofer player lower.
You don’t really know how the interaction between the different sections in a speaker performs, and therefore it’s nice to have the option of different Q values.
The mox filter is indented for prototyping only, the final filter is easily made a lot smaller on either breadboard or on any second order filter module.
\Jens
Many people think you need a LR response of the active filter to achieve the” perfect” system response. In fact IMHO the electrical response of the filter is not that interesting, because it is the total system response that you hear.
Changing the Q to achieve a “bump” in a HP filter in a SUB is one way of making the subwoofer player lower.
You don’t really know how the interaction between the different sections in a speaker performs, and therefore it’s nice to have the option of different Q values.
The mox filter is indented for prototyping only, the final filter is easily made a lot smaller on either breadboard or on any second order filter module.
\Jens
JensRasmussen said:
Thanks for the reply Jens.
I had never considered the above! I'm guessing that the higher ranges of Q (ie above 1) will be of limited use to those staying away from a drivers natural roll-offs, like me.
Interestingly, i currently use an LR aligned active Xover, so there should be some interesting comparisions when i get the MOX built.
Thanks again.
Hi MOX users!
I'm not using the MOX, but I'm working on an active flter too.
I wanted to know: how's your grounding shceme in your active XO? Since there's multiple boards, each one having an output+gnd, input+gnd and power supply+gnd connector, can't this create some ground loops?
Thanks
Alex
I'm not using the MOX, but I'm working on an active flter too.
I wanted to know: how's your grounding shceme in your active XO? Since there's multiple boards, each one having an output+gnd, input+gnd and power supply+gnd connector, can't this create some ground loops?
Thanks
Alex
HF rolloff with 100x settings?
Hi all,
I finally got to test my 8 boards. All filter sections are completed, inputs and buffers still to customize.
The results are very good and consistent for the 1x and 10x frequencies.
But at the 100x settings (the 1 nF cap) I have a consistent HF rolloff for the HP section, ca. -1 dB at 20k. The rolloff starts well below 20k, so that the crossover frequency shifts significantly and the HP and LP lines don't cross at -3 dB anymore.
Any ideas as to the cause? I eliminated the following:
- op amps: tested OPA 2134, TL072, NE5532
- input: buffered or unbuffered
- all 8 boards react the same
- soundcard has reference (its other channel)
- input buffer alone, or any HP settings with the 100 nF and 10 nF caps (1x and 10x) are perfect at 0 dB at 20 kHz - so it couldn't be in/out problems anyway.
Whay should the smallest cap cause a HF rolloff in a HP filter?? I can only think of lead inductance or parasitic capacitace to ground.
Pics attached, self explaining I hope. Note especially where the 2000 Hz HP, using 200 Hz x10, has 0 dB at 20 kHz, while the 2500 Hz HP, using 25 Hz at 100x, has -1 dB at 20 kHz. The 25 Hz at x1 or at x10 is perfect at 0 dB at 20 kHz.
MBK
Hi all,
I finally got to test my 8 boards. All filter sections are completed, inputs and buffers still to customize.
The results are very good and consistent for the 1x and 10x frequencies.
But at the 100x settings (the 1 nF cap) I have a consistent HF rolloff for the HP section, ca. -1 dB at 20k. The rolloff starts well below 20k, so that the crossover frequency shifts significantly and the HP and LP lines don't cross at -3 dB anymore.
Any ideas as to the cause? I eliminated the following:
- op amps: tested OPA 2134, TL072, NE5532
- input: buffered or unbuffered
- all 8 boards react the same
- soundcard has reference (its other channel)
- input buffer alone, or any HP settings with the 100 nF and 10 nF caps (1x and 10x) are perfect at 0 dB at 20 kHz - so it couldn't be in/out problems anyway.
Whay should the smallest cap cause a HF rolloff in a HP filter?? I can only think of lead inductance or parasitic capacitace to ground.
Pics attached, self explaining I hope. Note especially where the 2000 Hz HP, using 200 Hz x10, has 0 dB at 20 kHz, while the 2500 Hz HP, using 25 Hz at 100x, has -1 dB at 20 kHz. The 25 Hz at x1 or at x10 is perfect at 0 dB at 20 kHz.
MBK
Attachments
Well, now, this is interesting. Ages ago I posted a similar pic showing the same odd behavior and was really scratching my head over it. To me, it did not look like rolloff--Fc was dead on, and the slope looked to be zero outside the crossover. It just looked like it was -1dB down.
Looking at your post, I also do not think Fc has been shifted when the level is corrected for.
http://www.diyaudio.com/forums/showthread.php?postid=383881#post383881
But Jens was unable to duplicate my results! What could be different about his setup and ours?
Looking at your post, I also do not think Fc has been shifted when the level is corrected for.
http://www.diyaudio.com/forums/showthread.php?postid=383881#post383881
But Jens was unable to duplicate my results! What could be different about his setup and ours?
Tiroth,
You're right, FC shifts only insofar as now the curves intersect only at higher F - in fact it is due to a loss of gain in the HP. It doesn't shift w/ frequency after it reaches its maximum, so it is likely not stray inductance/capacitance as I speculated. Could be that the formulas for Sallen key make some hidden assumptions which become invalid for certain ratios of caps and resistors... I didn't know/remember that you had the same results...
In retrospect it would have been better (for me) to cut all R values by a factor of, say, 5 (better also for low noise and for op amp balance regarding R on negative vs positive inputs...) and to multiply the caps by 5 (I don't need the 0.1 x setting anyway which would then require a very large cap). Hmm, or I can use an intermediate cap at the 0.1x position. But that would make for odd frequencies.
I guess now the only quick hack is to raise the Q for the HP to compensate. Or to live with it
MBK
You're right, FC shifts only insofar as now the curves intersect only at higher F - in fact it is due to a loss of gain in the HP. It doesn't shift w/ frequency after it reaches its maximum, so it is likely not stray inductance/capacitance as I speculated. Could be that the formulas for Sallen key make some hidden assumptions which become invalid for certain ratios of caps and resistors... I didn't know/remember that you had the same results...
In retrospect it would have been better (for me) to cut all R values by a factor of, say, 5 (better also for low noise and for op amp balance regarding R on negative vs positive inputs...) and to multiply the caps by 5 (I don't need the 0.1 x setting anyway which would then require a very large cap). Hmm, or I can use an intermediate cap at the 0.1x position. But that would make for odd frequencies.
I guess now the only quick hack is to raise the Q for the HP to compensate. Or to live with it
MBK
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.