 Re: To measure or not to measure (and what good is it anyway?) [message #68582 is a reply to message #68137] |
Sat, 16 July 2011 21:04   |
 |
gofar99
Messages: 1947
Registered: May 2010
Location: Southern Arizona
|
Illuminati (5th Degree) |
|
|
Hi Everyone, I just saw this thread and thought I would add some thoughts to it. I see two separate concepts, one regarding testing and measurements and the other subjective quality of the end result. I feel both are needed. I have seen and heard numerous pieces of gear that measure well and sound like last Christmas's fruit cake. I have also seen and heard stuff that doesn't measure well and sounds fine. In a perfect world we ought to be able to get both. The issue of testing and measuring is quite valid to what I like to do - design valve audio equipment. I fortunately have a well equipped shop with now three digital storage scopes, two signal generators, a HP distortion analyzer and more meters than I can count. Still the process of testing a piece of equipment can be problematic. Often the results are ambiguous. Is the measured signal to noise that of the unit under test, the generator, the AC mains, stray EMI or what? How do you measure things like distortion, there are no hard and fast rules? I think in some ways I am particularly fortunate in that numerous diyers have built some of my projects (and there are many of the commercial kits out there as well) and I get lots of feedback on the good, and occasionally not so good aspects. This helps solidify the test and measurement process to validate the things that others want to hear. If for example I get a negative comment on distortion in a piece of gear when it is connected to some other piece of equipment - like a particular type of speaker on an amplifier, I can test for that problem and if valid come up with an improved design, or at least warn others not to do that. Another related issue is simulation of designs. This was mentioned in one of the early posts. I use some simulations in the early stages of design. I find they are at best only approximations of how valve gear will actually work. There are too many inexact parameters (tube variations between brands, and even ones of the same brand, component tolerances and such) to accurately do a final design on paper. This is where the art of design takes over. Also goes back to the measuring process.
Sorry for being so long winded  , but the issues here are fundamental to audio performance. I like this thread as it has potential applications to anyone in the diy community.
Good Listening
Bruce
|
|
|
|
| Re: To measure or not to measure (and what good is it anyway?) [message #68584 is a reply to message #68582] |
Sat, 16 July 2011 23:32 |
 |
Wayne Parham
Messages: 18783
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
It's a very important discussion, I think, because so many DIYers are interested in moving to the next level with their projects, going past kits into mods and even onto their own creations.
The biggest problems I see are in not having the right gear and not having the baseline to compare with, so making invalid interpretations of what they do see.
I think the modeling software available today makes hobbyists able to get really good results, but of course, measurements trump models any day. The problem, of course, is knowing what tools to use (right tool for the job) and knowing how to use them (and how to interpret the results).
I guess like everything else, it takes time. Study, study, then practice, practice, practice.
|
|
|
|
|
|
| Re: To measure or not to measure (and what good is it anyway?) [message #68608 is a reply to message #68595] |
Sun, 17 July 2011 23:00 |
|
Wayne Parham
Messages: 18783
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
I'm always happy to help when I can, or to just hang out and watch a guy post his progress on ART, sort of cheer him on. I really love this stuff. I'll admit my time is much less my own these days, so I am not nearly as active as I was, say five years ago. But I kind of have my loudspeaker line nailed down anyway, mostly working with mature designs.
Still, when I see a guy start trying to move from blind cut-and-try to modeling to measurements of physical models, it's gratifying. I always love to see their progression over the years, and I've made a lot of really good friends here along the way. This forum tends to draw good folks.
But I've seen some of the other forums get kind of crazy. It's pretty easy to get sideways in acoustic measurements, because there's a lot of ways to get them wrong. And with all the misinformation out there, a fellow can get going down the wrong road, even with all the best tools.
Amplifier measurements seem like they'd be a little more cut and dried, but I have seen a few amplifier measurement threads on various audio messageboards that I thought were sort of goofball-on-crack. One in particular was this guy obsessing about passing a "perfect square wave" through his SET amplifier. It had a slanted top and he just couldn't get it straight. 
The poor guy was going through all these circuit hacks and amplifier mods, and of course, lots of people were giving their "advice" on how to get that perfect square wave to go through. Then finally, one small voice appeared, with a short post that went completely overlooked, saying "it's the natural band-pass of the coupling components, particularly the output transformer." He said the slant was actually normal and expected. But was this good comment taken? No. The thread went on in its meandering insanity. 
|
|
|
|
|
|
| Re: To measure or not to measure (and what good is it anyway?) [message #68624 is a reply to message #68622] |
Mon, 18 July 2011 14:16 |
|
Wayne Parham
Messages: 18783
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
It's interesting, isn't it? I depend on measurements more than I trust my ears, but yeah, I think it's important to make the right interpretation of what you see. A good example is that square wave thing. There's no need for bandwidth outside the audible range, so the 10x below and 10x above factor is important to know when looking at the output signals. If it passes a good square wave from 200Hz to 2kHz, it's fine, and even better if it does like yours do - from 100Hz to 4kHz. But to expect an amplifier with coupling components to pass a square wave above or below that is unrealistic. Unimportant too. Like you said, I've heard many terrible sounding amps that passed DC to way up, but who cares? That's not the only metric, or even the most important one. Use a sweep to get the response curve, don't look at square waves. That's just what comes in vogue every now and then on messageboards, silly stuff that gets meaningless chatter.
|
|
|
|
Posting Rules
Members
Search
Help
Register
Login
Get a T-Shirt!
. Not so with any that have coupling caps or transformers. A good rule of thumb that someone passed along to me is that in order pass a clean square wave the response must be linear at both 10 time above and 10 times below the test frequency. It seems to hold true. Some SS gear ( I have one amp like this) can pass DC to 400K. It produces nice square waves at any audio frequency you choose... it just sounds dull and uninteresting. Everything I cook up with the exception of the recent phono preamp can do nice squares from about 100HZ to 4K. Below that you will find some top slanting and above that some slight rounding or an occasional overshoot. I design for -3db at 5HZ and 30K so this is about what you would expect to see. Gear like phono preamps (not IC ones though) will have rather wild looking squares (more like triangles) because of the method of equalization. I have considered what would happen if you used a linear phase IC to do the equalization in a tube based phono preamp. Not a purist thing for sure, but might be nice. 
