Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
174.68.76.189
In Reply to: RE: My First Maggies posted by wazoo on July 27, 2014 at 11:30:51
One of the things you pay for in 'better' amps (=$$$) is better matching of components if ANY is done at all.
And I'll disagree about the output being some 'average' function. When a transistor turns 'on', it has NO way of knowing if the other transistors in parallel with it are on OR off. I suspect if you use a O-scope, you could SEE the turn on of the transistors in the output.
It may also be possible that even though I am right, it may not matter due to the rise time or slew rate of the music being an order of magnitude or better BENEATH the variation in transistors.
Too much is never enough
Follow Ups:
My point is really that of Douglas Self who is the author of the Audio Power Amplifier Design Handbook, to name just one source. Doubling the output device complement nearly halves THD (a measured effect which is consistent with mathematical averaging) and greatly reduces crossover gain wobble, while also spreading the effect, thus generating lower order harmonics. Reducing the crossover distortion of an AB amp is a big deal. The only downside is in the control of parasitic oscillations.
I understand your theoretical point and think the escape from it which you posited is correct.
I too am generally a believer in the benefit of massively parallel devices. There are, however, problems in matching that many devices closely. The crossover distortion problem is much less of an issue with a high bias setting for the class A portion so the benefit of smearing the distortion is really not there in these designs.
The reason multiple devices are paralleled is not to improve distortion characteristics. It does, but that's a side/added benefit to the primary reason, yes? :)
Dave.
Let me try one last time than I'll drop it.
Sure, Distortion for multiple devices drops.
But let me try a thought experiment. Many people say that the way the driver moves FIRST during a drum whack makes a difference. I read about an experiment truncating the first rise of the waveform and it messed up the sound.
I'd submit that a bunch of output devices turning on at what I'll admit are slightly different times DOES make a difference.
I'd like someone to LIST the various distortions which can be measured in a modern amp. I'd suggest that most are measured steady state? no?
It is very difficult to discount the opinion held by many that the lower powered amp in a series where the main difference is 'scale', sounds somewhat better.
A worthwhile experiment would be to thoroughly characterize an amp with off-shelf output devices THAN the SAME amp with devices chosen and highly matched.
I wish Nelson Pass posted here and would insert a few brief comments about output device matching, how many parameters are 'matched' and possible sonic consequences.
Too much is never enough
As you previously mentioned, these devices slew quite a lot faster than any audio signal.In the first place, many people say lots of things which simply aren't true. A meme doesn't have to be true in order to spread. In the second place, we're talking about Class A/AB amps (and I can tell you from experience than an idling X150 or X350 makes a great space heater). In the third place, the Pass X150 utilizes 40 HexFETs in the output section and its bigger siblings only add 8 more (until we get to the X1000, which uses 80). Finally, every one of each output device type is drawn from the same lot and tightly matched nine ways to Sunday.
I think it's pretty easy to discount the opinion held by many that the lower powered amp in a series, where the difference is 'scale', sounds somewhat better. On the other hand, as 'sounds' is a subjective opinion, and a hell of a lot more than the actual sound factors into that opinion, it is patently impossible to say that those who find it to be true by actually doing the comparison are wrong. Of course, that raises the question: How many who make that claim have actually done the comparison? For the record, I have and I disagree (but, that doesn't make me right ;-).
Edits: 07/29/14 07/29/14 07/29/14
Not a THING you can tell me about HexFets. I MADE them for about 15 years as a process tech in the fab.
And I can assure you that 'lot number' is nearly meaningless. While 25 wafers fit in a standard cassette for MOST processing, a LOT consisted of 100 wafers. When processing, you can either process a SINGLE wafer at a time, which would be photo and dry etch processes, to IMPLANT which processed 25 wafers at a time. Certain 'reworks' while 'approved' and having ZERO impact on what is called 'yield', can and DO effect different wafers in such a way as to cause a variance across a lot. For this reason and others, MilSpec devices FORBID certain reworks.
