I would be interested in burning a few boards. Would it be possible to make LR24 for bass and But12 for high?
SF Pyromaniac
SF Pyromaniac
I am also in for two (or one stereo, three way) boards (red).
What about using submini tubes?
6021? 6111?
Franz
What about using submini tubes?
6021? 6111?
Franz
Or this one:
Distortion in Positive- and Negative-Feedback Filters
It is known that the harmonic distortion of an active filter is greater than the distortion of the operational amplifier itself. Positive- and negative-feedback linear filters, namely, Butterworth, Chebyshev, Bessel, and band-pass filters, are analyzed. The distortion multiplication factor K(d) is defined and plotted versus frequency. The maximum values of K(d) found for several filter configurations by means of a computer program are given in a form useful for filter designers.
Author: Bonello, Oscar J.
Affiliation: Solidyne SRL, Buenos Aires, Argentina
JAES Volume 32 Issue 4 pp. 239-245; April 1984
The nice thing about this paper is that he plots the 'distortion aggravation factor' vs freq.
jd
Last edited:
I think these will be 2nd order boards, separate for bass and MR/treble. That way, you can cascade them to get 4th order or use an input RC to get 3rd order. Many people have sealed box units that have an acoustic 2nd order high pass, so it makes sense to have a 4th order LP and a 2nd order HP. Q is, of course, easily changed with component values.
After considering pentode cathode followers very carefully, I'm getting less enamored of that- the only advantage they bring is a lower input capacitance, but the input C of a plain old triode CF is already extremely low. Pentodes DO make lovely current sinks, and potentially excellent gyrators. Nonetheless, I'm running some distortion spectra to see if a pentode CF has any advantages that aren't obvious from theoretical considerations.
Sorry, no sub-minis. I'm only using cheap, easy-to-get tubes for this. 6AU6 seems to be the leading candidate, given my inventory.
O.K., when they are not easy-to-get.
I have a stock of 6111WA, thats why I was asking. But I just looked at the availability and the prices today.
I am still interested in two boards, even when the colour should not be red 😎
Franz
The 6AU6 has K and G3 on seperate pins. If these pins (7 and 2) are tied together externally it opens the possibility of literally hundreds of possible tube choices, triodes, tetrodes, pentodes and even hexodes. I posted a list somewhere in Petes "engineers amp" thread. That list was just the tubes that I have, there are lots more.
So I vote for the 6AU6 (I have hundreds of those) and everything else that I can plug into that socket.
I'd love to see one of the lonely triode/pentode tubes used, like the whole 6U8 family, or one of the many others... all easy to get and dirt cheap. Problem is, looking at curves, most of the triodes aren't all that linear...
6BA8 and 6CX8 don't look too bad... 6BA8 is a "$1 tube" at RES, 6CX8 pricey at $4. And 6JW8/ECF802, another $1 wonder.
6BA8 and 6CX8 don't look too bad... 6BA8 is a "$1 tube" at RES, 6CX8 pricey at $4. And 6JW8/ECF802, another $1 wonder.
Last edited:
Have you decided to do a 3-way board? I see lots of people requesting, but no "yes's" from Sy or Pete. I'd be in for two 3-way stereo boards, if that were the case.
Distortion depends on the filter
"Harmonic Distortion in a Class of Linear active Filter Networks"
Peter J. Billam
JAES v26 n6, p426, 1978
Reading these 2 papers reminds me of a similar issue I experienced on a particular filter design where an intermediate signal was being clipped near the transition frequency due to amplifier gain. I never thought of a similar effect happening in positive feedback filters with g=1 amplifiers e.g. Sallen-Key; a "duh" moment for sure.
Also shows how to calculate k(d) for a number of popular topologies and alignments. For example. if I use a Sallen-Key topology to implement a 2nd order Butterworth filter with Q=.707; which is 1/2 of the very popular Linkwitz-Riley crossover alignment, the distortion amplification is only 6db, or 2X, worst case. No where near the 38db, 83X worst case for the 3rd order Chebychev used as the example in the paper.
My conclusion is that *If I want to implement the LR4 crossover alignment with Sallen-Key topology*; then by using augmented CF stages I can reduce the amplifier distortion by at least 6db, gaining back the advantage of a passive (gyrator) circuit. In addition, I would expect the gyrator or coil to add some of it's own distortion. Not to mention the cleverness required to implement a LPF using gyrators.
Maybe it's expected that people will want to use 3rd order chebychev filters, or other alignments that don't lend themselves so well to Sallen-Key as LR4.
I'm still not sure which way to go... maybe I'll need to try both and see if there is a difference for me with LR4.
Cheers,
Michael
"Harmonic Distortion in a Class of Linear active Filter Networks"
Peter J. Billam
JAES v26 n6, p426, 1978
Reading these 2 papers reminds me of a similar issue I experienced on a particular filter design where an intermediate signal was being clipped near the transition frequency due to amplifier gain. I never thought of a similar effect happening in positive feedback filters with g=1 amplifiers e.g. Sallen-Key; a "duh" moment for sure.
Also shows how to calculate k(d) for a number of popular topologies and alignments. For example. if I use a Sallen-Key topology to implement a 2nd order Butterworth filter with Q=.707; which is 1/2 of the very popular Linkwitz-Riley crossover alignment, the distortion amplification is only 6db, or 2X, worst case. No where near the 38db, 83X worst case for the 3rd order Chebychev used as the example in the paper.
