• 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!

Complex Load for Power Amplifier torture testing

Hugo9000

Addicted to Fun and Learning
Joined
Jul 21, 2018
Messages
574
Likes
1,749
Location
U.S.A. | Слава Україні
Hey it was the Mark Levinson that Audio magazine actually used to weld with. Shot 1 khz through it at max power and welded two pieces of steel together. An ML 3 I believe.
It was in the August 1987 issue of Audio, using a pair of Mark Levinson No. 20 amplifiers. (I had to look it up, had no clue what model it was haha, but fortunately most issues of that magazine are online! :) )

Here is a pdf of the sidebar to the review:
Audio-1987-08-OCR-Page-0068.pdf

And a link for the issue it appeared in:
https://www.americanradiohistory.com/Archive-Audio/80s/Audio-1987-08.pdf


An enthusiastic reviewer might call a beefy amplifier an "arc welder" as an exaggerated compliment to its ruggedness and current -handling capacity. However, no one would really expect an amplifier to actually melt steel. Almost no one, that is, except this reviewing team. We say this amplifier is an arc welder and back up this statement with a photograph (Fig. B1) of two 0.05 -inch steel plates welded together by a pair of Mark Levinson No. 20s.
 

Attachments

  • Audio-1987-08-OCR-Page-0068.pdf
    77.6 KB · Views: 876
Last edited:

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,278
Location
Gold Coast, Queensland, Australia
But, I thought they were from New Zealand, not Australia, senility...

You're not senile! That brochure was from the Aussie distributor at the time. Perreaux of course were based and manufactured in Napier, NZ.

I heard that amp with a pair of NZ made Lambert Audio speakers in a ballroom at the Gold Coast International Hotel. Hardly anyone was looking at the big Perreauxes, but after reading about them, I sought them out to listen. The 5150b had a perspex top to see the guts. The output inductor was about as big as a can of coke...

(not my pic)
1576702628544.png
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,278
Location
Gold Coast, Queensland, Australia
Hey it was the Mark Levinson that Audio magazine actually used to weld with. Shot 1 khz through it at max power and welded two pieces of steel together. An ML 3 I believe.

Was never sure if it was actually true or an urban legend. Those huge MLs, Krells and Perreauxes were referred to as arc-welders. Amplifiers to be scared of. Ones that could swing enough at the speaker jacks to kill you etc.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Here is the reason why my KA-5010 clone passed the test (with the complex load at 4.6Vrms output voltage) and Yamaha AX-396 did not and triggered off at 3400Hz. 1st, my KA-5010 clone has supply voltage only 2x27V, though Yamaha has 2x46V, the higher voltage stresses much more the output pair. 2nd, KA-5010 clone uses MJL21193/94 output devices which have much higher SOA (safe operating area) than the 2SA1695/2SC4468 used in Yamaha. I do not understand how Yamaha designers could use only one pair of these not very rugged devices in the amplifier rated at 2x95W/4ohm.

KA-5010 clone SOA and complex load Ic/Vce of one output transistor. Blue curve must be below the red line.
complex_load_SOA_ka5010.png


Yamaha AX-396 SOA and complex load Ic/Vce of one output transistor. Blue curve must be below the red line, which is not fulfilled.
complex_load_SOA.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
I have decided to "compromise" the complex load a bit by reducing C1 from 33uF to 6.8uF. In case of Yamaha AX-396 tested at 4.6Vrms output voltage, it did not help much. The amplifier did not like it again, triggered off itself at 15820Hz, SOA was again exceeded. No oscillations during the test.

complex_load2.png


ax396_complexload2_dist.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
This is impedance plot of another dummy load that I use to test amplifiers. This is not a "torture" load as described in #1 post, but rather a load that may simulate impedance of a simple 3-way conventional loudspeaker box.

PMA_dummyload_impedance.png
dummyload_PMA_cir.png


dummy_real.jpg
 
Last edited:

anmpr1

Major Contributor
Forum Donor
Joined
Oct 11, 2018
Messages
3,722
Likes
6,406
Was never sure if it was actually true or an urban legend. Those huge MLs, Krells and Perreauxes were referred to as arc-welders. Amplifiers to be scared of. Ones that could swing enough at the speaker jacks to kill you etc.
I came across this at Salvatore's site. I have no idea about it, either, but it was fun to read.

The [Bob Carver] Sunfire was designed to drive any load, do it with very few output transistors, very simple circuitry, and to emulate the sound of tubes. I'll let the shock of what I'm saying sink in. If you can find a Sunfire amplifier... buy it! You will simply not believe how it transforms any of the Apogees [low impedance loudspeaker popular back in the day].

We actually did little parlor tricks with the Sunfire. With the protection circuitry disabled, you could weld copper sheet together! It would literally keep going right through a dead short! Bob put a 1/4 Ohm protection circuit in, in his words, to protect himself from lawsuits in the event of a voice coil short circuit. The amp would keep on going as a result of the tracking down converter and cause the voice coil to overheat and possibly catch fire...
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,341
Likes
688
Audiograph Powercube and it costs US $15,000 for a stereo unit with up to 500 watts capacity...
Yeah, that's the problem. When I was at a humongous audio company that shall remain nameless to protect the guilty, I just badgered people until they bought the thing. On the one hand, the PowerCube let me make some marketing hay to beat up on competitors, on the other hand widening the stability versus complex loads doesn't increase the raw dollar per watt that a lot of customers focus on. [flashback vignette from another company: Me: "Why is this amp rated at 63 watts, it's such a weird number?" Naïf: "Uh, because that's how much power it makes!" Me: [hopes for world amp domination fading] "Uhhhhhh...never mind."

Later when I was starting a company with a buddy, I wanted to make cost-reduced just-LF version for checking if Chinese stuff was cr@p or decent. I never did decide what a good frequency would be-50 and 60 are mains frequencies. I was thinking 70 Hz for fewer common harmonics, however just did not have time and budget and other things happened.

I didn't read this whole long thread yet, but do think the test load is pretty extreme at high frequencies (UNLESS it's to sort out amps for those rare quasi-capacitor speaker models, whose buyers I hope realize they need special amps).

I am still wondering the truth between Matti Ottala's assertion that an 8 ohm speaker could appear like a 1 ohm load under transient conditions, and a much later paper asserting that they saw zero evidence that real speakers ever drew more current than their D.C. resistance would indicate. (Ottala's paper was based on computer simulation. It led IIRC to "high-current" designs like Harman and others, whose benefit was maybe not related to the instantaneous load but more to generally bigger SOA?). Some fine day I will try to rig something up to test this and write another AES paper.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
Thanks. One way to cut costs relative to Powercube is to just focus on one impedance with different phase angles. Even then, my single resistive load cost hundreds of dollars (had to be ultra low distortion). I shutter to think of replicating that four ways....
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,341
Likes
688
[Bob Carver]...put a 1/4 Ohm protection circuit in, in his words, to protect himself from lawsuits
In college, my new rich Texas roomie decided someone else's stereo was almost as loud as mine (a little NAD 3020 driving speakers I built to 117 dB, mind you!). So he bought a Carver touring sound amp, like 500 watts or such. It had two limiter knobs, like 30 seconds and 1 second or something. The next generation removed those. When I asked why, the answer was that the amp would shut off and sound guys would be pissed. With the limiters removed, the speakers would smoke and burn, and the sound guys were now way impressed how powerful the amps were...
...almost as funny as my friend who babied Aura 1808 subs through seasons of Lollapalooza and IIRC Def Leppard and many others, strapping the large heavy bulbous magnets in place as they had a tendency to just snap off. Which was because some fool designed the basket so that at one point the wall of the basket legs was 1 mm (not 10! ONE!) thick aluminum...
...I couldn't make this stuff up if I tried.
 

anmpr1

Major Contributor
Forum Donor
Joined
Oct 11, 2018
Messages
3,722
Likes
6,406
Thanks. One way to cut costs relative to Powercube is to just focus on one impedance with different phase angles. Even then, my single resistive load cost hundreds of dollars (had to be ultra low distortion). I shutter to think of replicating that four ways....
Power Cube as a research tool is one thing. To discuss it in the context of a 'real world' amplifier-speaker interface has its own issues. I am not sure how flexible the device is, but when Audio Critic was using it their protocol was "1% THD limited to a 1KHz sine wave, along with instantaneous peak current (10 KHz into 0.1 ohm)." See examples below.

While this tells more than the usual 8 ohm test, I can imagine some folks asking whether what is graphed represents what is actually going on at the amp/speaker interface. And 1% distortion is pretty forgiving in today's amplifier context. In any case, at the price it is a tool that is really out of the question for reviewers not having a huge purse. Finally, although I do not know the details, in one of his posts Aczel wrote that the machine broke down, and needed a part or something. If that sort of thing is what can be expected for fifteen or twenty thousand dollars...

The_Audio_Critic_20_r.jpg
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,341
Likes
688
Power Cube...I am not sure how flexible the device is, but when Audio Critic was using it their protocol was "1% THD limited to a 1KHz sine wave...

While this tells more than the usual 8 ohm test, I can imagine some folks asking whether what is graphed represents what is actually going on at the amp/speaker interface. And 1% distortion is pretty forgiving in today's amplifier context.
Oh, even Lars Ohlen would tell you the PowerCube is hardly fully representing the amp-loudspeaker interface. The main point is you certainly get a much fuller picture of how robust the amp is compared to resistors, particularly (thinking on the fly) if you connected a storage scope to capture waveform misbehavior at the same time. Distortion was set at 1% since that's often used in lesser specifications, and by Stereophile, and to not exclude various quality amps which by low-feedback design (a Pass amp tested here at ASR for instance) cannot hit lower distortion. 1 kHz is "too easy" of a frequency; I would have used 41 Hz or 82 Hz or something. However over dinner with Lars once he said those loads for high power would get super expensive. The PowerCube is a truly pricy beast, it's the same old story that any limited manufacturer has to charge a lot to support living wages for the entire staff. There's a load which will connect to AudioPrecision now though I don't know the price.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,278
Location
Gold Coast, Queensland, Australia
sell popcorn and take for several buck non-inductive resistor with bifilar wire winding. Such resistors don't have any current noise and have only thermal noise. And this thermal noise will be hundred times less then Audio presicion has.

VCR. It becomes a serious consideration where high voltages (high powered amplifiers) and extreme precision are required. Basically, you get non-linear behaviour that varies with the applied voltage to the load instead of being a perfectly linear relationship.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
"I'm sure there're plenty of amplifiers that show incredible 0,00001% THD on resistors and suck when are loaded on real speakers. "
proof?
I agree real speaker load is useful. But wtf is this? This is getting into completely opposite territory.
 

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
sell popcorn and take for several buck non-inductive resistor with bifilar wire winding. Such resistors don't have any current noise and have only thermal noise. And this thermal noise will be hundred times less then Audio presicion has. For very very scientifiс measurements put resistor in a water with ice.
@amirm will be along soon ;)
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,278
Location
Gold Coast, Queensland, Australia
Unfortunately admins like @amirm here still performing useless tests of amplifiers on resistors

I'd hardly call a decades old industry standard testing protocol "useless". Does it tell the whole story? No. Does it provide a way to compare products on a somewhat level playing field? Yes.

Is it perfect? No. That's why complex/simulated loads are used. No load can simulate every situation can it?
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Measuring on resistor - what is completely opposite territory in comparison to real world usage.
Simply:
1, Have you measure any amplifier by yourself or seen from a valid source that many amplifiers have -140db distortion?
2, Do you have any proof that there's no correlation between good performance measured using resistor load and using real speaker load?
3, Do you have examples with evidence that the name you called, NCore, measures poorly with real speaker load?
4, Do you know how to calculate/interpret dynamic impedance, phase, and SOA related to the topic?
 
Top Bottom