Every once in a while, you'll find 25 out of 100 wafers of a 'lot' which 'zero' at probe! Big investigation which will center on ANY reworks or 'non conforming' notes attached TO the lot number in the inventory control system.
And I'd agree, MATCHING them is the only hope. Now, Pass sells for a premium, in part because of such attention to detail.
See how well output devices match in other, NON-cost-no-object designs. I doubt that anything more than an 'incoming' material inspection is done on those with MAYBE samples pulled at random and check on a curve tracer to see if they 'meet spec'.
Pick your meme and category? Cables? Room treatments? Crystals? 'A' vs 'AB'? Feedback? (gasp!) and so on.
The fact that transistors are typically quicker than the transient they are reproducing is necessary to their function. If they all don't turn on together? Well? Audible or not? Crazy? I think my idea has as much validity as those who can 'hear' silver wire or tell the difference between dielectrics. Many people say that the amp with fewer output devices sounds better. Who am I to argue? I have just Suggested a reason.
Too much is never enough
I think you might be underestimating how fast said devices can slew relative to your drum whack.....or the speakers trying to reproduce that drum whack. :) And, it's more than likely the devices don't need to turn on....they already are on. You don't see very many Class-B designs nowadays. :)
Regardless, dynamically induced distortions have been considered for many decades. As an example, you may remember the buzz around low-TIM amps back in the 70's after the Marshall Leach article in Audio buzzed up interest in this particular issue. Even earlier, the fellas at McIntosh and Marantz were considering dynamic issues designing some of their memorable valve amps.
Certainly, steady-state distortion specifications are more commonly quoted in sale sheets, but any good amplifier designer will certainly be considering dynamic performance in his designs.
Cheers,
Dave.
The primary reason is simply to get more power (divide the current demand), but it was discovered that doing so reduced distortion - a win, win (a rare scenario) with a few caveats.
Obviously.
The question is if you don't need the extra power do you want the extra devices?
Dave's question was rhetorical, but I answered it anyway. I'd take his point even further and say that, at the outset, not only was reducing distortion not 'the main' purpose, but it wasn't even 'a' purpose. However, once the benefit was demonstrated, it became a design option. Even the 'little' X150 has 40 MOSFETs in the output stage - the X350 (or X600) only adds 8 more.Your question is too open ended to answer.
---
My only reason for starting this sub-discussion is that I disagree with the statement that the lower powered members of a series sound better as a result of having fewer devices in the output stage. If done properly, more devices in parallel should always result in a measurable improvement. Whether or not such an improvement is audible is another matter. Is a reduction of an inaudible distortion somehow more inaudible? ;-)
Edits: 07/28/14
Maybe if I could pose the question better I would have an answer for it.
Nearly 30 years ago a DIY audiophile friend built two amps. One was a 32 transistor/ch amp for headphones but with speaker binding posts to drive his Shahanian like multidirectional monstrosity - running class A at 2 watts, the other was a single push pull pair of gigantic devices. He ran that at AB and played around with feedback, I don't know how much power it generated but it was obviously much more than the other amp. It had a much more vivid presentation and at some (low?) feedback setting there was noticeable distortion but the music just clicked and sounded real. The little 2W amp did clean detail and correct tone but lost something of the music's vividness even with AKG340 headphones.
The 2 device amp reminded me of my cousin's bin diving find of a KLH bookshelf speaker with a rotted surround that made the most communicative music I ever heard just off a 1970's SS Marantz receiver doing radio. It was obviously distorted but something there was "right".
The question is what was it that was "right" despite distracting distortion?
The other question was what is wrong with the beautiful sound of the little 2 watt headphone amp that it didn't click despite the lack of audible distortion.
