|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.61.99.36
In Reply to: RE: Where is the magic in tubes? posted by Tadlo on October 24, 2014 at 05:07:27
try this: (I apologize that the tables are a litle screwed up by reformatting. I tried a reply to your e mail address; but it was returned.
Cable Properties and the 5% Rule
The reference for what follows is an article by Nelson Pass from a Feb. 1980 Speaker Builder article. You can find the whole six page article on Pass' website at
www.passlabs.com/pdf/spkrcabl.pdf
This is probably the best and most relevant article around.
In the article Pass measured 9 cables. Several sizes of zip, and several "high end" cables. What I will attempt to do here is summarize some of the key points, and point out some related issues.
The two key factors in the audio frequency range (under 100 kHz) is the resistance of the cable, and the inductance. For the cables that are side by side (zip cord format) the inductances ranged from 0.19 to 0.25 micro Henrys per foot, not much variation for 24 gauge to 12 gauge cable. Resistance measured for these cables follows very closely to standard tables of resistance for wire of the appropriate gauge. Note that the inductance is for the paired wires, but the resistance is the sum of both legs of the cable (out and back). Electrically, the resistance and the impedance due to the inductance are in series and add. But since there is a phase difference, they add as the square root of the sum of the squares. The vector sum. In other words 1 ohm of inductance and 1 ohm of resistance have a sum of 1.414 ohms not 2.
Resistance does not vary with frequency, inductance does. And the impedance due to the inductance rises linearly with the frequency.
The following table shows how the inductance has a significant effect at higher frequencies.
12 gauge Impedance for an inductance of 0.25 microH per ft.
-------------------------------------------------------------------------------------------
Inductive Impedance
Run length R (ohms) 1000 Hz 10,000 15,000
-------------------------------------------------------------------------------------------
10 ft. 0.033 0.016(.037) 0.157(.16) 0.236(.24)
20 ft 0.067 0.031(.074) 0.314(.32) 0.471(.48)
30 ft. 0.1 0.047(.11) 0.471(.48) 0.706(.72)
Resistance of the 12 gauge is based on 1.6 ohms per 1000 ft.
The figures not in parentheses are the inductive impedances in ohms.
The figures in parenthesis ( ) are the vector sums of resistance and inductive impedance.
It is generally believed, at least by Bose explicitly, and by B&W by inference; that the cable impedance plus the amplifier source impedance should not exceed 5% of the minimum impedance of the speaker. This is an approximation based on the impedance curve of a typical speaker. A more thorough and rigorous treatment is given in another paper of mine. This 5% figure keeps the ripple in the speaker's frequency response to +/- 0.25 dB. Generally considered inaudible. For a speaker like a B&W 602, the minimum is 4.3 ohms. 5% of that is 0.21 ohms. An amplifier like the NAD C350 has about 0.05 ohms of internal impedance at the output. So the cables can only have 0.16 ohms impedance before the effect on the speaker's sound is just barely audible.
If you look at the figures in parenthesis in the table above, we get there with only 10 feet of 12 gauge at 10 kHz. Not at all at 1000 Hz. In my original posting I referenced Dunlavy and some others. In Dunlavy's paper he says that in their tests, up to about 12 feet, there was no audible difference between 12 gauge and any audiophile cable they tried. From the table you can see why. With 10 or 12 feet of cable any effect is at or below the threshold of audibility up to 10 kHz. And above that frequency, your ability to make fine discriminations in sound is much, much less.
Now let's look at the exotic cables from Nelson Pass' paper:
Mogami cable had a resistance of 2.1 ohms per 1000 ft. Inductance was 0.023 micro Henry.
Mogami Impedance for an inductance of 0.023 micro H/ft.
----------------------------------------------------------------------------------------------
Inductive Impedance
Run length R (ohms) 1000 Hz 10,000 Hz 15,000 Hz
-----------------------------------------------------------------------------------------------
10 ft. 0.042 0.002(.042) 0.014(.044) 0.022(.048)
20 ft. 0.084 0.003(.084) 0.028(.088) 0.043(.096)
30 ft. 0.126 0.004(.126) 0.042(.132) 0.061(.144)
If we use the same speakers and amp as in the 12 gauge, you can see we don't exceed our just audible impedance of 0.16 ohms anywhere in the table, even though the resistance is higher.
