Sound Quality Vs. Measurements

[Wavebourn]

Sounds interesting, but how precise are sine to cosine waves in phase? What insures this precision?

However, when measurements are done on exponential scale so strict amplitude stabilization is not needed at all. When measuring on linear scale sine wave purity is not significant, so an ordinary functional generator is the way to go.
2 different inexpensive generators are the way that I prefer to go.

In 1070'th I used a huge HP signal generator, vacuum tube based. It had couple of RF oscillators, multiplier, and CRT with 1 KHz camertone fork - based oscillator to set frequency by Lissajous figures. Quite interesting toy, huh?

[1audio]

Originally Posted by Wavebourn

When measuring on linear scale sine wave purity is not significant, so an ordinary functional generator is the way to go.

Actually sine wave purity is very important since it causes significant errors in AC sine to DC comparisons. It also causes errors in different AC to DC conversion methods. Things that seem unimportant become really important at .01% accuracy.

[Wavebourn]

Originally Posted by 1audio

Actually sine wave purity is very important since it causes significant errors in AC sine to DC comparisons. It also causes errors in different AC to DC conversion methods. Things that seem unimportant become really important at .01% accuracy.

Actually to me it is important when measuring spectrum or frequency response on log scale. But when I observe on oscilloscope what happens in linear scale I don't care about signal purity so much. It is like to have good GPS in a car, good radio/mp3/whatever - player, and good CCTV, VS all in one toy that can do everything a little bit deciding for me if I want to search for nearby gas station, or I have to stop a first right on the freeway to do so.

What's the point in meeting contradictory requirements in the same tool when combining them results in worse and more expensive end result?

[bcarso]

Originally Posted by Wavebourn

Sounds interesting, but how precise are sine to cosine waves in phase? What insures this precision?

The matching of each integrator can be tweaked pretty effectively. Once thus matched, quadrature is assured. The input resistors can be adjusted to compensate for the difference in capacitor values. If the amplifiers are sufficiently high frequency the performance at audio frequencies can be very good. One of the advantages of the topology is the absence of common-mode swing of the integrator amplifiers. With one additional amplifier as an inverter it's possible to make all the amplifiers have a very small common-mode swing.

But usually one wants tunable frequency, so probably the best way is via code-programmable step attentuators (i.e., multiplying D-A converters) in place of the fixed integrator resistors. Then the requisite codes can be saved to keep things in quadrature.

[Wavebourn]

I like the idea of saved codes, when completely analog tracts are controlled digitally.

[1audio]

Actually as long as the resistors match to close tolerances and the caps match the amplitude stays constant and the phase is constant. The phase is constant even if the amplitude shifts and the sample pulse is triggered by the zero crossing. Its a very elegant solution.

A scope is rarely better than 3% if analog. Digital scopes can be better but precision long scale meters are .005% or better for AC and the small distortions are important for the calibration process. I'm looking for a reference but most are behind paywalls or buried in long articles. If I can find one I'll post it.

You can learn a lot about analog accuracy from repairing these types of instruments or studying the circuitry. The details of construction and design will show a lot of things learned from experience and analysis.

[dvv]

Originally Posted by Wavebourn

Placebo effect can be caused and directed by hypnotic inductions. You know that.

"And now while you sitting here comfortably relaxing with each slow calm breath listening to this music please tell me does it relax you or not"

True, no disputing that.

However, I still stand by what I said - if you significantly upgrade your power supply, no matter what you have except perhaps the best of the best, you WILL be able to hear a positive difference, although it may have a bigger or smaller impact overall.

I say this from persopnal experience, although I must add that the total number of devices I have worked over is not very large, say around 20 over the last 20 years. I don't do it for a living, usually I do it for friends only, hence the small(ish) number.