My objection was to the statement that lower powered members of a series (sharing the same topology) sound better. Your example is quite a different situation and far more details would be necessary in order to proffer an answer, which I'm not likely qualified to do, in any case. :-)
Getting back to the Pass amps, the output devices of each type are drawn from the same lot# and then very, very tightly matched.
If I read correctly, you said HexFet in another post.
As a former IR Employee for over 15 years, I'll say, without fear of contradiction that lot# is nearly irrelevant, due to possible lot rework. Unless you somehow SPECIFY clean (ZERO rework at any step other than photo) lots, you are pretty much wasting your effort.
The assumption being that all wafers (100 in the case of IR) in a lot receive identical processing.
This ain't so, since some processes are SINGLE WAFER and microscopically different from wafer to wafer. Machines doing such processes are serviced at regular intervals (by wafer count, typically) and precess thru phases between such service. Even mass batch process, like diffusion, can have temp gradients across the length of the tube which effect outcomes at probe and some dopine steps.
Too much is never enough
I wouldn't presume to argue against any of your points. I'm certainly no authority on the semiconductor manufacturing process (I've spent my working life as a technician). Perhaps, part of the high cost of the Pass amps covers rejected output devices. Frankly, I have no idea, but Pass does match the devices for each amp to exceedingly close tolerances. Even if the lot code is essentially meaning less , the tight matching is still meaning ful .
I should also qualify a statement I made regarding the quantity of paralleled devices in the output section. The X150 uses different devices from its bigger siblings. The X250 has 32 output devices. So, compared to the X250, the delta is 16 for the X350 and X600 - a 50% increase instead of 20%. There is a theoretical advantage to that and measurements suggest that also translates to practical advantages.
I still maintain that comparing the slew rate of those devices to anything in the audio realm is like comparing a lightning strike to a sprinter and, as Dave pointed out, they're never really 'off'. I can honestly state that I regret selling my old X350 - it was a superb amplifier. I'm not saying that I'm at all unsatisfied with the valve amp which replaced it, but the difference in sound between the two isn't exactly overwhelming. I think that's high praise for the X350. While my Phi 300.1 was at VAC for repair and some upgrades (all at zero charge), I tried using my X150 and hardly listened to that system - same power rating (at 8 ohms), but it was anything but equal at the task of driving my 3.7s (which are spared the grunt work of sub 60Hz bass reproduction - well, that's the crossover point @18dB/oct). Even with my MMGs, I thought the X350 was significantly better than the X150 - but, that's probably not the best comparison (it would have been interesting to compare it to an X250).
I didn't intend for this to become a hot debate. I just find the position (which has been forwarded by numerous individuals) regarding the difference between the members of a series favoring the juniors hard to support (based on my understanding of the theory and my experience, albeit limited), at least for such amps as the Pass X-series. I admit that my personal experience could just as easily be affected by my belief as others who hold the opposite opinion. Perception is a brier patch.
This is an interesting hobby. Pick any issue in audiophilia and opinions will always be divided. Who is right? That's probably impossible to answer. I tend to gravitate to engineering theory, in which I do have some background (and I have a deep love for science). As long as the discussions don't get ugly, I enjoy a bit of debate. That's the best way to expose the foundations of opinions. In the end, we don't have to agree on this one - not that big of a deal. I just wanted to express my opinion on the matter, so I did.
Breir patch, indeed! No question about it. Some of these debates is like jumping Neckid into a patch of Poison Ivy. Or more to MY experience, picking figs off my tree without wearing long sleeve shirts. Man, what a RASH from that sap and whatever. Must immediately SHOWER.
Anyway, my point was simply to come up with a REAL or Realistic explanation for what some have noted, that being in a series of amps which are essentially scaled UP as you go 'upline', some actually prefer the lesser powered amp.
I HAVE to give those guys as much credence as the hobby (as a whole) seems to give those who note all sorts of differences, from Insulators (teflon=good / polystyrene=bad) or perhaps WIRE (Silver is bright sounding) or any of a dozen OTHER things of various degrees of scientific proof to outright wackiness. There are various levels of expanation ranging from 'reasonable theory' to 'let's have a s`eance!' Name your poison, stake your claim or whatever.
I floated an idea some time ago about biamping with radically different amps. My thought was that the time it takes a signal to go THRU the amp would have an audible effect. Sure enough, I got little support, but than I spotted a post from Davey which seemed to support the idea. Progress? !
cheers:
Too much is never enough
I don't know if I "supported" that concept or not, but I might have pointed out that indeed there are some amplifiers that do have built in group-delay (and other characteristics) for varying purposes. The Sunfire amps being the classic example...to facilitate the operation of the tracking power supply.
If you utilize disparate amps when bi-amping the world is not going to end, but when performing subjective evaluations it's important to understand any differences so as not to extrapolate incorrectly.
You have to really be careful when forming objective conclusions based on subjective testing. There's a couple of members here who do it incessantly. :)
Dave.
Dave
I am assuming you include me in the speculative extrapolation group. I will say that it is true that is my first tool to investigate anything. Then you delve into the physics in parallel to finding other people's observations and designer's pet theories. From there I go to picking modsters and DIY'er minds and go on to more listening experiments and figure things out or start a project, or post the speculation online for comments and to generate more ideas, or drop an issue as either too big to tackle or postpone giving it more attention.
The main limit is that I don't want to become an electronics engineer or audio engineer (or maybe I do?...). But I still want to be able to improve my own system's performance and advise friends on improving that of their systems.
"The main limit is that I don't want to become an electronics engineer or audio engineer (or maybe I do?...)."
You should. It allows to see through much of the self deception, intellectual dishonesty, and nonsense that pervades this industry.
If your first tool of investigation is speculative extrapolation then you're starting from a premise that can immediately lead you in the wrong direction....in a variety of different ways. :)
There's nothing wrong with subjective evaluations, but you can't make any objective conclusions from it. And any subjective conclusions you make from it are only valid for you and no one else.
Dave.
Unfortunately Davey, just about every industry is like this these days. Too many people hired to spin and BS their products rather than hired to make them better, combined with an uneducated public that typically does not spend any time really researching things before they buy. This industry is particularly difficult because it is technical and few understand any of the guts of these things at all (although my kids are learning about it at the elementary level at their STEM school).The A/V industry pales in comparison to the mattress industry. Talk about lying manipulating crooks.
I though this was about my first pair of Maggies....
Edits: 08/04/14
There is little alternative for an audiophile. There are aspects of sound quality we don't have effective measurement methods for so we can't rely on by the book design guidance principles and technical evaluation alone. We also need to be able to weigh the results of particular tradeoffs for our purposes.
There is no choice but to speculate on what technical aspect results in a paricular aspect of SQ being one way or another. Hence the constant need to change items in the chain.
Finally, there are simplifying assumptions behind the engineering equations that enable quantitative design that some designers try to take into account when designing with particular components in mind. Pass is one in particular who pays attention to these things in his white papers and we presume he does so in his designs, particularly for the First Watt designs.
Sorry:
It was Satie who made a related comment: QUOTE:
Davy (UK) had a listening based observation on coherence going with biamping vs single and found that for him there was something that sounded wrong with the use of different amps on bass and mids on 3.x models. For my own observation it is an issue, but careful matching of amps and XO will solve it and what little time issues remain can be adjusted for by ear with relative distances to the drivers (toe in - on the order of < <1").
The 'relative distances to drivers' remark speaks to me of time delay to fix the 'problem' of amp latency.
Too much is never enough
Yeah, that's not me. Sorry.
That guy is in the UK somewhere....I'm not.
Additionally, everything I read from Satie I take with a large grain of salt. :)
Dave.
You noted that the discrepencies between amps could often be adjusted for by panel toe which changes the distance from the listen to the various drivers…making up for 'lead' or 'lag' of a signal.