My conclusion is that *If I want to implement the LR4 crossover alignment with Sallen-Key topology*; then by using augmented CF stages I can reduce the amplifier distortion by at least 6db, gaining back the advantage of a passive (gyrator) circuit. In addition, I would expect the gyrator or coil to add some of it's own distortion. Not to mention the cleverness required to implement a LPF using gyrators.
Maybe it's expected that people will want to use 3rd order chebychev filters, or other alignments that don't lend themselves so well to Sallen-Key as LR4.
I'm still not sure which way to go... maybe I'll need to try both and see if there is a difference for me with LR4.
Cheers,
Michael
Consider the gyrator, implemented using a nonideal active device (triode, pentode, MOSFET. BJT, opamp...) which has some nonlinearity in it's current vs. voltage curve.
This would be expected to have it's maximum contribution to the signal when the gyrator is working hardest, but before it presents a low impedance to the signal, i.e. *the gyrator's active device nonlinearity will contribute it's greatest to the signal at the filter's transition frequency* ironically the same region of largest K(d) seen for example in the Sallen-Key topology...
Are there any measurements of any distortion added by different kinds of active devices used to implement the gyrator?
Seems like Spice would be a good platform for evaluating this if the device models express the nonlinearity well enough.
Cheers,
Michael
My Acheron measurements (cheap opamp gyrator, triode CF) showed distortion below my reliable measurement threshold (0.002%) at crossover (110Hz in my case). No pentode measurements yet. The LPF will probably be implemented in solid state- that really works just fine for bass.
I have just read SY's Acheron paper and found it very interesting. Thanks, SY.
I then modelled with LTSpice my tweeter filter originally designed using opamp / Shallen-Key, and it only took 5 minutes to produce the exact transfer function using the gyrator. No big deal. I guess I may next try a low pass filter for my midrange bandpass.
Next I searched the forum for more information about the Acheron, and first found this: http://www.diyaudio.com/forums/solid-state/119548-op-amp-gyrator-oscillates.html
There is still an unanswered question why it did not like the 100nF cap (quite large already) and oscillated.
I am mirroring Michael in his critical thinking, and have long been wondering if an opamp can work well simulating an inductor and what sorts of precautions I need to take to make it work optimally.
I am just a learner.
Regards,
Bill
I then modelled with LTSpice my tweeter filter originally designed using opamp / Shallen-Key, and it only took 5 minutes to produce the exact transfer function using the gyrator. No big deal. I guess I may next try a low pass filter for my midrange bandpass.
Next I searched the forum for more information about the Acheron, and first found this: http://www.diyaudio.com/forums/solid-state/119548-op-amp-gyrator-oscillates.html
There is still an unanswered question why it did not like the 100nF cap (quite large already) and oscillated.
I am mirroring Michael in his critical thinking, and have long been wondering if an opamp can work well simulating an inductor and what sorts of precautions I need to take to make it work optimally.
I am just a learner.
Regards,
Bill
Michael,
Your posts are very enlightening! Thanks a great deal.
I have an issue with what you proposed in the above though, as right now I am designing my active filters with my 4 way active speakers and am thinking the issues through.
In real life, a filter with an elctrical LR4 transfer function won't work unless the loudspeaker driver response is flat, which is never the case. I guess I can only design a filter that shapes the mounted driver response to acoustic LR4, in which case, the filter itself is more like a second or third electrical order.
For example, consider that we design a filter for the tweeter XO at 2kHz. Most tweeters start rolling off at 2kHz, and the rolloff is usually low Q. In my case I found the only way to shape the acoustic response to LR4 using a single opamp stage with Shellen-Key is to use a high Q 2nd order similar to Chebychev, which combines with a 1st order filter and the driver roll off yielding exactly LR4 acoustic response. But the Q of the Shellen-key is quite high and I would be worried that the distortion is 38dB.
I have not found a solution yet.
Regards,
Bill
Michael,
Please ignore my post above, it was due to my ignorance. 😀
I spent a bit of time looking at my HP LR4 filter with LTSpice.
I initially used a high Q Shellen-Key because I wanted to use a single opamp to achieve an electrical 3rd order yielding acoustical LR4. In that case, the 2nd order HP must have high Q.
If I use an additional opamp intead of a single opamp, then the 2nd order needs not to be high Q.
So the problem is solved. In fact, I don't use an additional opamp, but simply replace the 1st order HP by putting a cap at the output of the opamp to form a 1st HP with the power amp input impedance. This saves an opamp, It is not good for a commercial product but good enogh for my own DIY system.
🙂
Regards,
Bill
Please ignore my post above, it was due to my ignorance. 😀
I spent a bit of time looking at my HP LR4 filter with LTSpice.
I initially used a high Q Shellen-Key because I wanted to use a single opamp to achieve an electrical 3rd order yielding acoustical LR4. In that case, the 2nd order HP must have high Q.
If I use an additional opamp intead of a single opamp, then the 2nd order needs not to be high Q.
So the problem is solved. In fact, I don't use an additional opamp, but simply replace the 1st order HP by putting a cap at the output of the opamp to form a 1st HP with the power amp input impedance. This saves an opamp, It is not good for a commercial product but good enogh for my own DIY system.
🙂
Regards,
Bill
- Status
- Not open for further replies.