Nelson Pass has plotted some nice curves, which I urge you to visit.
Now let me close this much too long treatise by adding a few caveats.
1) You may find you like one cable over another not because it's flatter in response, but because of the changes it makes in the sound.
2) The output impedance of the amplifier plays a part in this. So one cable may work better with one amp/speaker combo than it does with another.
3) The size and location of the speaker's impedance peaks will have a very significant effect on how the cable interacts.
4) Even long runs of 12 gauge have no inductive effect below 3-4 kHz. and a low inductance cable will have no inductive effects below about 10 kHz.
5) DC resistance is the dominant factor below 3-4 kHz for zip cord types, and below 10 kHz for low inductance types.
6) The 0.21 ohm criteria we set was for 8 ohm speakers with a minimum impedance of 4.3 ohms. Many speakers go well below that, and the fundamental principal is still 5% or less of the speaker's minimum impedance.
7) Many low inductance cables have high resistance and may still cause significant bass and midrange effects. You still have to worry about the DC resistance. Two of the exotics that Pass tested had the resistance of 18 gauge. Unacceptable!
8) All the exotic cables had very high capacitance. Some had 20-30 times as much as zip cord. That may not affect the speakers in the usual way, but it causes some amplifiers fits. (edgy or gritty sound) Pass' article covers this as well, and he should know about amps.
9) The speaker designer may have anticipated the effects of cable and amplifier impedances, and may have compensated the sound of the speaker to some extent.
10) If the speaker had a perfectly flat impedance curve, the DC resistance would have no effect on the sound. It would only make the speaker a tiny bit less loud. But the effect of cable inductance will still have an effect since the highs would be attenuated more than the mids and lows.
11) Transmission line effects and characteristic impedance are not a factor, unless the amplifier is upset by the very high frequency (1-10 MHz) effects. Pass' feels this should be a rare effect.
11) These are pretty subtle effects, and you won't hear big changes. Unless you're using 30 ft runs of cheap 24 gauge and switch to a good low resistance cable.
I do hope this sheds a little light on the subject. Nelson Pass' article is by far the best I've seen, he has measurements to back it up, and it stands the test of reasonableness based on known physical principals. The terrible thing is this was published in 1980, and the knowledge has still not become common.
Further, I feel cable manufacturers should be supplying data on the DC resistance, inductance and capacitance of their cables. It would help a lot. But I doubt they will as they'd rather sell mystique at high prices and very high margins.
I have written a longer second paper on the topic and included the effects of source impedance. In that paper, the 5% “rule of thumb” is not used, but a more rigorous treatment.
Interactions between speakers, amplifiers, and cables.
In this paper, V will be used instead of E for voltage.
The elements between the amplifier output are shown in a simplified diagram below. The elements are all in series and the load current flows through all of them.
Vo--------------- Za--Va--------------Zc--------------Vs--Zs-------------GND
---------> I
Vo is the voltage output of a theoretical amplifier with no internal losses. If the amplifier has a gain = G, then the output is Vo = Vi x G where Vi is the input voltage and G is the gain of the amplifier.
Va is the actual amplifier output voltage after accounting for internal losses.
Va = Vo - Za x I
Za is the internal impedance of the real amplifier at the output. It is usually expressed in terms of Damping Factor. DF = 8/Za for SS amps. For tube amps it is the tap impedance instead of just 8. Za is usually not strongly frequency dependent, but 50 to 100% changes are seen in solid state amplifiers with a series inductor in the output for stability. Amplifiers with capacitively coupled outputs can change over a very large range depending on whether negative feedback is taken from the output of the coupling cap or not.
Zc is the series impedance of the speaker cables. It is the vector sum of the DCR and the inductance as explained in the first paper.
Vs is the voltage at the speaker input terminals.
Zs is the speaker impedance. It is complex and varies widely with frequency in most speakers.