The word "significantly" is also open to discussion. As an example, I'd say that if I throw out 15,000 uF caps and put in 22,000 uF caps, which is around 50% more, that would be a significant upgrade. If in the process I also take out a say 300 VA transformer and put in a 500 VA transformer, that starts to look more like a rebuild than an upgrade, but I still call it an upgrade. To me, it starts to be a rebuild when I do all that, and then start swapping transistors, typically exchanging lower for higher power ones.

My point is, when someone tells me he has upgraded whatever, I cautiously ask exactly what has he done, just to be sure, to know what we're actually talking about.

[nigel pearson]

If you look at a spectrum analyzer at the ripple of a PSU it argues well for any improvements that can be made . Harmonics are strong way past 1 kHz . What astonishes me is that transistor people almost never fit chokes ( not even me ) . I haven't calculated it yet and perhaps the gains are not large . The thing to note is not only lower but nicer harmonics . I would say 2 x 10 000 uF with choke between might out gun 47 000 uF and be no more expensive , one can even do the valve trick and have a lower ripple rating on the second cap if you please . I assume 4 caps and 2 chokes . When I get a moment I will try it . As for chokes . I would try off the shelf toroid's ( a mains transformer ) . Funny thing I find is valve people obviously eat caviar for breakfast as the prices are outrageous . A choke can be made with a hand drill . If you have some M6 metal and ordinary metal you can play with the inductance . My rule of thumb is make the wire double what it would be in free air as a start point . I doubt if greater than 1 mm for most users ( 14 A free air ) . You can pot the chokes as it helps heat a bit . I found that be accident . I use Robnor 1113 g polyurethane from Farnell . Cheap and good . You will have to make you own safety checks . If the potting doesn't melt is isn't a problem usually ( i.e . hot ) .

Julian Vereker said to me his amps got better and better whilst he had done exactly nothing . He went on to say unrequested by him the capacitors just got better and better . His observation being the computer industry perhaps demanded it .

I got the PSU bug from Naim . Basally nothing except the transformer was different in most of their designs . Big sonic difference . Night and day I would say . That's where teaching goes out of the window . I have measured everything regarding that . Mostly what people say can be disproved in seconds when measuring . There is no simple explanation . I suspect the iron core in a class B amp is an energy store and that all there is to say . Better regulation and noise simple do not come into it . Noise is dreadful as with all transformers almost regardless . If not how would we use them for valve output stages ( you can use mains transformers , they work better than you would think ) ?

My friend got me to design a new PSU for his phono stage . The box had already been designed . I could have 80VA max with 1 watt load . I felt a bit cheated ! The design has 3 mm to spare . I am using the small terminal IEC inlet as it gets me 2 mm .

DVV , when I said we were poor I meant the nation . I have worked hard all my life ( from 13 on the farm , now 56 ) . I always had money .

My Wavebourn , any good links on hybrid cascodes ? ECC82 bottom ECC81 top ( all ECC81 ) ? Valve Wizard want's me to play guitar . I was wanting something above the pseudo pentode knee .

Any better than 5534 ? http://www.ti.com/lit/ds/symlink/opa627.pdf

[jacco vermeulen]

Originally Posted by nigel pearson

What astonishes me is that transistor people almost never fit chokes

Not these days, but how many years does the average audio DIYer curriculum span ?
In the early to late 1980s, CLC was rather popular in some areas.
Example : http://www.albs.de/Spec/Netzteile1_Info.pdf

[Wavebourn]

Originally Posted by nigel pearson

My Wavebourn , any good links on hybrid cascodes ? ECC82 bottom ECC81 top ( all ECC81 ) ? Valve Wizard want's me to play guitar . I was wanting something above the pseudo pentode knee .
]

What do you mean?

I don't remember any famous musicians that used hybrid cascodes. What kind of special effect would you expect from them?

A first guitarists found that overdriving amps made from datasheet examples they were getting sounds they liked. Then what they liked was shaped in specific guitar amp topologies. May be if datasheet examples contained cascodes they would go this way, I don't know. I played a lot with guitar effects, but all effect that guitarists liked were similar to what they already heard on records of famous musicians.