You may have considered listing the on-line resources such as Texas Instruments "Filter-Pro" and "TINA", LT-Spice from Linear Tech, and the free-ware version of MultiSim at Analog Devices.
Accurate Mechanical Time-Alignment
I hope your book includes all the advantages of achieving correct mechanical time-alignment of all the drivers and the improved sound character that results when phase / time shift correction circuits are removed.
I hope your book includes all the advantages of achieving correct mechanical time-alignment of all the drivers and the improved sound character that results when phase / time shift correction circuits are removed.
The same for me !
I will send you that today in French on your website new email:
drgself@dsl.pipex.com
It contains a long chapter on achieving time-alignment in one way or another.
Single (opamp) stage 3rd or 4th order filter? I don't have any idea how to do that. Suppose it is a first order filter or a cascade of two first order filters plus a Sallen-Key, the Sallen-Key needs to have a very high Q to yield a useful Q of the total filter. I read from some source that the distortion associated with high Q Sallen-Key is astronomical. I am wondering if your book provides a nice alternative or a remedy.
Regards,
Bill
Regards,
Bill
SK can do up to about Q=3 and still be OK, and that is good enough for your typical loudspeaker crossover filters. For instance, Butterworth 3rd order only requires Q=1 from the second order section. I'm not sure what type of filter you had in mind...
You can use a Multiple-Feedback filter instead of Sallen Key, if you need high Q's. The MFB is also a single op amp filter.
I am developing a PC board that can do both SK and MFB, up to third order, if you are interested. See this post and the thread for more info:
http://www.diyaudio.com/forums/anal...bilities-i-want-your-input-9.html#post2460607
-Charlie
With regards to the distortion of high Q filter implemented as Sallen-Key, you can refer to this post:
http://www.diyaudio.com/forums/tubes-valves/161644-sy-pete-millett-crossover-6.html#post2098961
I most certainly don't want the distortion amplification of 38db, 83X worst case for the 3rd order Chebychev.
Do you mean MFB filter has much lower distortion figure than Sallen-Key for high Q filters? Could you point me to a link with a single op amp MFB?
Thanks in advance.
Regards,
Bill
http://www.diyaudio.com/forums/tubes-valves/161644-sy-pete-millett-crossover-6.html#post2098961
I most certainly don't want the distortion amplification of 38db, 83X worst case for the 3rd order Chebychev.
Do you mean MFB filter has much lower distortion figure than Sallen-Key for high Q filters? Could you point me to a link with a single op amp MFB?
Thanks in advance.
Regards,
Bill
Last edited:
The issue I have with a generic, active, configurable analogue XO board is that it could possibly produce nice sounding loudspeakers but not top notch sounding loudspeakers because it must have too many opamp stages yielding long power supply tracks, i.e. higher inductance that can not be totally surpressed with local bypass.
I found opamps (I use opa627 for tweeter and midrange) won't sound good unless with an extremely well designed power supply with diminishing output impedance to high frequencies. Local bypass with film caps, if not carefully done, can cause ringing due to LCR resonance. Very short tracks between the output of the power supply regulator and the opamps are essential. I had previously 3 opamps in the tweeter circuit (acoustic LR4 + a notch), 7 opamps in the midrange (acoustic LR4 LP, acoustic LR4 HP, 3 notches and 1 Q boost) and 7 opamps in the woofer circuits, each driven by its own power supply. When I reduced them to 2 opamps in the tweeter, 3 opamps in the midrange and 5 opamps in the woofer circuits (by ditching all notches and Q boosts, and accept some irregularities in the frequency domain), there was an significant increase in sound clarity, probably because I could then shorten the path between the last opamp to the power supply regulator.
Yes I confess it is the hard way, but a DIYer who looks for the best result can not avoid spending days, weeks and months to design and experiment with his custom designed active XO, the only way to achieve top notch result if going the analogue way. By the way, I installed 24 x 25 turn trimpots for fine adjustments of levels, XO points and Q, which I found to be extremely handy.
Regards,
Bill
I found opamps (I use opa627 for tweeter and midrange) won't sound good unless with an extremely well designed power supply with diminishing output impedance to high frequencies. Local bypass with film caps, if not carefully done, can cause ringing due to LCR resonance. Very short tracks between the output of the power supply regulator and the opamps are essential. I had previously 3 opamps in the tweeter circuit (acoustic LR4 + a notch), 7 opamps in the midrange (acoustic LR4 LP, acoustic LR4 HP, 3 notches and 1 Q boost) and 7 opamps in the woofer circuits, each driven by its own power supply. When I reduced them to 2 opamps in the tweeter, 3 opamps in the midrange and 5 opamps in the woofer circuits (by ditching all notches and Q boosts, and accept some irregularities in the frequency domain), there was an significant increase in sound clarity, probably because I could then shorten the path between the last opamp to the power supply regulator.
Yes I confess it is the hard way, but a DIYer who looks for the best result can not avoid spending days, weeks and months to design and experiment with his custom designed active XO, the only way to achieve top notch result if going the analogue way. By the way, I installed 24 x 25 turn trimpots for fine adjustments of levels, XO points and Q, which I found to be extremely handy.
Regards,
Bill
There is a practical application of a high Q HP filter.
Considering a typical open baffle loudspeaker where there is a 6dB/oct dipole roll-off towards lower frequencies, for the high pass filter of the midrange, a typical implementation would include a shelving low pass filter lower down, or notch filter higher up, or both to counter the dipole roll-off, plus a high pass at the XO. But this could well be done with a single high Q high pass filter in a simple Sallen-Key, provided that if the distortion is not over the top. All this could be done with a single opamp as a second order electrical filter combining with the acoustic roll-off of the driver could yield a typical, desirable LR4 response.
Regards,
Bill
Bill,
I'm sure that there are many practical applications for high Q filters. I'm not sure why you mentioned Sallen Key again, or why you are fixated on op amp distortion. There are op amps now like the LME49720 that have 0.00003% distortion (although this varies with signal level and frequency like any other amplifier).
The problem with the Sallen Key filter for realizing high Q sections is not distortion as much as it is the increasing sensitivity to component values as Q increases. This means when you assemble the filter, even from 1% components, the performance is way off when Q is "high". There are many mentions of this in the literature and a simple Google search about the SK filter would give you plenty to read about it. For instance, you can read a section from an online Walt Jung "Google book" at this link.
High component sensitivity also implies high temperature sensitivity, so even if you trimmed out the values, they would likely drift.
If you keep reading the Jung book, you will get to a section on MFB fiters. Although it warns about problems realizing "high frequency high Q" filters, this is not a problem for audio so much, and the Q can easily be up to 20, which is way more than you would use in any crossover.
-Charlie
Probably because of Billam's analysis in JAES. I'd be interested to see experimental verification (or refutation) of his claims.
I used to take the published figures as all I needed to know but I now look at them from different angles.
The diminishing low distortion figures do not always correlate to subjective sound quality. For example, the datasheet of LME49720 says 120dB PSSR. Further examination showed at 20kHz it is only -70dB for the negative rail. At 200kHz it is only -50dB. Junks at high frequencies at the rails would mix into the signal. This applies to all opamps.
I have spent over two years playing with opamp and have now just managed to have reasonably clean sound (i.e. can be bettered) out of the tweeter circuit, relying on a shunt regulator with sub milli ohm output impedance way up to hundreds of kHz, and using two opamps only in the circuit and placing them almost directly at the output of the regulator. Longer tracks would degrade the sound. For midrange and woofer circuits, they are less critical. It is for this reason, I don't know how a generic board that consists of half a dozen or more opamps work.
So I think I will buy the new blameless book from our great master to learn more about low noise implementation of opamps, etc. It will be helpful.
Regards,
Bill
I see this is a non-issue for audio applications. What amplifier or speaker is going to reproduce this 200k Hz "junk"? Where is the problem, exactly?
You betcha.
-Charlie
Documentation exists that shows RF entering an active gain stage (op amp) through the supply rails, or any input / output, can cause LF IM products that clearly degrade distortion performance in the audio band.
With op amps that have bipolar input stages, the input clamping diodes act as detectors and you can hear radio broadcasts if you are near a transmitter ! ! !
Last edited:
Hmm, sounds fishy to me. From what I know about IM, in order for these IM products to be at a high enough level, you need two or more closely spaced relatively strong signals (correct me if I am wrong here). I could see noise entering from the PS rails, but just not sure how that kind of broadband signal leads to significant IM. Can you point me to this "documentation" that might explain (not just report) this?
I can relate to this one... My father bought a really, really poor quality receiver on "super close out" from the local all-in-one retailer around 1982. Every now and then, the FM radio circuit would pick up some communication from a plane to the local airport tower!
Anyway, I thought that the main path through which this kind of signal entered the amplifier was via the output (and back from there, via feedback paths). I wasn't aware that any decent power supply could also be a source of this kind of pickup. I definitely need to do some reading on this. I recently picked up a book on amplifier design and I need to review the relevant chapters.
-Charlie
Don't remember the reference, probably data sheet for op amp. I agree, RF can't enter through supply rails with even half-way descent power supply and close decouple caps.
My experience with bipolar input op amps was in RIAA preamp where gain was very high. Adding 50 to 100 pf across + and - inputs usually suppresed all artifacts.
I live near the TV transmitters in our area and I used to have to wait until channel 6 signed off at night to really enjoy the music. My phono preamp had hum that I just could not seem to find source of. One evening I happened to look at the signal on a scope and WOW -- there was the vertical retrace pulse drifting across the screen that was synced to 60 Hz line. TV 60 Hz retrace is slightly different than 60 Hz line freq. I watched as channel 6 pulled the B+ and like magic the hum vanished. Same thing happened with my car cassette player; as I drove nearer to the xmitter towers hum would increase. All of this was due to bipolar transistor inputs. When I replaced bipolar OP37 op amps with FET input OPA637, no more problems like this.
My experience with bipolar input op amps was in RIAA preamp where gain was very high. Adding 50 to 100 pf across + and - inputs usually suppresed all artifacts.
I live near the TV transmitters in our area and I used to have to wait until channel 6 signed off at night to really enjoy the music. My phono preamp had hum that I just could not seem to find source of. One evening I happened to look at the signal on a scope and WOW -- there was the vertical retrace pulse drifting across the screen that was synced to 60 Hz line. TV 60 Hz retrace is slightly different than 60 Hz line freq. I watched as channel 6 pulled the B+ and like magic the hum vanished. Same thing happened with my car cassette player; as I drove nearer to the xmitter towers hum would increase. All of this was due to bipolar transistor inputs. When I replaced bipolar OP37 op amps with FET input OPA637, no more problems like this.
I bypass my supply rails . . . don't you?
I lowpass my inputs . . . don't you?
And I don't put low (signal) level devices in unshielded boxes . . . do you?
Yes to all, had no effect. RF entered through the input cables from the turntable. Cassette player was bought, not built, so what can you do? The 50 to 100 pf across + and - inputs was LP function.
By close to xmitter towers, I mean a few 1000 feet. Worst case situation. At least now all of my equipment has very robust RF rejection.
By close to xmitter towers, I mean a few 1000 feet. Worst case situation. At least now all of my equipment has very robust RF rejection.
Last edited:
Oh, *that* problem. But not quite worst. I worked briefly at a FM station in the '60s . . . transmitter (250 watt) in the back room, antenna on the roof. RF everywhere. Interconnects become antennas, RF all over the "shields". We had to re-work almost everything that came into the studio. It was essentially impossible to get a FM tuner to work (to listen to anyone else). The "good old days" . . . it is *so* much easier now . . .
- Status
- Not open for further replies.