I is the current that flows through the circuit.
1) Vs = Vo - I (Za + Zc) This equation shows that the voltage arriving at the speaker is reduced by the losses in the amplifier and the cable.
2) I = Vo/(Za + Zc + Zs) the load current is simply the amplifier voltage divided by the total circuit impedance. V = IR or Ohm’s Law.
If we substitute equation 2) into equation 1) and manipulate the terms we get:
3) Vs = Vo (Zs/Za + Zc + Zs) this is simply the expression for a voltage divider.
Since Zs is the largest impedance in the circuit, it has the biggest influence on current flow in the circuit. In a typical 8 ohm speaker Zs will vary from a maximum of 25 ohms to a minimum of 5 ohms. (large Advent). So the current through an Advent will vary by almost 5:1 over the full frequency range.
Since there is a loss of voltage across Za & Zc as we show in equation 1), the voltage at the speaker input Vs will be less than Vo. But since the current flow varies a lot with frequency, so do the losses, and Vs varies with frequency.
To express this variation in dB as a measure of the effect of a particular amplifier and cable on a system, we need to identify the maximum Zs and the minimum Zs. We’ll call these Zs-max and Zs-min, and the voltages at the speaker as Vs-max and Vs-min.
4) Vs-max = Vo (Zs-max/Za + Zc + Zs-max)
5) Vs-min = Vo (Zs-min/Za + Zc + Zs-min)
If we now take the ratio of Vs-max/Vs-min and then convert to dB, we will get a number of dB by which that the amplifier and cable are changing the output of the speaker.
6) dB = 20 log [Zs-max/(Za + Zc + Zs-max)]/[Zs-min/(Za + Zc + Zs-min)]
note that Vo has dropped out since it occurs in both the numerator and denominator of the ratio, and it is constant with frequency.
We have assumed that Za and Zc do not change with frequency. If the frequency at which Zs-max and Zs-min occur are less than 3 kHz, this is a good assumption.
OK, now lets try this with our Advent speakers and 32 feet of 12 gauge cable.
Zc = 0.1 ohms
Zs-max = 25 ohms @ 43 Hz
Zs-min = 5 ohms @ 100 Hz
For my NAD C350 integrated amplifier DF = 150, so
Za = 8/150 = 0.053 ohms.
dB = 20 log (Vs-max/Vs-min) = 20 log (0.994/0.97) = 0.21 dB
This is the shift in the response of the Advent. It is added to the frequency response curve of the speaker. Since there is almost no loss at the frequency of Vs-max (1.0 would be zero loss, we’re at 0.994), we find a 3% loss at the minimum. Since 0.21 dB is inaudible we’re fine.
What is audible? 0.25 dB under very careful listening conditions is generally agreed to be the threshold of detection.
Now let’s try my Onkyo Receiver. DF = 80, so Za = 8/80 = 0.1 ohms.
dB = 20 log (0.992/0.962) = 0.27 dB I can’t hear this change.
Here’s a tube amp, the Dyna Stereo 70. DF on the 8 ohm tap is 12,
Za = 8/12 = 0.67 ohms.
dB = 20 log (0.97/0.867) = 0.98 dB
That I can hear. It shows up as a little more thump in the bass, and a little more forward midrange, because there’s an impedance peak at 700 Hz.
Another tube amp. This example is typical of the Manley Stingray and the 40 W Hoveland Sapphire. DF = 4, Za = 8/4 = 2 ohms.
dB = 20 log (0.923/0.704) = 2.35 dB
That’s plainly audible.
Last example. The Cary 300 integrated SET amp. DF = 2 Za = 4 ohms.
dB = 20 log (0.859/0.549) = 3.89 dB
A huge variation. Wonder why SET’s sound different?
If you doubt these numbers, the calculations for the Cary and Hoveland agree very well with the measurements made by Stereophile and available on their web site. If you look at the data for the tube amps, you’ll see that the amplifier impedance is much larger than the cable impedance, so even doubling the cable impedance would not make a significant change in the result.