I'll see if I can find the entry again……I saw it maybe 45' ago.
Too much is never enough
Satie, these things are intriguing, are they not?It jives with an observation of mine. One of the most alluringly musical sounds may come from a basic alarm radio, or an old transistor radio. For all their relative SQ shortcomings, they have one or a few basic strengths.
The alarm radio has one speaker. Just that simple driver is capable of issuing a very cohesive -- time-aligned -- musical stream.
The transistor radio, in addition, may have the benefit of DC electronics, as opposed to AC. The same goes for an ancient cassette recorder that I used to have (in days when not even Walkmans were around). When traveling, I could enjoy Herbert von Karajan and the BPO to a surprising extent...out of a 3" speaker!
Edits: 07/28/14 07/28/14
My Tivoli #1 table radio sounds terrific. I lifted it up the other day and discovered it uses a ported enclosure! They now retail for 150$
Too much is never enough
An observation...
It is kind of interesting to read the opinions on how bi amping is the way to go on this forum and contrast it with the prevalent opinion on the HK/Limage discussion group on how bi-amping destroys the sound by supposedly sending two signals to a shared Mylar.
Just commenting on how there are, as with many things in audio, two camps on this. Luckily I am not qualified to comment either way. I have two set of Maggies.... Perhaps I will try both.
SC, please allow me to take advantage of your report: "...HK/Limage discussion group on how bi-amping destroys the sound by supposedly sending two signals to a shared Mylar."
I am so in love with the planar imaging experience that I worry when stuff like this is spread. Whoever spoke about the destruction of imaging at the hands of bi-amping is a comedian or is talking of bi-amping at speaker level (and really badly done at that). I have never heard imaging better than from properly matched LINE-LEVEL multi-amped planars. To be sure, we are talking of either Passive (PLLXO) or Active (ALLXO) setups...often generically lumped together as "going active". (wher a power amp channel directly controls each driver .)
With proper line-level bi-amping, the added powerful AND direct amplifier control of each segment of the Mylar, allows each driver segment to deliver more of its potential. And boy, each driver has a ton of untapped potential! Mid/bass drivers rock, punch & still deliver delicate textures with full-bodied presence. QR tweeters become Ribbon-like in performance very often.
Thus, IMHO it is PRECISELY a shared Mylar sheet which benefits the most from bi-amping. With proper care, bi-amping significantly reduces the disadvantage in mechanical intermodulation that a shared Mylar imposes on many planars. Importantly, in the process, bi-amping actually allows the advantages of "shared Mylar" to become much more dominant PLUS it prevails at much higher SPL than stock designs can allow. On a "frame reinforced" Maggie, this stuff adds up synergistically. As a result, one gets far more of the "cut from the same cloth" sound integrity and less distortion.
Some of it can be readily measured at home. On my PLLXO'd MMGs, harmonic distortion is lower. The last time I measured them with a normal xover was in late 2012. Among other surprises, the PLLXO bi-amp even bested the factory parts on "group delay behaviors" (but it was by sheer luck, not any design prowess on my part : - )) .
Of course, the ears had always said the very same thing. After many a full day of continuous listening enjoyment without fatigue -- actually wishing for more time -- it is fair to say that distortion is not much of an issue. Importantly, in my case, it is also being helped by the superb richness of imaging that prevails.
These old flatties will unroll a rich concert hall that overlays the room boundaries. Then they will place all the elements at discreet, specific locations along with delightful textures and proper 3D solidity. This is stuff that you can turn your head an look at directly, even when it "appears" outside the room. Your eyes actually refocus accordingly, as the walls vanish. The eyes scan the width, the depth, the FULL center fill and the elements within; many in their layered positionings.
There are things that your eyes cannot "see" even if you turn your head. These are "sound items" thay may be perceived as directly to your sides, above or behind you. THESE will stay in place only as long as your head is looking at anywhere BEHIND the plane of the speakers (even when that plane is virtually extended outside the room sides).
