• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

ASR dummy load configuration

From the left hand column of page 20 of the Hofer article you posted the formula for the Resistor Voltage Coefficient. You neglected to mention or quote, “In most common electronics, the VCr is safely assumed to be negligible”. Still keep your eye on it for ultra-low THD + N applications.
The heck does this mean? This is what he says:

1632121014393.png


In this thread and application we do care and so does he. So why on earth would you quote the "we don't care" bit?
 
On the right hand column of the same page Hofer explains the Thermal or Power Coefficient. This formula is provided. Note that (V squared/Ro) or Watts is included. I did not make it up.

R(Vs) = Ro x [1 + TCr x Z (w) x (V squared/Ro)] *
The last term is Vsource squared. So it absolutely matters. Yet you claimed I was mistaken to worry about "voltage" and its relation to distortion? What could be more important factor than it being squared?

If you are having trouble understanding what I write, ask. Don't take a misunderstanding and turn it into an insult.
 
So what I did was use resistors of higher value (1k Ohm)

You really should specify your measurement parameters as the results are presented. I thought you were using (4) paralleled 16Ω 50W resistors linked to previously.

Regardless, Hofer's paper was aimed at low power applications involved with the design of AP measurement gear, while this thread aims at high power amplifier dummy load resistor applications. That doesn't change the physics presented, but it does change probable cause behind the tell-tale 3rd harmonic distortion observed to strongly favor VCR. This is simply because we are dealing with much larger (thermal inertia) resistors while simultaneously testing at relatively very low power levels. This makes it extremely unlikely that any kind of thermal nonlinearity is involved, leaving voltage as the only guy standing when the music stops.

The simple VCR definition from the earlier cited Barth paper states that "The voltage coefficient of resistance may be expressed as the ratio of the resistance change in ohms to the corresponding increase in applied voltage in volts when the temperature is held constant." Their focus involved instantaneous and negative changes in resistance during a pulse as short as 1ns in which they concluded there wasn't enough time for thermal change, thus laying the blame entirely at the feet of VCR.

Interestingly, in some cases they saw instantaneous drops (and recoveries) in resistance by as much as half the resistor's normal value with no damage to show for it afterwards. This intuitively supports Hofer, et al. calculations of the 3rd harmonic tell-tale as such a drop should equally affect the peak and trough of the voltage waveform. It also helped me understand Hofer's requirement that there be no "no significant DC component" involved as this would bias the distortion in the direction of the DC offset, thus affecting the waveform unevenly, thus generate even order harmonics like tube amps tend toward.

God bless you and your precious family - Langston
 
Regardless, Hofer's paper was aimed at low power applications involved with the design of AP measurement gear, while this thread aims at high power amplifier dummy load resistor applications. That doesn't change the physics presented, but it does change probable cause behind the tell-tale 3rd harmonic distortion observed to strongly favor VCR. This is simply because we are dealing with much larger (thermal inertia) resistors while simultaneously testing at relatively very low power levels. This makes it extremely unlikely that any kind of thermal nonlinearity is involved, leaving voltage as the only guy standing when the music stops.

The simple VCR definition from the earlier cited Barth paper states that "The voltage coefficient of resistance may be expressed as the ratio of the resistance change in ohms to the corresponding increase in applied voltage in volts when the temperature is held constant." Their focus involved instantaneous and negative changes in resistance during a pulse as short as 1ns in which they concluded there wasn't enough time for thermal change, thus laying the blame entirely at the feet of VCR.


Barth paper was specifically referencing carbon composition resistors of which wire-wound power resistors obviously are not. A more detailed study looking at thick-film resistors shows a variety of mechanisms and that dependent on the voltage range the R(v) is effectively linear so any model used as "standard" is likely only correct for a particular resistor type, and the coefficients must be known with some accuracy or the impact of V is effectively unknown and a guess at best. The conductive pathways could even be near step-function of applied voltage.

https://www.mouser.com/pdfdocs/Welwyn-VCR-Characteristics.pdf

Within the limits of the test equipment though, it should be easy to tell with some certainly whether the effects are thermal or applied voltage.
  • Thermal effects will be frequency dependent. Decrease the frequency and the distortion due to thermal effects will go up.
  • Applied voltage effects, at this low frequencies, will be relatively immune to frequency effects.

If you are testing for thermal effects, due a sweep from low to high frequency, and high to low frequency to negate or at least illustrate any effects of self heating. Inductance and capacitance can come into play, so I hope you are using non-inductive resistors.

Their focus involved instantaneous and negative changes in resistance during a pulse as short as 1ns in which they concluded there wasn't enough time for thermal change, thus laying the blame entirely at the feet of VCR.

In one of our tests, the resistance of a 2
watt carbon composition resistor was observed to decrease from 390
ohms to 200 ohms during pulse testing. The high voltage resistance
was approximately 51% of the resistance at low voltage during
application of a 2kv, 100ns wide pulse. In pulse tests at 3kV with the
same value resistors, a 1 watt resistor decreased approximately 15%,
and a 1/2 watt resistor decreased approximately 6%.*

Keep in mind this is far beyond the voltage rating of these parts. We laugh at subjective audiophiles when they come up with terms for wire like "micro-arcing". At these voltages, arcing is a very real concern. You would expect more for the 1W than the 1/2 watt, but that could imply there are other effects here that the authors did not properly account for. It is not like there was a lot of peer review and replicated experiments.

From Rhopoint,
Voltage coefficient, measured in ppm/V, is negative for most materials. Voltage coefficient is generally measured between 10% and full rated voltage.

Voltage coefficient (VC) is usually experienced with film resistors as the electrostatic field generated by the voltage tends to align polarised molecules in the resistance film, much in the same way as a capacitor dielectric.

Wirewound resistors or metal foil resistors tend not to suffer in the same way as their resistance elements are in a much stronger chemical structure. However these technologies tend not to go beyond 1M, and most high voltage applications need to go well beyond this resistance range.

Voltage coefficient, measured in ppm/V, is negative for most materials. Voltage coefficient is generally measured between 10% and full rated voltage.

VCs for high voltage resistors vary from immeasurable to over – 10ppm/V for some film resistors. A VC of -10ppm/V is equivalent to 0.001%/V or –1%/kV. This is unsatisfactory for precision applications where total error is maintained below 1%.

Given the statement that Wirewound resistors tend not to suffer, and that we are dealing with rather small voltages and very large structures, i.e. Barth referenced data of Kilovolts over 5-10mm of carbon composition, versus say 10-40V over what would be meters in wire-wound resistor for comparison coupled with their construction advantages .... then it is quite the stretch to state unequivocally that the issue is voltage dependence of resistance when using wire-wound resistors.

That does not mean that thermal is the reason either, though it should be possible to characterize inductance and capacitance well enough at lower frequencies to put these into a model and look at the frequency dependent change on distortion and see if it matches what the thermal model predicts.

It could very likely be neither of these.
 
Barth paper was specifically referencing carbon composition resistors of which wire-wound power resistors obviously are not.

Incorrect. The Barth paper was not specifically referencing carbon resistors and it did discuss wire-wound types. And cermet, metal film, physical attributes and construction methods. Unlike yourself, they performed real experiments and developed test procedures to detect VCR for their applications. Then they wrote about their observations. That's a better model than just talking and minimizing the contributions of others.

The Welwyn link was quite helpful to me, thank you.

Edit: I apologize for feeding the troll referenced in this post, I mistakenly thought we were mutually interested in pursuing truth. I don't have time for both thoughtful forum engagement and dealing with anonymous egos that sport in disparagement. In the future I won't respond to this type of bait.
 
Last edited:
Incorrect. The Barth paper was not specifically referencing carbon resistors and it did discuss wire-wound types. And cermet, metal film, physical attributes and construction methods. Unlike yourself, they performed real experiments and developed test procedures to detect VCR for their applications. Then they wrote about their observations. That's a better model than just talking and minimizing the contributions of others.

The Welwyn link was quite helpful to me, thank you.

Yes, let's see what the "Barth" paper says ....

Bulk metal resistors have almost unmeasurable voltage
coefficients. However, due to the low resistivity of metals, wire
wound resistors must be used to achieve reasonable resistance
values. The combined inductance and capacitance effects of wire
wound resistors prevent their use either at high frequencies or with
fast pulses.

In case you don't understand that quote, wire wound resistors are bulk metal resistors and their voltage coefficient is almost unmeasurable. However, since their expertise is high speed pulse measurement, they can't use wire wound resistors due to inductive and capacitive effects.

I provided you quotes from actual resistor manufacturers. I am thinking they know their products pretty well. Heck, even Barth agrees with them.

The Barth paper measurements are pretty much exclusively Carbon.

No one is minimizing the Barth work, though again, it is not peer reviewed and I don't see any replication of the results. However, the Welwyn paper (actual resistor manufacturer) goes into detail on what impacts the coefficient, most of what they discuss is not at all applicable to wire wound resistors, and Rhopoint got their start in high performance resistor supply, though has much expanded and changed their model. I.e. they know what they are talking about. That makes two expert opinions, Barth and Rhopoint, that say coefficient of voltage for wire-wound is not an issue, and Welwyn in discussion of what the issue is, gives evidence that wire wound again, does not have an issue with resistance changing with voltage.

Soooo, I will go back to my conclusion, based on 3 expert opinions essentially (and I could dig up more, but why bother), and that the dV/length is exceeding small, that the coefficient of resistance w.r.t. voltage is exceedingly small for wire wound resistors used for audio testing and would not contribute to an increase in distortion ... and I am not aware of any experiments presented as yet that come close to properly isolating this variable to prove it is an issue. That is simply the conclusion presented.
 
Hello All,

This Afternoon I am running some impedance sweeps; 20 hZ to 80khZ at 1 Volt by 3 seconds.

First the 4R, 300 Watts, single wire wound, weighing in at 1.25 ish pounds.

Thanks DT

Edit Note:
This is the standard model resistor, not the low inductance "N" version. Still not too bad for a bench tool.

4R 300 Watt Impedance Imainary.PNG
4R 300 Watt Impedance Magnatude.png
4R 300 Watt Impedance Phase.PNG
4R 300 Watt Impedance Real.PNG
 
Hello All,

These impedance plots are for the 4, 16 Ohm resistors in parallel. These resistors are not the low inductance "N" version. The scales of the plots are the same as the previous plots for the 300 Watt 4R resistor. The inductance for the 4 parallel resistors is lower than for the 300 Watt resistor.

Note: inductance in parallel adds like resistance in parallel. The 4 16 Ohm resistors in parallel have a total resistance of 4 Ohms The Total inductance is 1/4 the inductance of each resistor alone.

in terms of inductance the 4 resistors in parallel is a better bench tool than the 300 Watt resistor.

Thanks DT
4R 4 50 Watt resistors  Impedance Imainary.PNG
4R 4, 50 Watt Impedance Magnitude.PNG
4R 4 50 Watt resistors Watt Impedance Phase.PNG
4R 4 50 Watt resistors  Impedance Real.PNG
 
Nice work. Your paralleled resistor impedance measurements are as good as I've ever seen. : )

Consider (for example) loudspeaker impedance plots with 5Ω or so between each horizontal line. If you used that kind of vertical resolution, your traces would be flat to Buzz Lightyear.

My boat anchors appear to be a bit better than your boat anchor, possibly due to mine being the non-inductive types, but your paralleled setup takes the transfer function cake. Any of these responses are far better than the amp load application requires.

I measured the impedance of my (4) resistor jig pictured a couple of times earlier in this thread. 2Ω, 4Ω, and 8Ω complex (meaning magnitude and phase) traces are shown on the following plot with 0.5Ω vertical resolution on a 10Ω overall scale. The spectrum of interest (DC - 20kHz) is lovely. The greater the load resistor value, the flatter the traces due to the higher terminating resistance swamping out the fixed source inductance in the cabling between the voltage sense of the 4-Wire (Kelvin) impedance* jig and amp load jig.

RLoad v3 1ft 2-4-8Ω.png


The above plot used about 1ft of 14AWG (round trip is 2ft of course) between my Kelvin type impedance jig and the amp load jig. I could show flatter results by connecting the resistor(s) directly to the impedance jig with shorter cabling, but again, that's unnecessary for amp load applications.

My normal setup (for convenience) will use 5ft of 14AWG cable between the amp under test and amp load jig. I did a comparison between the 1ft and 5ft cabling at 4Ω using a 2Ω overall scale to highlight differences.

RLoad v3 4Ω 1ft vs 5ft.png


God bless you and your precious family - Langston

* Ideally you're supposed to take the voltage sense directly at the DUT (amp load jig in this case), but my impedance jig cheats for sake of convenience and something else that's OT in this thread. My impedance jig takes the voltage sense at the point where the cabling to the DUT is connected, thus that cabling's impedance is included in the measurement, thus that cabling needs to remain short enough that the results aren't skewed beyond the tolerance required of the measurement. My guess is that DT used a proper Kelvin connection with the voltage sense directly attached to the resistor(s), partially accounting for his flatter traces.
 
