• 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!

Building A Reactive Load for Amplifier Testing

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
That's a good idea! But those Randall boxes are expensive at $450.
But at least not 25 grand right haha. The third one is to use phase cancellation instead of isolation box. Putting two speaker face to face playing inverted signal. If you stand far enough it's really quite..
 

Murrayp

Member
Forum Donor
Joined
Oct 30, 2016
Messages
42
Likes
86
Location
Napier, New Zealand
I wonder if dynamic loads mightn't be much more telling for an amplifier than mere static loads? Speaker cones have mass and stored energy - I suspect this can't be entirely modelled with reactances and resistances. What if a lf step is applied to a speaker at the same time as a delicate cymbal sound? The power devices are busy trying to accelerate and then brake a hefty woofer cone into position while perfectly reproducing some hf content - damn tricky if the bulk current is actually flowing the wrong way due to the overshoot of the woofer as it possibly rings to it's new position part of the time. It's remarkable the things manage this at all. Sheer current or dissipation is one aspect but response to the dynamic nature of music is possibly much more interesting. I guess IMD tests are a step in this direction but personally I'd be much happier if complex waveforms could be used with real speaker loads for testing. I guess then measuring accurately might take some interesting maths....... Anyway - here's another vote for using real speakers as a load for part of the testing at least.
 

jazzendapus

Member
Joined
Apr 25, 2019
Messages
71
Likes
149
Potentially stupid question ahead: if an amplifier behaves well under extreme conditions (a speaker with low nominal impedance, lowest impedance in areas where music has a lot of energy, wild swings in both impedance and phase etc.) can it be assumed that it will behave just as well or better under less extreme conditions? Or are there numerous possible and different combinations of "extreme conditions" that should be taken into account to get a clear picture? If the answer is yes then maybe go with what JohnYang1997 suggested, find a couple of garbo speakers that answer to that extreme criteria, integrate them into future amp tests and be done with the whole ordeal.
 

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,440
Likes
9,100
Location
Suffolk UK
The problem with using real loudspeakers for testing is that of standardisation. For measurements to mean anything, they have to be standardised so comparisons can be made. That's why 4 and 8 ohm pure resistive loads are the norm, they're the same for everyone.

For a reactive load, one can do as the Power Cube tests do, with known inductive and capacitative loads, and that is a reasonably accepted standard. Alternatively, something simpler like the old 2uF 8 ohm, but that has its limitations.

How difficult would it be to make a switchable 2-4-8 ohm load with switchable inductive and capacitative loading?

S.
 

somebodyelse

Major Contributor
Joined
Dec 5, 2018
Messages
3,682
Likes
2,959
Speaker cones have mass and stored energy - I suspect this can't be entirely modelled with reactances and resistances.
The equations used to model electrical, mechanical and acoustic systems are in fact the same, but with different names and units. The voltage in the electrical system is equivalent to mechanical force and acoustic pressure. Inductance is equivalent to mass, capacitance to compliance and so on. The speaker's motor is modelled as a transformer with a ratio determined by the magnetic field strength and the length of wire in that field. You will find more detail about the relationships in any textbook on speaker analysis.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
How difficult would it be to make a switchable 2-4-8 ohm load with switchable inductive and capacitative loading?
That's easy. Problem is when you add just reactive loading, you change impedance with frequency now. The result is rising distortion proportional to that change so doesn't teach us anything. This is what happened when I added capacitors to my dummy load.

What powercube does is to keep impedance the same while changing the ratio of reactance to resistance variable. This now requires essentially building a mini-crossover circuit for every load combination we want to test which is a lot of work given the inductors and capacitors for each load we want to create.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
The problem with using real loudspeakers for testing is that of standardisation. For measurements to mean anything, they have to be standardised so comparisons can be made. That's why 4 and 8 ohm pure resistive loads are the norm, they're the same for everyone.
I am not beyond creating our own standard! :) It is not like there is an army of people doing measurements anyway. I have already deviated by standardizing on 5 watt for baseline results as to have hope of distortion showing up as opposed to just noise.

So what if we find a stout speaker that anyone can source for this type of testing? We could initially just use it for 5 watt testing if it can't handle all the power. Or sweep up to 20 watts or whatever.
 

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,440
Likes
9,100
Location
Suffolk UK
I am not beyond creating our own standard! :) It is not like there is an army of people doing measurements anyway. I have already deviated by standardizing on 5 watt for baseline results as to have hope of distortion showing up as opposed to just noise.

So what if we find a stout speaker that anyone can source for this type of testing? We could initially just use it for 5 watt testing if it can't handle all the power. Or sweep up to 20 watts or whatever.
Yes, that could create an ASR standard , but that loudspeaker would have to present a difficult load, and be commonly available. Can't think offhand of anything, but I haven't looked too closely at loudspeaker impedance graphs.

Thanks also for the explanation as to what powercube does with impedance. I had thought it was rather less complicated.

S
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,278
Location
Gold Coast, Queensland, Australia
That's easy. Problem is when you add just reactive loading, you change impedance with frequency now. The result is rising distortion proportional to that change so doesn't teach us anything. This is what happened when I added capacitors to my dummy load.

What powercube does is to keep impedance the same while changing the ratio of reactance to resistance variable. This now requires essentially building a mini-crossover circuit for every load combination we want to test which is a lot of work given the inductors and capacitors for each load we want to create.

We have the purely resistive loads 8/4/2/1 plus 16 other dedicated networks to build. If we skip 1 ohm resistive, it saves 4 of those networks.
The individual networks are optimized for one frequency, 1KHz.
It's 20mS per 500mS duty cycle.

Can the AP ramp up its output and resolve a trigger point THD of 1% from the DUT and store that value based on the 20mS bursts separated by 480mS, then move onto to the next test and store the whole lot of results via the data logging function? How long would it take to get a reliable lock on 1% THD using only 20mS bursts?

Is the burst test (CEA/IHF) built-in already or do you have to create it with this? :

https://www.ap.com/download/apx-burst-waveform-utility/
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
Can the AP ramp up its output and resolve a trigger point THD of 1% from the DUT and store that value based on the 20mS bursts separated by 480mS, then move onto to the next test and store the whole lot of results via the data logging function? How long would it take to get a reliable lock on 1% THD using only 20mS bursts?
Yes. You give it the range of input voltages into the amp and it will then hunt between them until it finds the optimal value that gives whatever you specify as distortion. I have tested it for continuous tone. I will have to experiment with bursts.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
Is the burst test (CEA/IHF) built-in already or do you have to create it with this? :

https://www.ap.com/download/apx-burst-waveform-utility/
It can do burst:

1562715270499.png


Those files are useful when the analyzer can't directly feed the device under test. We don't need that in this instance.
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,341
Likes
688
The Apogee Scintilla was infamously difficult, wasn't it? A long time ago, however, and the company is long dead.
Ah that was SUCH a lovely speaker! I was passing by a room with those playing and it really sounded like someone was playing a drum kit. IN the room, it sounded like there was another room behind the speakers. Yes, down to 1 ohm http://www.lippaudio.org/old/MySystems/Scintilla/impedance.html
the one speaker I ever heard cables sound different on (though one set was MIT, which had a passive network, so maybe not a big mystery about hearing a difference)
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,341
Likes
688
Basically you need to know the amps current, and voltage operating envelopes. And you don't need reactive testing to do that.
I worked for a big manufacturer and got them to buy the PowerCube, and I can't agree at least for automotive stuff. Unless maybe we're not thinking quite the same thing. Some lesser designed amps (not like the Krell and Rotel stuff you show) would freak out with non-resistive loads.
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,341
Likes
688
This now requires essentially building a mini-crossover circuit for every load combination we want to test which is a lot of work
Yeah. So based on a bunch of fiddling around back in the day, I think
(1) ANY reactive testing is way better than the NONE that goes on now
(2) I think it IS interesting to go down to 1 ohm (though grant this is more relevant for automotive, with more low-impedance speakers and frequent paralleling of subwoofers)
(3) I think one bass frequency (some non-multiple of 60 Hz) is sufficient to be way more interesting than just resistors. Because if the alternative is "OMG to build this for 50 and 1k and 15k is WAY too hard, forget it" then I'd rather have something than nothing. And the bass frequencies stress the amps the hardest with music signals. Yes that probably means very large capacitive and inductive elements.
Maybe Neurochrome could build something nice :)
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
I am afraid that the goal here is fast automated testing rather than deep investigation of one product or topology that take days or weeks. It is understandable based on the fact that it is to make living.
 

Ralph_Cramden

Major Contributor
Joined
Dec 6, 2020
Messages
2,574
Likes
3,468
I wonder if @John Atkinson still uses the load that Ken Kantor designed years ago?

 

John Atkinson

Active Member
Industry Insider
Reviewer
Joined
Mar 20, 2020
Messages
165
Likes
1,022
I wonder if @John Atkinson still uses the load that Ken Kantor designed years ago?

Still do, and use it for every amplifier review. See, for example, figs.1 & 2 at https://www.stereophile.com/content/vac-sigma-170i-iq-integrated-amplifier-measurements . The only change I made to Ken's circuit was to add a Zobel network---a resistor and capacitor in series---across the input terminals so that the top-octave impedance more closely resembled real-world loudspeakers.

John Atkinson
Technical Editor, Stereophile
 

MC_RME

Addicted to Fun and Learning
Technical Expert
Audio Company
Joined
May 15, 2019
Messages
854
Likes
3,564
Values please?
 
Top Bottom