Anyway, this is just a pair of old modded MMGs, without a subwoofer. Therefore, your Maggies could turn into veritable monsters.
It does take some work, but it works. Since it appears that you are eager to do some work along these lines, a few of us will be very happy to help where we can.
Thanks for sharing, JBen. Like I said, I have two pairs and would love to try bi amping the older ones some day (currently they are sitting there sadly with no amp),
The person in question is indeed the guy known as Limage in Hong Kong. He has some non traditional ideas, to say the least, some of which are better received than others. Rumors are he really has some great sounding 3.5s.
Since getting the new 3.7i's I have been experimenting daily with set up and positioning, and no I do not have them dialed in yet. As described elsewhere (iow apologies for repeating myself).
I too can get holographic imaging, especially using a "Limage" style set up (out toward side walls, 40-60% out from FW, tweets in and little or no toe in. My issues are with getting the best sound with the fewest tradeoffs.
In my room,I have so far been unable to get the new speakers to have as much dynamic snap as is possible using Limage variations. Closer to the FW, I get all the "snap," but less depth and wider images (fat acoustic guitars?). It sounds great but is a bit claustrophobic compared to what is capable.
In addition, tweets in gives me sharper imaging and substantially more mid and upper bass and power range (with toe in it gives me muddy bass). Tweets out give me better time alignment, tighter sounding but lower amounts of bass, and something indescribably and coherently magical which tweets in misses.
I have kept careful notes on each location, with scores. I have several nines, but have yet to find a ten with the best of everything.
Though the IIIa's and 3.7i's sound extremely similar overall, the major differences are a substantially cleaner, smoother and less grainy midrange on the new guys, and substantially more low bass on the old guys (the new guys have apparently sacrificed low extension for bass quality -- a good decision overall, IMO). I continue to wonder how the tape on the back of the mids is affecting things, one guess is that it is reducing back wave energy and may have something to do with my relative "snap" issues when more than eight feet out from the FW.
Frankly, the HK/Limage setup sounded like lunacy to me...until I tried it. I'm still enamored with it. In fact, I tried undoing it over this weekend and went right back to it.
The idea that biamping could destroy the perceived benefits of that arrangement seems even more like lunacy, but nothing surprises me anymore (referring to the existence of illogical ideas). The idea makes about as much sense as the suggestion that a series crossover unifies the drivers, whereas parallel designs divide them - the function of a crossover is the same, no matter how it's implemented.
I read all the posts/comments on the HK forum and there were some other interesting orthodoxies over there. This includes:
1) Maggies should be tweets in
2) chop sticks are necessary to protect the frames from warping.
3) bi amping messes up the coherency as does any surgery on the cross over
4) later models of Maggies are not better than earlier models. It is primarily a marketing thing.
Again, I am making no comment on the soundness of these arguments. Just sharing...
".....discussion group on how bi-amping destroys the sound by supposedly sending two signals to a shared Mylar."
What do they think the stock, passive crossover does? :)
Sometimes you just have to laugh at the observations made on discussion groups. :)
Dave.
Davy (UK) had a listening based observation on coherence going with biamping vs single and found that for him there was something that sounded wrong with the use of different amps on bass and mids on 3.x models. For my own observation it is an issue, but careful matching of amps and XO will solve it and what little time issues remain can be adjusted for by ear with relative distances to the drivers (toe in - on the order of < <1").
Yeah, if you're going to use a disparate combination of amplifiers then that's another variable introduced and not related (directly) to bi-amping.
Yet again you veer off on a tangent from the main topic. You're uncanny in your ability to do this. :) Are you sure you haven't missed your vocation in politics? :)
Dave.
Yes, I tend to veer off, but the point discussed on the peking road planar group is something someone else raised and I just chimed in with my $0.0002 which I had on hand.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: