Series-Parallel
Gentlemen,
Lucky me!
With my Ashly MOSFET power amp I get to use series connected speakers without volume loss (compared to parallel connected speakers).
Thank you very much for the detailed discussion. I don't know of any other discussion forum that would have provided this explanation.
dobias
Gentlemen,
Lucky me!
With my Ashly MOSFET power amp I get to use series connected speakers without volume loss (compared to parallel connected speakers).
Thank you very much for the detailed discussion. I don't know of any other discussion forum that would have provided this explanation.
dobias
Quick note on parallel feed - I've done this for the input/driver stage on a power amplifier and was not happy with the result. For one thing, you must use a fairly large capacitor (2 uF or larger) otherwise there will be a large LF peak (several dB) as a result of the cap resonating with the (quite large) inductance of the transformer.
This large cap connected in the direct signal path (which also develops a significant voltage across it at LF thanks to its reactance) has a major impact on the overall sonics of the amplifier. I tried many caps (and combinations of caps) and was not able to find any that had less coloration than the entire amp/preamp combination. This particular circuit has the property of subjectively magnifying cap-coloration, an unexpected and novel result.
You may be one of the people that doesn't hear cap-coloration - if so, good for you! But I've worked with speakers long enough I find this particular coloration quite offensive and annoying, since (for me) it represents a substantial loss of low-level resolution, a quite strong "plastic" coloration, and sometimes outright graininess (with metallized polypropylene at any price level).
It's part of the reason I try and keep crossover complexity to a reasonably low level - finding good-sounding caps is just another hassle in subjectively balancing a speaker, and it doesn't help that subjective cap-coloration can mimic driver colorations.
By way of example, when I developed the Ariels and originally started out with Sprague 730P caps, I found I had to make a 1/2 dB level adjustment when switching to the same-value Hovland film-and-foil caps. I should add this subjective adjustment was required not just with music, but pink-noise stimulus as well. The difference between these two quite-good polypropylene caps was so large I was having to re-balance the entire crossover - not a lot, but some, and the re-balancing was not optional, the speaker sounded out-of-balance without doing it.
Yes, I've tried paralleling dissimilar-construction caps to "balance" the sonics, but that can quite easily result in the worst aspect of each cap coming to the fore, not the best. That practice is pretty much like rolling the dice, no way of telling what it'll sound like. It's the kind of thing parts-rollers can spend weeks, months and years doing - no thanks!
I do like Teflon caps, but finding them in several-uF values is astronomically expensive, sometimes more than the driver itself.
This large cap connected in the direct signal path (which also develops a significant voltage across it at LF thanks to its reactance) has a major impact on the overall sonics of the amplifier. I tried many caps (and combinations of caps) and was not able to find any that had less coloration than the entire amp/preamp combination. This particular circuit has the property of subjectively magnifying cap-coloration, an unexpected and novel result.
You may be one of the people that doesn't hear cap-coloration - if so, good for you! But I've worked with speakers long enough I find this particular coloration quite offensive and annoying, since (for me) it represents a substantial loss of low-level resolution, a quite strong "plastic" coloration, and sometimes outright graininess (with metallized polypropylene at any price level).
It's part of the reason I try and keep crossover complexity to a reasonably low level - finding good-sounding caps is just another hassle in subjectively balancing a speaker, and it doesn't help that subjective cap-coloration can mimic driver colorations.
By way of example, when I developed the Ariels and originally started out with Sprague 730P caps, I found I had to make a 1/2 dB level adjustment when switching to the same-value Hovland film-and-foil caps. I should add this subjective adjustment was required not just with music, but pink-noise stimulus as well. The difference between these two quite-good polypropylene caps was so large I was having to re-balance the entire crossover - not a lot, but some, and the re-balancing was not optional, the speaker sounded out-of-balance without doing it.
Yes, I've tried paralleling dissimilar-construction caps to "balance" the sonics, but that can quite easily result in the worst aspect of each cap coming to the fore, not the best. That practice is pretty much like rolling the dice, no way of telling what it'll sound like. It's the kind of thing parts-rollers can spend weeks, months and years doing - no thanks!
I do like Teflon caps, but finding them in several-uF values is astronomically expensive, sometimes more than the driver itself.