For the Dyna ST-70 with 64 feet of 12 gauge,
dB = 20 log (0.966/0.851) = 1.1 dB a difference of 0.12 dB from 32 feet. Not significant
You can play with the numbers yourselves. Lower impedance speakers make the cable and amplifier more critical. For more background I would strongly suggest reading the paper on cables on Nelson Pass’ website: www.passlabs.com
Gerald W. Crum gwcrum@oh.rr.com
Follow Ups:
Wow that is pretty complicated, a little over my electonics level. But it looks like ultimately it is a matter of alteration of frequency response. Isn't that basically what Carver said? If you made a SS amp have the same frequency response as a tube amp it would sound like a tube amp. Are tubes and transformers just pricey tone controls? Why doesn't somebody make a tube-simulating un-equalizer? I have a feeling I am opening myself up to ridicule and charges of stupidity!
Exactly. The significant difference in the source impedance of tube Vs SS amps causes a voltage divider effect between the amp & cable and the speaker impedances that is frequency dependent and calculatable.
I have equalized a solid state amp to match the response calculated for a tube amp, and it did sound very much the same as the tube amp used with that speaker which was designed for tube amplification.. I have also done the reverse using a speaker designed for SS amps and EQ'd the tube amp - again the sound was pretty close.
The first published paper I know of on this "voltage divider effect" was in 1954 by the chief electrical engineer for E-V. It is also mentioned by John Atkinson of Stereophile, and he shows the net frequency response changes from amps using a speaker simulator. As expected, the response variations nicely match the theory. You can find this in Stereophile's archives for amplifier tests under Measurements.
Basically, the data (and theory) show more output where the speaker impedance is high, and less where the impedance is at a minimum. Effectively modifiying the frequency response of the speaker by the shape of the speaker's impedance curve. So, if the speaker impedance does not change with frequency, there is no effect. Unless, of course, the amp and cable impedance changes a lot with frequency.
In any event, it's all about frequency response, and you can EQ most of that. I say most; because there are other reasons besides the voltage divider effect why amps sound different.
Jerry
" If you made a SS amp have the same frequency response as a tube amp it would sound like a tube amp."
This is what Sony engineers supposedly did when they designed/invented their VFET SS amps (e.g., TA-5650) - the idea was to emulate tube amps. Per TVK: "... V-FET transistors are the best audio solid-state devices ever made." I'm currently using one and can verify it sounds very much like a tube amp. :^)
Well, yes and no. The V-FET has response characteristics similar to tubes, so less feedback is needed to get linear (low distortion) output; but the amp's source impedance will still look like an SS amp unless that is also compensated for. I don't know what Sony did overall in the design, so I can't really comment. However, I do know that several SS amps in the late 60's and early 70's were specifically designed to be tube like in their impedance characteristics. Most notably the MacIntosh amps with autotransformers on the output. It was also true of the Pioneer SX990 I once owned and the Marantz 2230 receiver. Because most speakers of that day were voiced for use with tube amps.
As an experiment, put a 0.5 to 1 ohm resistor in series with an SS amp and see how tube-like it sounds.
Jerry
...for some interesting reading.
Later Gator,
Dave
Hi Airtime, I'm not Professor Samra or Professor Bold Eagle but I just read this under inductors in How to Service Your Own Tube Amp: An interesting point about the use of an inductor in the power supply is that it effects the sound quality.The inductor in the power supply contributes in a small way to the (warm) sound of tube amplifiers. I don't know if your Sherwood uses a inductor or choke, my Dyna Mark 3 does, my Dyna ST-35 does not...maybe Mike and Bold got some more on it....Mark Korda
I think that view tends to misdirect ones attention from the real causes of the differences in sound between SS and tube amps. However, the effects are real.
In an amplifier, the power supply is in series with the load. The signal path only serves to modulate the current flow through the output devices going from the power supply to the speakers and back. So changes to the power supply certainly can affect the sound quality. I much preferred the sound of a Dyna Mk II over a Mk III, and at one point back converted a Mk III to a Mk II by taking out the choke and adding a series R to match the Mk II circuit. But I did change the rectifier in the real Mk II to the GZ34.
Jerry
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: