Peter Daniel said:
Another example would be the chip amps that you make and sell (Audio Sector?).
But for me, these products fall into the same category as a Wilson loudspeaker. Sounds pretty good, but measures not so good. I don't think they sound good because of their mediocre measurements, but despite them. In other words, they have "heroic" cabinets, high quality parts, and other attention to details that other companies overlook. So I think they would sound even better if they measured better.
And I wonder how good your chip amp or the Walker phono stage would sound with a well-designed discrete circuit. Clearly you are doing a lot of things right, as they are highly regarded products. Would they sound even better with discrete circuits? Only one way to find out...
I started with discrete circuits through this forum, built many amps and preamps, but after switching to op-amps I couldn't go back to those circuits, they were missing certain qualities that I only found through chips (I still have those amps sitting in different corners of my house doing nothing). It's only recently, after building F3 (with carefully selected parts), that I regain confidence in discrete circuits again.
I guess it's all matter of parts choices and implementation. One can build bad sounding descrete circuit as easy as bad sounding op-amp circuit. With a proper supply and part choices op-amps do not sound half as bad as some try to make them. And it's usually easier to tweak such circuits as there are much less parts to play with.
I guess it's all matter of parts choices and implementation. One can build bad sounding descrete circuit as easy as bad sounding op-amp circuit. With a proper supply and part choices op-amps do not sound half as bad as some try to make them. And it's usually easier to tweak such circuits as there are much less parts to play with.
Peter Daniel said:
That is actually a huge advantage. Tweaking makes at least as much difference as the circuit itself. And with a discrete circuit, the number of permutations can be daunting.
I've been designing discrete circuits commercially for 15 years. Each generation keeps getting better, but it's only been in the last few years that I've really been happy with their sonic performance. Persistence pays off. (By design, all of the older products can be updated to the current versions for a reasonable price.)
I think that Phase Accurate's point was well made and supports your experiences:
phase_accurate said:
GRollins said:
Sorry Grey for the late reply, I missed your post somehow.
That's a good question, but I don't have a good answer. Only a few data points. So in semi-random order:
1) One way to test this theory would be to wire the circuit with point-to-point methods so that there is no PCB to muck things up. The problem with this is that a PCB also provides mechanical support. It turns out that this is critically important to the sound. Here are two examples:
a) One time we were experimenting with a preamp. We doubled the amount of filter capacitance in the power supply. The extra capacitors were connected by about 8" of high-quality wire and set on the floor next to the chassis. I liked what I heard from the extra capacitance, but one of the other guys listening pointed out that some aspects of the sound went downhill. So we set the extra capacitors on top of wood blocks instead of the carpeted floor in the listening room. I was stunned to hear that putting the extra caps on wood blocks made a bigger difference to the sound than whether or not they were in the circuit.
b) Another time we were comparing brands of capacitors. We had ordered a dozen or so different brands and models and were swapping them in and out of the circuit to see which one sounded best. At first we left the leads at the full length (as supplied) and just let the capacitors sit up in the air off of the PCB. The idea was that if we ever wanted to use those caps in a different application that they would still be usable. We were most of the way through the test when we realized that it made a significant sonic difference whether the caps were suspended in the air or pulled tight against the PCB. So we had to re-do all of the tests using a more consistent methodology.
2) I used to believe that simpler was always better when it came to circuits. In the last several years I have come to find that this is not necessarily so, and my circuits have become considerably more complex than they used to be. They also sound much better. But at the same time, I have also been migrating to better sounding PCB materials and improved layout techniques that minimize the dielectric stress on the PCB. So maybe these things tend to cancel out.
The bottom line is that I can't answer your question directly, as I haven't done a controlled experiment where only one variable was changed. One good way to test your hypothesis would be to make one PCB with a very compact layout, and then another PCB with the identical circuit but a large layout with long traces. Then listen to the two and see what differences there are.
The bottom line for me is that I think you need to optimize all of the parameters you can. So in my case it would amount to making a fairly complex circuit, but spending a lot of time to optimize the layout, and finally, using the best PCB material available. How's that for a non-answer?
GRollins said:
We have all of our prototype PCB's made by Advanced Circuits in Colorado. They are very quick and very reasonably priced.
www.4pcb.com
They have two different ordering methods -- prototype and production. The reason they are so quick and cheap for proto boards is that they will combine several different jobs onto one panel. But that means that your choices for materials are quite limited.
On the other hand, for production boards they will use other materials. From their FAQ page:
Can you build RF applications?
Yes. We stock several RF materials such as Rogers 4000, Teflon, Duroid, and Polymide.
But they will also build just one PCB and call it "production". It's just that the costs will be higher than for their standardized prototype boards. Please see:
http://www.4pcb.com/index.php?load=content&page_id=80
Hope this helps,
Charlie
Charles Hansen said:
Well I know I have tried before and failed to measure differences between things I believed sounded different, but that doesn't mean the differences weren't there, just that I wasn't looking in the right place or with the proper tools / technique.
To me it is simple, assuming that I am not imagining the difference, if my ears hear a difference between one thing and another, then the sound field from the speakers must have changed. Ergo the signal driving the speakers must have changed, and that should be traceable to some change in the electrical properties of some piece of the circuit.
do you have any thoughts re the magnitude of the change in dielectric constant? In the work I am presently involved with, we have to worry about the change in dielectric constant caused by the variation in resin thickness due to the weave of the fibers in the pcb, as well as the surface roughness of the pcb traces since it is on the same order as the skin depth of the signal. (a good overview can be found here: download.intel.com/education/highered/signal/ELCT762/Class21_InterconnectI.ppt ) But this is all in the 100s of MHz and up, and I know my ears roll off well before that. So I am trying to come up with a theory that relates this back to what I can hear.
Grey,
you might want to try Teflon PCBs.
You can get "PTFE-laminat 0,8 mm/35 ì Cu " in small quant.
from http://www.elfa.se/en/
Sigurd
you might want to try Teflon PCBs.
You can get "PTFE-laminat 0,8 mm/35 ì Cu " in small quant.
from http://www.elfa.se/en/
Sigurd
GRollins said:
Charles,
I also use only Vishay's bulk metal foil SMD resistors as I find them superiour to any other SMD type resistor. I would not touch any other SMD resistor for critical signal path (or indirect signal path) duties.
When choosing a hole monted resistor it mostly is the nude Vishay S102 from Texas Components, the TX2352.
Surrounding circuitry must be made to sonically fit with these very revealing resistors, IMO.
Regarding packaging I cannot agree. There is a lot surrounding an OPamp: type of components, how to connect the OPamp for optimal sound quality, PCB layout, connectors, wires, and as you say below, the PS is very important.
Thus,
two designers can use the same OPamp in their designs, and we will get two very different sounding amplifiers.
I would really call it an egineering effort to create a CD-player's analog section that can be manufactured.
Sigurd
Electronics Engineer
I also use only Vishay's bulk metal foil SMD resistors as I find them superiour to any other SMD type resistor. I would not touch any other SMD resistor for critical signal path (or indirect signal path) duties.
When choosing a hole monted resistor it mostly is the nude Vishay S102 from Texas Components, the TX2352.
Surrounding circuitry must be made to sonically fit with these very revealing resistors, IMO.
Regarding packaging I cannot agree. There is a lot surrounding an OPamp: type of components, how to connect the OPamp for optimal sound quality, PCB layout, connectors, wires, and as you say below, the PS is very important.
Thus,
two designers can use the same OPamp in their designs, and we will get two very different sounding amplifiers.
I would really call it an egineering effort to create a CD-player's analog section that can be manufactured.
Sigurd
Electronics Engineer
Charles Hansen said:
Charles, et. al.,
Anyone know how the copper foil is bonded to the substrate? I've heard epoxy and heat (presumably leading to partial melting of the substrate, in turn leading to bonding when it cools), but do not know if either or both are true. It wouldn't surprise me if different methods were used depending on whether it was fiberglass, Teflon, etc.
When I etch boards I find a thin, white film remaining under the copper. It buffs away easily enough when I burnish the boards with steel wool, but it seems reasonable to ask what electrical properties it might have, either if left on the board between traces or the part that's still underneath the remaining copper.
Charles,
Yes, I'd thought of the two-boards-experiment. However, the looser board would introduce other variables: Signal trace length (think in terms of the sound of wire--in this case it just happens to be flat [Quick! Somebody call Nordost!]), stray capacitance, moving parts further away from each other (in turn leading to less electromagnetic interaction--a subtle effect, at best, but obviously we're talking about subtle things anyway), and the possibility that one layout or the other might be more prone to oscillation or RF interference. I could probably come up with ten other variables if I put my mind to it. In other words, this isn't going to be a simple question to answer because it will be difficult to isolate just one variable in proper scientific fashion.
Grey
EDIT: And while we're on the subject of boards...does anyone know whether someone makes boards with silver foil? That would be an interesting experiment to try. At least in principle you could have a proper comparison, with identical layouts and substrate--only the conductor need change.
Anyone know how the copper foil is bonded to the substrate? I've heard epoxy and heat (presumably leading to partial melting of the substrate, in turn leading to bonding when it cools), but do not know if either or both are true. It wouldn't surprise me if different methods were used depending on whether it was fiberglass, Teflon, etc.
When I etch boards I find a thin, white film remaining under the copper. It buffs away easily enough when I burnish the boards with steel wool, but it seems reasonable to ask what electrical properties it might have, either if left on the board between traces or the part that's still underneath the remaining copper.
Charles,
Yes, I'd thought of the two-boards-experiment. However, the looser board would introduce other variables: Signal trace length (think in terms of the sound of wire--in this case it just happens to be flat [Quick! Somebody call Nordost!]), stray capacitance, moving parts further away from each other (in turn leading to less electromagnetic interaction--a subtle effect, at best, but obviously we're talking about subtle things anyway), and the possibility that one layout or the other might be more prone to oscillation or RF interference. I could probably come up with ten other variables if I put my mind to it. In other words, this isn't going to be a simple question to answer because it will be difficult to isolate just one variable in proper scientific fashion.
Grey
EDIT: And while we're on the subject of boards...does anyone know whether someone makes boards with silver foil? That would be an interesting experiment to try. At least in principle you could have a proper comparison, with identical layouts and substrate--only the conductor need change.
It's relatively easy to build any circuit p2p using any type of wire. I did it on few occasions but the results were not really worth the effort. I also recall similar report from another forum member: http://www.diyaudio.com/forums/showthread.php?postid=423953#post423953
http://www.diyaudio.com/forums/attachment.php?s=&postid=33970&stamp=1022802435
http://www.diyaudio.com/forums/attachment.php?s=&postid=34127&stamp=1022869486
http://www.diyaudio.com/forums/attachment.php?s=&postid=34130&stamp=1022869824
http://www.diyaudio.com/forums/attachment.php?s=&postid=36080&stamp=1023927370
http://www.diyaudio.com/forums/attachment.php?s=&postid=264893&stamp=1068835307
If I build for myself, I still favour p2p, but the returns are rather diminishing: http://www.diyaudio.com/forums/attachment.php?s=&postid=1123226&stamp=1170435055
http://www.diyaudio.com/forums/attachment.php?s=&postid=33970&stamp=1022802435
http://www.diyaudio.com/forums/attachment.php?s=&postid=34127&stamp=1022869486
http://www.diyaudio.com/forums/attachment.php?s=&postid=34130&stamp=1022869824
http://www.diyaudio.com/forums/attachment.php?s=&postid=36080&stamp=1023927370
http://www.diyaudio.com/forums/attachment.php?s=&postid=264893&stamp=1068835307
If I build for myself, I still favour p2p, but the returns are rather diminishing: http://www.diyaudio.com/forums/attachment.php?s=&postid=1123226&stamp=1170435055
mightydub said:
One would think so. The problem is trying to figure out what to measure. The last time I made a semi-serious attempt at this was with video. I spent several days trying to make a measurement (any measurement!) that would correlate with the obvious differences I was seeing, but to no avail.
I've pretty much given up on these types of experiments. It is far more productive to spend the time just doing listening tests to find out what kind of advances can be made. Eventually I have enough results to create a structural framework in my head that reflects my experiences, but I have no confidence that I know the actual mechanisms involved.
Kind of like gravity. We know enough about its effects to send a probe to Pluto with unbelievable accuracy, but nobody knows how it really works.
mightydub said:
Well, you might be able to find something if you have some exceedingly sensitive test equipment. But it will be very tricky. The problem is going to be to replicate the actual conditions involved.
It will be easy to come up with negative results. But if you come up with any positive results, my guess is that the "objectivists" will dismiss them out of hand as being far too small to be perceptible by human beings. (Oddly enough, those same subjectivists are now saying that it is important to reduce THD+N to 0.00003%, even though there is no evidence whatsoever that distortion levels even a hundred times larger than this are audible under the "double-blind" conditions they consider de rigeur.)
But I think I would attack this problem sideways, and from several angles. Here are a few suggestions (assuming you have the proper test equipment):
1) Measure Dk at both 25 C and 50 C to see if there are any changes with respect to temperature. A positive result here would demonstrate that it is at least possible for Dk to be variable. If there is a change, then see if there is any persistence of the effect when the material is restored to room temperature.
2) Repeat the above tests with respect to dissipation factor.
3) Take a piece of double-sided PCB material and measure Dk as supplied. Then "break it in" by applying a 10 Vrms white noise signal across the two copper foils for a period of 30 days. Then immediately retest the Dk value. If there is a positive result, then retest the Dk after letting the PCB sit undisturbed for 1 hour, 2 hours, 4 hours, 8 hours, 24 hours, and one week to see if the effect dissipates over time.
4) Repeat the above tests with respect to dissipation factor.
5) Take a piece of PCB material and measure the Dk. Then send it off to be cryogenically treated. Re-measure the Dk when it comes back.
6) Repeat the above tests with respect to dissipation factor.
Should you be in a position to conduct any of the above tests, please report the results (positive or negative) to me, as I would be extremely curious as to what you find. I don't always keep up with this forum, so please e-mail me directly at my first initial concatenated with my last name at the four-letter name of my company dot com. Please be sure to include the word "ayre" somewhere in the subject header to make sure it gets past my spam filter.
Good luck!
Sigurd Ruschkow said:
Yes, it is true that someone had to do a board layout, even if the schematic was stolen from someone else. I'm still not sure that I would personally consider that the work of a designer.
But what about another real world example. What about a company that makes a power amplifier that consists of an off-the-shelf Vicor switching power supply and a B&O ICE module. Is this an example of engineering or packaging?
GRollins said:
I always assumed that an adhesive was used. But the Rogers Duroid material uses an electrodeposition process.
http://www.rogerscorporation.com/mwu/pdf/rt241.pdf
The first paragraph is the relevant one. But obviously there aren't many (any?) adhesives that bond well with Teflon. So maybe a different process is used with more normal substrates. I looked at some Isola data sheets, but they don't specify the process. But the Rogers RO-4000 series (non-Teflon) data sheet also specifies electroplating.
Hope this helps.
Charles Hansen said:
I think that even if you gave a "packager" the schematics of any of your amps, he/she will never get it to sound as good as you do! I saw the inside of an Ayre C5 a few weeks back, and there is little chance to make it sound as good as you do it with just the schematics. Or do you think otherwise?
A PCB engineer would actually call it design work to make the layout from a schematics
In my view, the power supply - both raw and regulated - are major contributors to the sound af a high-end amplifier. As are the component choices and box design (incl material choices to make). So even a "packager" has lots of design work to do. Would you not agree?
Your example of an ICE Power amp module is interesting. The first thing I think about is if it would be just packaging this module into a box, would it then be possible that all the finished amplifiers using this module sound exactly the same? They would probably be very similar but I am not sure if all of them would sound the same. That would actually surprise me.
I think that there are different levels of design work. Some people have the skill to design an amp from scratch, some do not have that skill so they use OPamps/modules.
To me they both do design work. Albeit in different ways.
Sigurd
Sticking to ptfe
just as in bonding metals to metals, an etching process is almost always used, be it wet chemical or CV based... with teflon/fep this involves mettallic sodium in a rather noxious mix..
http://www.actontech.com/fluor7.htm
John L.
jacco vermeulen said:
just as in bonding metals to metals, an etching process is almost always used, be it wet chemical or CV based... with teflon/fep this involves mettallic sodium in a rather noxious mix..
http://www.actontech.com/fluor7.htm
John L.
Charles Hansen said:
Hi Charles,
Thank you for explaining the background of your email. Nevertheless, it was not an interview. It was a communication initiated by you in which you, without provocation, attacked audio designers who use op amps as mere "packagers". I think that does a great disservice to many fine designers out there who are merely guilty of not sharing your design philosophy.
Charles, the exception does not make the rule. Yet you have based your arguments on that in citing the example of the Sony CD player. Another example of that kind of specious argument appeared in your letter to the editor.
You said, "For a good example of how extremely high levels of feedback can affect the sound quality of an audio circuit, we have to look no further than the review in Stereophile a few years ago of an ultra-low-distortion power amplifier. In that review, the reviewer noted that 'dry, dull, or aggressive recordings sounded much more so' through these amplifiers. The very idea that such an amplifier could make a mediocre recording sound worse than it actually is contradicts the idea that low distortion leads to sonic accuracy".
One can always find examples of bad practice with any approach, and poor amplifiers have been made with and without negative feedback. That proves nothing.
Charles, could you please tell us which amplifier that was, so that we can read the review and draw our own conclusions, in the context of the whole review, not just a snippet?
Finally, you argue that the National amplifiers merely achieve much lower distortion by applying hugely more negative feedback. This is a simplistic argument that misleads the reader. The new National op amp has a gain-bandwidth of about 50 MHz. This is quite high but it is only about ten times that of the old 356 op amp. Although I am only speculating as I don¡'t know the inner details (nor do you), it seems that this amount of implied additional negative feedback alone cannot account for the extraordinary improvement in performance. The National op amps use an advanced complementary bipolar process and probably numerous innovative circuit design approaches to obtain exceptional open loop linearity.
Of course, those nasty op amps must certainly suffer from the fact that their metallization is likely aluminum rather than Oxygen-free copper
.Cheers,
Bob
Bob Cordell said:
I guess it's the Halcro DM58 review by Paul Bolin, October, 2002
http://www.stereophile.com/amplificationreviews/683/index1.html
last paragraph.
- Status
- Not open for further replies.