Another quick note - the SPL at the listener position is the vector sum of the driver(s) radiation. Thus, if there is a phase difference due to mismatched phase response of the drivers or dissimilar path-lengths to the listener, you get less than the simple voltage-addition of the direct electrical equation. (Although drivers working in the piston-band should have response that is both flat and minimum-phase - the only exception should be a very narrow resonance from the spider and/or surround in the 300~500 Hz region.)
The quartet of the drivers for the bass-module can take advantage of the floor image, but this is most applicable for the drivers closest to the floor. By the time we get to the mid drivers, the floor reflection is sufficiently delayed (about 2.5~3 mSec) it is no longer contributing to the subjective increase in level - so, by industry convention, the FR measurements and level calculations do not include the contribution from the floor. Don't count on absorption from the carpet, either - I found that carpets don't attenuate significantly until you get to 8 kHz.
Although this whole business of floor reflections is intellectually disturbing - taking advantage of it at some frequencies, and ignoring it at others - it is a fact of everyday experience, and its absence is what is unacceptable in real-world terms.
Remember, most recording microphones are either close-miked (thus a very weak reflection) or are suspended high above the performing ensemble, and the initial recording was equalized with soffit-mounted professional monitor loudspeakers that had a generous floor reflection from the studio floor. Take away that floor reflection from your listening room, and it won't sound right. For that matter, just speaking in a room without a floor reflection is going to sound very odd and I'd guess more than a little unnatural.
The floor reflection is predictable (because we already know the height of the driver locations), but the reflections from the rear and side walls are going to depend on each listener's room and the location of the speaker within it. These reflections (and associated standing-waves) are the best single argument for bi-amping and the separate Left and Right EQ for the bass module, as well as the efficiency argument previously demonstrated by JohnK's modelling.
As posted previously, I don't believe in using EQ to correct for room abberations above 300 Hz - above that frequency, I'll only correct for deviations from the drivers themselves, and preferably in the passive domain using the crossover. Below that frequency, though, EQ on a per-channel basis is a good idea, provided boost EQ is kept to a minimum (because it stresses both amplifiers and drivers, and reduces system headroom).
The quartet of the drivers for the bass-module can take advantage of the floor image, but this is most applicable for the drivers closest to the floor. By the time we get to the mid drivers, the floor reflection is sufficiently delayed (about 2.5~3 mSec) it is no longer contributing to the subjective increase in level - so, by industry convention, the FR measurements and level calculations do not include the contribution from the floor. Don't count on absorption from the carpet, either - I found that carpets don't attenuate significantly until you get to 8 kHz.
Although this whole business of floor reflections is intellectually disturbing - taking advantage of it at some frequencies, and ignoring it at others - it is a fact of everyday experience, and its absence is what is unacceptable in real-world terms.
Remember, most recording microphones are either close-miked (thus a very weak reflection) or are suspended high above the performing ensemble, and the initial recording was equalized with soffit-mounted professional monitor loudspeakers that had a generous floor reflection from the studio floor. Take away that floor reflection from your listening room, and it won't sound right. For that matter, just speaking in a room without a floor reflection is going to sound very odd and I'd guess more than a little unnatural.
The floor reflection is predictable (because we already know the height of the driver locations), but the reflections from the rear and side walls are going to depend on each listener's room and the location of the speaker within it. These reflections (and associated standing-waves) are the best single argument for bi-amping and the separate Left and Right EQ for the bass module, as well as the efficiency argument previously demonstrated by JohnK's modelling.
As posted previously, I don't believe in using EQ to correct for room abberations above 300 Hz - above that frequency, I'll only correct for deviations from the drivers themselves, and preferably in the passive domain using the crossover. Below that frequency, though, EQ on a per-channel basis is a good idea, provided boost EQ is kept to a minimum (because it stresses both amplifiers and drivers, and reduces system headroom).
Lynn, I have made speakers in the past with vertically distributed multiple woofers, that had multiple path lengths to the floor and well spread out mid bass cancellations. Those measured with almost no floor related 100-200Hz usual SPL notch. They had a very welcome body and ambient cues benefit in the power range. Nothing sounded weird. Roy Allison criteria OK speakers sound weird? Recordings are normally produced over highly suspended studio monitors, and the engineers listen to their ceiling notch, but they never release if not sounding good on headphones and near fields too. Its more complex.
The floor reflection is naturally acceptable in my view, but it still detracts.
The floor reflection is naturally acceptable in my view, but it still detracts.
re: Lynn's parallel feed/capacitor comments
I note his observations. MSL (MQracing)/Mike LeFevre (whose transformers ARE wonderful- I have one of his Peerless preamp power transformers) is a strong advocate for parallel-feed (see AA, Bottlehead, etc.) and notes the presence of the last PS cap in the circuit of a series-feed arrangement (though this is not necessarily applicable to Lynn's differential designs-see his output-circuit current loop article). I have resisted this approach, primarily because of this low-frequency resonance-it is apparently model-able(?) (though I admit I haven't pursued these models with rigor), but the idea of using/having to account for a resonance to augment/tailor the low-frequency response seems inelegant (i.e. analogous to a ported low-frequency alignment-something this entire thread is predicated on avoiding).
Bill
I note his observations. MSL (MQracing)/Mike LeFevre (whose transformers ARE wonderful- I have one of his Peerless preamp power transformers) is a strong advocate for parallel-feed (see AA, Bottlehead, etc.) and notes the presence of the last PS cap in the circuit of a series-feed arrangement (though this is not necessarily applicable to Lynn's differential designs-see his output-circuit current loop article). I have resisted this approach, primarily because of this low-frequency resonance-it is apparently model-able(?) (though I admit I haven't pursued these models with rigor), but the idea of using/having to account for a resonance to augment/tailor the low-frequency response seems inelegant (i.e. analogous to a ported low-frequency alignment-something this entire thread is predicated on avoiding).
Bill
Back on topic
Hi Lynn, thanks for your comments. I take your advice on capacitors seriously!
Anyway, back to the task at hand: I was fortunate enough to be given a pile of Speaker Builder magazines and found gold in 6/98 where there is an article on arraying OB midrange drivers (Focal 7K2, 96dB) in different arrangements to get directivity control. Sorta like Magnetar has done with his horizontal paired midranges, but using Earl Geddes-type foam to selectively attenuate certain drivers for pattern control. The tweeter used was a Raven ribbon.
Are you familiar with the article? Perhaps you could comment.
Best regards,
Mike Spence
Hi Lynn, thanks for your comments. I take your advice on capacitors seriously!
Anyway, back to the task at hand: I was fortunate enough to be given a pile of Speaker Builder magazines and found gold in 6/98 where there is an article on arraying OB midrange drivers (Focal 7K2, 96dB) in different arrangements to get directivity control. Sorta like Magnetar has done with his horizontal paired midranges, but using Earl Geddes-type foam to selectively attenuate certain drivers for pattern control. The tweeter used was a Raven ribbon.
Are you familiar with the article? Perhaps you could comment.
Best regards,
Mike Spence
Re: Back on topic
Ah! Finally someone mentioned this. It's been in my memroy for a long time and I failed to find any soft or hard copies to proof it's not my own dream or imagination.
As I remember, this is a special OB with 4* Focal 15 inchers on V shape bass baffle per side. And the mid-high module is a separate small baffle with 4* 7 inchers arranged as a cross with Raven in the center....
I've been searching this on web for years! Anyone has related info for sharing?
mikey_audiogeek said:
Ah! Finally someone mentioned this. It's been in my memroy for a long time and I failed to find any soft or hard copies to proof it's not my own dream or imagination.
As I remember, this is a special OB with 4* Focal 15 inchers on V shape bass baffle per side. And the mid-high module is a separate small baffle with 4* 7 inchers arranged as a cross with Raven in the center....
I've been searching this on web for years! Anyone has related info for sharing?
More on Speaker Builder
Basically they tried TM, MTM, MMTMM, and a diamond array and measured the directivity in the horizontal plane. The mids were 6" focal 7k2, 96dB, quite similar performance to Audax PR170M0 from what I can tell.
The goal was to narrow the pattern compared to a single mid at frequencies above 500Hz.
The best arrangement (ie gave smoothest directivity between 1-7K) was with MT and a mid either side of the T. In other words, the diamond array but with the bottom mid eliminated.
The "side" mids were covered in 4" reticulated foam to attenuate frequencies above 2K.
The side mids were wired in series, with the series pair wired in parallel with the "main" woofer.
The reticulated foam was also shaped around the ribbon tweeeter and lined with felt to give a minor waveguide effect.
After re-reading the article I think it may have some correlation with objectives expressed in this thread. Can someone advise on the ethics of copying or sharing the article?
Best regards,
Mike
Basically they tried TM, MTM, MMTMM, and a diamond array and measured the directivity in the horizontal plane. The mids were 6" focal 7k2, 96dB, quite similar performance to Audax PR170M0 from what I can tell.
The goal was to narrow the pattern compared to a single mid at frequencies above 500Hz.
The best arrangement (ie gave smoothest directivity between 1-7K) was with MT and a mid either side of the T. In other words, the diamond array but with the bottom mid eliminated.
The "side" mids were covered in 4" reticulated foam to attenuate frequencies above 2K.
The side mids were wired in series, with the series pair wired in parallel with the "main" woofer.
The reticulated foam was also shaped around the ribbon tweeeter and lined with felt to give a minor waveguide effect.
After re-reading the article I think it may have some correlation with objectives expressed in this thread. Can someone advise on the ethics of copying or sharing the article?
Best regards,
Mike
The reason I was interested in this, is basically I like big coaxes, but everything on the market has a problem that I don't beleive lets them work best. The two 'best' I reckon are the Altec 604 and the PAudio BM8CX, from those I've tried.
So, my idea was the diamond pattern of 18Sound 6ND410, say 400-2k active LR xover to a CD on the 18Sound XT120 flare or another flare design I have that has the same OD as the PDC of the 6ND410. The side drivers would be rolled out earlier.
The alternativewould be a pair of frame to frame side by side by side 6.5's above and below the HF flare. The more I think about it, this might be better.
So, my idea was the diamond pattern of 18Sound 6ND410, say 400-2k active LR xover to a CD on the 18Sound XT120 flare or another flare design I have that has the same OD as the PDC of the 6ND410. The side drivers would be rolled out earlier.
The alternativewould be a pair of frame to frame side by side by side 6.5's above and below the HF flare. The more I think about it, this might be better.
"Quick note on parallel feed - I've done this for the input/driver stage on a power amplifier and was not happy with the result. For one thing, you must use a fairly large capacitor (2 uF or larger) otherwise there will be a large LF peak (several dB) as a result of the cap resonating with the (quite large) inductance of the transformer.
This large cap connected in the direct signal path (which also develops a significant voltage across it at LF thanks to its reactance) has a major impact on the overall sonics of the amplifier. I tried many caps (and combinations of caps) and was not able to find any that had less coloration than the entire amp/preamp combination. This particular circuit has the property of subjectively magnifying cap-coloration, an unexpected and novel result."
I can appreciate that you tried this Lynn and it did not work out for you. I never did a head to head comparison because I never had another set of identical amps to compare to - but also because parallel feed just made sense to me.
For a conventional feed SE amps where the output transformers must deal with both the DC and AC, the AC travels through the transformer from the tube plate and is then shunted to ground through the power supply capacitors. Thus, the parallel feed capacitor is usually better than the power supply caps, unless if you realize this like Mikey the Audiogeek has and make your last cap in the power supply a very high quality cap.
My next point with respect to my parallel feed amps is how I have adjusted them to use with my OB system. I need to roll off my OB Lowthers at 150 hz (eventually I will increase that to something like 250 hz) to protect the drivers, and to avoid the drivers becoming congested and to avoid doppler distortion in general. To prevent the Lowthers from receiving the frequencies below 150 hz, I sized my coupling capacitors and parallel feed capacitors to start rolling off the signal at 150 hz. This not only prevents the Lowthers from getting that lower frequency stuff, but it also prevents the amps from having to work too hard. The size of my capacitors is now something like 0.22 and 0.04 UF, easily within the realm of the best teflon capacitors like V-Caps, at a more reasonable price. I biamp with a sand amp for the OB bass.
The parallel feed arrangement allows for using the output transformers as autoformers, a more benign way to use power transformers that avoids some of the losses that are inherent in transformers - and losses, of course, means distortion.
Retsel
This large cap connected in the direct signal path (which also develops a significant voltage across it at LF thanks to its reactance) has a major impact on the overall sonics of the amplifier. I tried many caps (and combinations of caps) and was not able to find any that had less coloration than the entire amp/preamp combination. This particular circuit has the property of subjectively magnifying cap-coloration, an unexpected and novel result."
I can appreciate that you tried this Lynn and it did not work out for you. I never did a head to head comparison because I never had another set of identical amps to compare to - but also because parallel feed just made sense to me.
For a conventional feed SE amps where the output transformers must deal with both the DC and AC, the AC travels through the transformer from the tube plate and is then shunted to ground through the power supply capacitors. Thus, the parallel feed capacitor is usually better than the power supply caps, unless if you realize this like Mikey the Audiogeek has and make your last cap in the power supply a very high quality cap.
My next point with respect to my parallel feed amps is how I have adjusted them to use with my OB system. I need to roll off my OB Lowthers at 150 hz (eventually I will increase that to something like 250 hz) to protect the drivers, and to avoid the drivers becoming congested and to avoid doppler distortion in general. To prevent the Lowthers from receiving the frequencies below 150 hz, I sized my coupling capacitors and parallel feed capacitors to start rolling off the signal at 150 hz. This not only prevents the Lowthers from getting that lower frequency stuff, but it also prevents the amps from having to work too hard. The size of my capacitors is now something like 0.22 and 0.04 UF, easily within the realm of the best teflon capacitors like V-Caps, at a more reasonable price. I biamp with a sand amp for the OB bass.
The parallel feed arrangement allows for using the output transformers as autoformers, a more benign way to use power transformers that avoids some of the losses that are inherent in transformers - and losses, of course, means distortion.
Retsel
Retsel said:
Hmm - what puzzles me is how you avoided the large subsonic peak formed by the combination of a low-value cap and large transformer inductance. Using tubes with an Rp of 2500 ohms, I found that any cap below 1uF had a substantial (several dB) peak below the passband. Maybe the lower Rp of your output tube is acting as a damper for the LC of the parafeed cap and transformer?
The reason I mention this is the Lowther is famous for its limited Xmax - 1mm for the 8-ohm version, 3mm for the 16-ohm version. They would not respond well to subsonic peaks.
Continuing with the discussion mentioned above, something as simple as a vertical pair of 6ND410 (at the top) and a 6ND420 (at the bottom) in a baffle that is 14" high and 14" wide with an X-shaped "pinch" in the middle might work pretty well. The Edge model for this shape turned out pretty well.
Weirdly enough, this X-shape doesn't work very well for a diamond-shaped array of 4 midrange drivers, not something I would have expected. A plain square baffle (well, taller than it is wide) seemed to be the best overall for this driver array.
Re Parallel feed caps and LF resonance... There is a way to have small parafeed caps and no resonance. What you do is parallel the small cap with a much larger cap in series with a resistor. Someone (Voltsecond?) coined the phrase "lossy parafeed " for this and it apparently works without the sonic penalty I always associate with paralleling caps.
Stephen Robinson of Izzy Wizzy Audio tried this in his preamp when coupling to a tx volume control and said he couldn't hear the large cap (black gate)/ resistor wired in parallel
to the output coupling cap. This arrangement has now changed and he now uses a more conventional transformer coupled output however.
Rob.
Stephen Robinson of Izzy Wizzy Audio tried this in his preamp when coupling to a tx volume control and said he couldn't hear the large cap (black gate)/ resistor wired in parallel
to the output coupling cap. This arrangement has now changed and he now uses a more conventional transformer coupled output however.
Rob.
Baffle thickness
I will check to see if I have a peak in the bass response of my Lowthers beause of the small parallel feed capacitors; the Lowthers are the 16 ohm variety which have the 3 mm xmax.
I don't recall much discussion about what the thickness and materials of construction should be for open baffle speakers.
I initially made my baffles out of 3/4 inch plywood (which I have not changed because of a lack of time) and I suspect that this is way too thin to avoid the vibrations which occurs in the baffles. I added several braces, but still the baffle seems to vibrate too much. I have a super tweeter, my midrange and one set of 15 inch woofers mounted on the same baffle, and I have a set of 18 inch woofers mounted in a separate W-baffle (there is one 15 and one 18 wired in series per side). When I only had the 15 inch woofers (before I purchased the 18 inch drivers), the baffles were like refrigerator boxs outside on a windy day.
For his Quasars, Bert Doppenberg made his baffles 4 inches thick. I suspect that he discovered a long time ago that thin baffle material does not cut it (actually, I parially copied his original design, which seemed to be made of 3/4 inch material) for my current setup.
I also have seen hollow baffle panels (two 1/2 inch panels with perhaps a 1/4 inch space between) filled with sand.
It seems that mounting the drivers by their magnet (which also removes the distortion caused by the basket), instead of to the baffle, could avoid much of the need to "overbuild" the baffles, but then you would need a sturdy structure for mounting the magnets to.
Retsel
I will check to see if I have a peak in the bass response of my Lowthers beause of the small parallel feed capacitors; the Lowthers are the 16 ohm variety which have the 3 mm xmax.
I don't recall much discussion about what the thickness and materials of construction should be for open baffle speakers.
I initially made my baffles out of 3/4 inch plywood (which I have not changed because of a lack of time) and I suspect that this is way too thin to avoid the vibrations which occurs in the baffles. I added several braces, but still the baffle seems to vibrate too much. I have a super tweeter, my midrange and one set of 15 inch woofers mounted on the same baffle, and I have a set of 18 inch woofers mounted in a separate W-baffle (there is one 15 and one 18 wired in series per side). When I only had the 15 inch woofers (before I purchased the 18 inch drivers), the baffles were like refrigerator boxs outside on a windy day.
For his Quasars, Bert Doppenberg made his baffles 4 inches thick. I suspect that he discovered a long time ago that thin baffle material does not cut it (actually, I parially copied his original design, which seemed to be made of 3/4 inch material) for my current setup.
I also have seen hollow baffle panels (two 1/2 inch panels with perhaps a 1/4 inch space between) filled with sand.
It seems that mounting the drivers by their magnet (which also removes the distortion caused by the basket), instead of to the baffle, could avoid much of the need to "overbuild" the baffles, but then you would need a sturdy structure for mounting the magnets to.
Retsel
Robert F said:
DrP - damped resonance parafeed (part 3 and 4 w/ spreadsheet) by VS:
http://www.siteswithstyle.com/voltsecond/Parafeed_fun/Parafeed_fun.html
http://www.siteswithstyle.com/voltsecond/Damping_ringing_XFMRS/Damping_ringing_in_xfmrs.html
Digital vs. Analog sound
Gentlemen,
I hope this question isn't considered to be asked in the wrong thread.
I've searched everywhere I can think of & can't find a definition for "digital sound" or "analog sound".
I've been looking for a low (35wpc) SS amp & have convinced myself that class A is the way to go.
I found a class A (35wpc) & also a Sony, built-like-a-battleship, (35wpc) that's not class A. The shop owner described the difference in sound as digital for the class A as opposed to analog ("tube-ie") for the Sony.
Can anyone enlighten me as to the difference in sound?
I need a small amp for my highly efficient full range Wharfedale 12CS-AL's so I can use the existing 125wpc amp for the highly inefficient Wharfedale woofers.
dobias
Gentlemen,
I hope this question isn't considered to be asked in the wrong thread.
I've searched everywhere I can think of & can't find a definition for "digital sound" or "analog sound".
I've been looking for a low (35wpc) SS amp & have convinced myself that class A is the way to go.
I found a class A (35wpc) & also a Sony, built-like-a-battleship, (35wpc) that's not class A. The shop owner described the difference in sound as digital for the class A as opposed to analog ("tube-ie") for the Sony.
Can anyone enlighten me as to the difference in sound?
I need a small amp for my highly efficient full range Wharfedale 12CS-AL's so I can use the existing 125wpc amp for the highly inefficient Wharfedale woofers.
dobias