Last edited:
You really should specify your measurement parameters as the results are presented. I thought you were using (4) paralleled 16Ω 50W resistors linked to previously.

Regardless, Hofer's paper was aimed at low power applications involved with the design of AP measurement gear, while this thread aims at high power amplifier dummy load resistor applications. That doesn't change the physics presented, but it does change probable cause behind the tell-tale 3rd harmonic distortion observed to strongly favor VCR. This is simply because we are dealing with much larger (thermal inertia) resistors while simultaneously testing at relatively very low power levels. This makes it extremely unlikely that any kind of thermal nonlinearity is involved, leaving voltage as the only guy standing when the music stops.

The simple VCR definition from the earlier cited Barth paper states that "The voltage coefficient of resistance may be expressed as the ratio of the resistance change in ohms to the corresponding increase in applied voltage in volts when the temperature is held constant." Their focus involved instantaneous and negative changes in resistance during a pulse as short as 1ns in which they concluded there wasn't enough time for thermal change, thus laying the blame entirely at the feet of VCR.

Interestingly, in some cases they saw instantaneous drops (and recoveries) in resistance by as much as half the resistor's normal value with no damage to show for it afterwards. This intuitively supports Hofer, et al. calculations of the 3rd harmonic tell-tale as such a drop should equally affect the peak and trough of the voltage waveform. It also helped me understand Hofer's requirement that there be no "no significant DC component" involved as this would bias the distortion in the direction of the DC offset, thus affecting the waveform unevenly, thus generate even order harmonics like tube amps tend toward.

God bless you and your precious family - Langston
Hello All,

To assist the understanding. This is the Hofer posted formula for calculating the 3 Harmonic Distortion for the Thermal Coefficient of Resistance model of distortion:

3HD = ( TCr x θR x Vs^2/Ro ) / 4

Vs^2/Ro is Watts

θR is thermal resistance

TCr is Thermal Coefficient of Resistance

Recall that you are retired and that you looked up thermal conductivity; where heat transfers easily thermal resistance is low. The Thermal conductivity of Aluminum is 239

For the formula above, θR is thermal resistance. The thermal resistance, θR, is = 1/239 or equal to 0.00418. (a small number)

That is why your boat anchor has a low 3rd HD.



Thanks DT
 
That is why your boat anchor has a low 3rd HD.

I'm pretty sure it's not that simple at audio frequencies, though it certainly makes sense near DC where things move verrrry sloooowly. That 3rd HD equation is found in the section titled "Thermal Modulation at Very Low Frequencies", which it states is "typically less than 0.2Hz".

Still, your apparent fixation* on heat has made me think about how the heck the magic VCR quantity (which I'm convinced is the real culprit) actually works from a physics point of view. I've read speculation that VCR at its root is nothing more than tiny thermal spikes within the resistor between unlike materials, bonding agents, etchings, composition particles, etc. And that these spikes occur at the intersections of these tiny junctions regardless of the size of the boat anchor. Some have done experiments that seem to correlate with this hypothesis.

Then again, these empirical confirmations remind me of the famous "study" published during the roaring 20's that found an inverse correlation between women's skirt lengths and stock prices. Shorter skirts, higher prices! : )

The great reveal of this thread is that resistors of various types exhibit various short and long-term non-ohmic behavior. We may not fully understand causation, but we sure as heck can measure it and avoid its effects to the extent required for a given application.

Just like Louis Armstrong said, What a Wonderful World! : )

God bless you and your precious family - Langston

* late Middle English (originally as an alchemical term denoting the process of reducing a volatile spirit or essence to a permanent bodily form). Completely irrelevant, but a fun word and no deprecatory invection intended. : )

BTW - where was the voltage sense derived in your impedance measurements? Did you take it directly at the resistors, or was it taken inside the APx1701, or does the APx555 assume the voltage level due to the known (10x) gain of the amp?
 
Last edited:
Hello All,

We have various types of resistors:

Thin metal film

Thick film

Metal oxide

Metal foil

Carbon Compound

And more

I suppose that each type could have a distortion profile of its’ own.

One profile is VCR which assumes constant temperature and no dependence on frequency.

Another profile is TCR that does have a relationship to temperature and frequency dependence as well.

In terms of combination profiles tests show CC resistors have a combination of both VCR and TCR resistor nonlinearity.

In our power amplifier test resistors we use wound wire resistors which have minimally measurable VCR, their primary mechanism of distortion is TCR.

Today for a reality check I ran some plots with the AHB2 amplifier and the 300 Watt wire wound Boat Anchor test resistor.

To warm things up a little I turned the test voltage to input about 12 Watts into the Boat anchor. Then I made a few test frequency steps starting at 250Hz and stopping at 12.0 kHz.

There was a 17dB increase in 3rd HD between a test frequency of 250Hz and 12.0 kHZ. This is not the result of VCR resistor nonlinearity alone. I was surprised; I expected decreasing distortion with increasing frequency

Before we post any more amplifier reviews we should check our assumptions.

Thanks DT

For testing impedance I use the APx500 speaker upgrade software. There is a current sense resistor inside the APx1701. I connect a XLR cable between the APx1701 Current Sense output and the APx555 input 2. I connect the APx555 Input 1 with a XLR cable and small alligator clips directly to the Boat Anchor test resistor as you would do testing TS/P’s of a speaker. This is per the illustrations in the APx500 software manual.
Screenshot (24).png
Boat Anchor 12 Watts 250 Hz.png
Boat Anchor 12 Watts 1.00 kHz.PNG
Boat Anchor 12 Watts 4.00 kHz.PNG
Boat Anchor 12 Watts 8.00 kHz.PNG
Boat Anchor 12 Watts 12.00 kHz.PNG
 
For testing impedance I use the APx500 speaker upgrade software. There is a current sense resistor inside the APx1701. I connect a XLR cable between the APx1701 Current Sense output and the APx555 input 2. I connect the APx555 Input 1 with a XLR cable and small alligator clips directly to the Boat Anchor test resistor as you would do testing TS/P’s of a speaker.

Perfect. Thanks.

Today for a reality check I ran some plots with the AHB2 amplifier and the 300 Watt wire wound Boat Anchor test resistor.

To warm things up a little I turned the test voltage to input about 12 Watts into the Boat anchor. Then I made a few test frequency steps starting at 250Hz and stopping at 12.0 kHz.

Great idea. I hope we're not screwing up this thread with our back and forth, if so, somebody in mgt. say something and I'll quit. Otherwise, I'm still learning and that'll likely end when we're done cranking up the heat.

There was a 17dB increase in 3rd HD between a test frequency of 250Hz and 12.0 kHZ. This is not the result of VCR resistor nonlinearity alone. I was surprised; I expected decreasing distortion with increasing frequency

I'm not surprised - the AHB2 has increasing output impedance at HF. I'll repeat your tests with both the AHB2 and the near zero output impedance NC400. Amir mentioned that VCR was source impedance dependent.* I bet all kinds of distortion products are. I bet this extends to loudspeakers with and without passive crossovers. I bet I'm going to learn something here. : )

* And.. I bet the mirror image of this applies as well - that increased load impedance decreases resistor based distortions and vice-versa. It'd be a bit of a challenge to separate this from the amp's behavior.
 
Last edited:
Perfect. Thanks.



Great idea. I hope we're not screwing up this thread with our back and forth, if so, somebody in mgt. say something and I'll quit. Otherwise, I'm still learning and that'll likely end when we're done cranking up the heat.



I'm not surprised - the AHB2 has increasing output impedance at HF. I'll repeat your tests with both the AHB2 and the near zero output impedance NC400. Amir mentioned that VCR was source impedance dependent. I bet all kinds of distortion products are. I bet this extends to loudspeakers with and without passive crossovers. I bet I'm going to learn something here. : )

Just for fun I ran the same stepped frequency plots with the APx1701 amplifier and got nearly identical results.

Thanks DT
 
I'm not surprised - the AHB2 has increasing output impedance at HF. I'll repeat your tests with both the AHB2 and the near zero output impedance NC400. Amir mentioned that VCR was source impedance dependent.*
Yes, this was confirmed with my testing showing a lot of sensitivity with AHB2 but not with Hypex. The VCR coefficient gets divided down by the feedback ratio according to John at Benchmark. With hypex having tons of feedback, that may be the reason it is not impacted as much.
 
Great idea. I hope we're not screwing up this thread with our back and forth, if so, somebody in mgt. say something and I'll quit. Otherwise, I'm still learning and that'll likely end when we're done cranking up the heat.
No, carry on. This is a specialized thread so digging deep in the topic in however way you feel is fine. This information/discussion doesn't exist anywhere else when I went researching it back when.
 
Preliminary Conclusion
At this point my ARCOL NHS300 (Non-inductive, 300W) 4Ω 1% power resistors appear to be innocent of bad behavior.

Summary
The only amp I used to test them with today is the Hypex NC400, which as Amir noted uses lots of feedback to achieve its world-class distortion specs. Therefore, it is possible that amps that don't employ as much feedback and/or have higher output impedances may reveal non-ohmic behavior in my power resistors that were absent from today's measurements.

Stay tuned for the AHB2, which employs reduced negative feedback in favor of a 2nd smaller amp in parallel with each channel that provides "feed forward" error correction. I don't pretend to understand this, but the takeaway is that it's a very different class AB animal that should provide a very useful contrast to the class D NC400.

Measurements
I decided to use 2 second continuous sweeps at 96k/24bit. DT produced some lovely FFT's at 250Hz, 1kHz, 4kHz, 8kHz and 12kHz also at 96k/24bit, but I felt sweeps would be nicer. I also took the power thing seriously and performed sweeps at 1W, 5W, 10W, 100W and 200W. Everything was done at 4Ω, but I used (3) different 4Ω loads; (1) of the 4Ω ARCOL resistors, (4) of the 4Ω ARCOL resistors in series/parallel for a net 4Ω load, and something entirely different - a parallel array of (3) water heater elements with a net 4.2Ω load.

Heat
Neither of the ARCOL resistor setups showed any detectable warming. They remained as cool to the touch after the measurements as before. The water heater elements (rated at 4,500W each submerged) remained on the bench air-cooled (shown below). I did notice a very slight warmth to the touch after the 200W sweep.

RLoad VCR TCR Test.jpg


Step One
The only player in today's distortion measurements was (as expected) the 3rd harmonic, so let's dispense with the insignificant 2nd harmonic. As mentioned above, I made (5) sweeps with each of the (3) 4Ω load configurations. All (15) 2nd harmonic measurements overlap and barely contribute to THD. Higher orders than the 3rd are so far down they are (as expected) irrelevant.

2nd Order Harmonics
HD2 All.png


Step Two
Now we compare the single 4Ω ARCOL resistor to the quad series/parallel array that also nets to 4Ω. All (5) power levels effectively overlap. If some level of VCR (voltage coefficient) or TCR (temperature coefficient) distortion were within the range of my measurements, the quad array would have shown significantly lower distortion. The fact that the distortion traces overlap tells me that:

(a) the resistors are innocent, or
(b) my measurement setup isn't sensitive enough to detect the distortion, or
(c) the NC400 has sufficient feedback to eliminate the distortion that did exist.

Level vs. THD
NC400 1R vs 4R.png


3rd Order Harmonics
HD3 1R vs 4R.png


Step Three
Now we compare the single 4Ω ARCOL resistor to the paralleled water heater elements. These traces do NOT overlap. The paralleled elements show the same distortion at 1W, but much higher (3rd harmonic) distortion at the higher power levels.

Level vs. THD
NC400 1R vs Wtr Htr Elements.png


3rd Order Harmonics
HD3 1R vs Wtr Htr Elements.png


Results for the AHB2 vs. each of the (3) 4Ω loads tomorrow!

God bless you and your precious family - Langston
 
Last edited:
Conclusion
My ARCOL NHS300 (Non-inductive, 300W) 4Ω 1% power resistors are innocent and have been released by the parole board with a work permit on my bench on account of good behavior with secondary duties as a boat anchor.

I'm convinced that amp characterization using this resistor series (such as those published by ASR) can be trusted. I also have come to believe that even though I've found papers dating from the '50's on VCR and TCR distortion in resistors, this problem is still ignored in much of the amp testing you'll find online and largely explains why Amir's distortion results are often lower.

Summary
Today's tests used the AHB2 amplifier, which employs much less feedback than the NC400 to achieve its world-class distortion specs. As expected, this resulted in reduced distortion suppression of the poorly performing test load constructed from paralleled water heater elements.

Measurements
2 second continuous sweeps at 96k/24bit were used. Power levels were 1W, 5W, 10W, 100W and 200W. The (3) test loads were; (1) of the 4Ω ARCOL resistors, (4) of the 4Ω ARCOL resistors in series/parallel for a net 4Ω load, and a parallel array of (3) water heater elements with a net 4.2Ω load.

Heat
Neither of the ARCOL resistor setups showed any detectable warming. They remained as cool to the touch after the measurements as before. The water heater elements (rated at 4,500W each submerged) remained on the bench air-cooled. I did notice a very slight warmth to the touch after the 200W sweep.

Step One
Again, the major distortion component was the 3rd harmonic, and while the 2nd harmonic was again insignificant in most cases, it did start to rise at the 100W and 200W levels with the water heater elements. The measurements performed are the same as on the NC400; (5) sweeps using each of the (3) 4Ω load configurations. Again, higher orders than the 3rd harmonic were so far down they are had no material affect on THD.

2nd Order Harmonics
AHB2 H2.png


Step Two
Now we compare the single 4Ω ARCOL resistor to the quad series/parallel array that also nets to 4Ω. All (5) power levels effectively overlap. If some level of VCR (voltage coefficient) or TCR (temperature coefficient) distortion were within the range of my measurements, the quad array would have shown significantly lower distortion. The fact that the distortion traces overlap tells me that the resistors are innocent.

Level vs. THD
AHB2 1R vs 4R.png


3rd Order Harmonics
HD3 1R vs 4R.png


Step Three
Now we compare the single 4Ω ARCOL resistor to the paralleled water heater elements. These traces do NOT overlap. The paralleled elements show significantly more distortion at each power level.

Level vs. THD
AHB2 1R vs Wtr Htr Elements.png


3rd Order Harmonics
H3 1R vs Wtr Htr Elements.png


God bless you and your precious family - Langston
 
Last edited:
Thank you for sharing really detailed investigation on "dummy load".

Even though I am not an expert in this kind of precision measurements, I use ARCOL 8Ω 100W resistors for "burn-in" (in Japan we call "aging") of my amplifiers as shared here in my thread.

I also use Jantzen Audio's 22.00 Ohm 10W Audio-Grade Non-Inductive Wirewound SuperRes Resistor for fine tuning of midrange drivers, tweeters and super tweeters in my multichannel multi-amplifier configuration as shared here and here. I asked several places (like here) in ASR Forum that whether the subjectively favorable effect of such a small constant and extra power load to amps can be explained and/or justified in terms of the zero-cross distortions in the amps, or not. So far, however, no one gave me proper insight in this regards...

BTW, I found/confirmed it is really important to avoid any magnetic susceptible metals (terminal strips, screws, etc.) in these SP level lines to keep excellent Hi-Fi level sound quality, as I shared here, here and here.
 
Last edited:
Thank you for sharing really detailed investigation on "dummy load".

Even though I am not an expert in this kind of precision measurements, I use ARCOL 8Ω 100W resistors for "burn-in" (in Japan we call "aging") of my amplifiers as shared here in my thread.

I also use Jantzen Audio's 22.00 Ohm 10W Audio-Grade Non-Inductive Wirewound SuperRes Resistor for fine tuning of midrange drivers, tweeters and super tweeters in my multichannel multi-amplifier configuration as shared here and here. I asked several places (like here) in ASR Forum that whether the subjectively favorable effect of such a small constant and extra power load to amps can be explained and/or justified in terms of the zero-cross distortions in the amps, or not. So far, however, no one gave me proper insight in this regards...

BTW, I found/confirmed it is really important to avoid any magnetic susceptible metals (terminal strips, screws, etc.) in these SP level lines to keep excellent Hi-Fi level sound quality, as I shared here, here and here.

Hello,

JBL uses a fixed L-Pad in the tweeter crossover for the M2 Monitor speaker.

An 11R 10 Watt resistor is used in parallel with the tweeter and another 13R 30 Watt resistor is in series with the parallel tweeter and 11R resistor.

What this does is operates the amplifier at a higher power output where the amplifier operates at a higher SNR. Also the amplifier operates into a flatter, more resistive load. I use this fixed L-Pad configuration in my tri-amp system. See page 14 https://rephase.org/projects/JBL_M2_crossover.pdf .

In terms of Class AB operation most AB amplifiers overate in Class A for the first few watts. In terms of Sound Quality try it and let us know what you think.

Thanks DT
 
Back
Top Bottom