Building A Reactive Load for Amplifier Testing

[Blumlein 88]

I hate to say it as I've been one of those asking for reactive testing. Looking more closely now, I think it is of very little utility. I think testing into 1, 2, 4, and 8 ohms is really all you need (though I'd like the addition of 16 ohms). It appears using tone burst testing will do it so very little danger to the DUT.

Basically you need to know the amps current, and voltage operating envelopes. And you don't need reactive testing to do that.

If for completeness sake one still wishes to actually test into a reactive load I think the following Power Cube results indicate it is a waste of time to test at 30 and 60 degrees. Things could be simplified by testing only at 0 and 60 degrees. Those inductive tests mainly point to amplifier instability. So you really only need the more extreme test to indicate it is happening in a given amp. So that alone would simplify what is needed for a reactive load.

Here are some results from the Power Cube test. I think you'll see even when the result wasn't close to great you didn't learn anything you couldn't learn from a straight resistive test. A simple graph of volts over impedance will show plenty about the amps capabilities.

KRELL KSA 80B

8Ohm
158W 60deg Cap
157W 30deg Cap
156W 0 deg
157W 30deg Ind
159W 60deg Ind

4Ohm
303W 60deg Cap
294W 30deg Cap
292W 0 deg
296W 30deg Ind
302W 60deg Ind

2Ohm
557W 60deg Cap
518W 30deg Cap
506W 0 deg
518W 30deg Ind
545W 60deg Ind

1Ohm
403W 60deg Cap
407W 30deg Cap
421W 0 deg
417W 30deg Ind
332W 60deg Ind

MCC POS: 70A
MCC NEG: 0A*

*could not measure ?
from:
KRELL KSA 80B

from: https://www.diyaudio.com/forums/solid-state/78129-krell-ksa-100mkii-clone-post1974700.html

from: https://www.biline.ca/audio_critic/audio_critic_web3.htm

from: http://erha.se/~ronny/hifi/xtz/a100.htm

from: https://www.biline.ca/audio_critic/audio_critic_web1.htm

 

[Blumlein 88]

https://www.biline.ca/audio_critic/mags/The_Audio_Critic_20_r.pdf

Good the Audio Critics volumes are available here.

In this one, starting on page 18 there is some detailed discussion of the Power cube and its operations. I wondered how they were getting peak current values. They use one cycle of a 10 khz tone into .1 ohm load. I actually don't see the usefulness of this particular measurement. I'd omit it.

Though not graphed in this issue several large amps are tested and Cube results described.

 

[amirm]

Years ago, when magazines still employed engineers to do reviews, HiFi News used an 8ohm load in parallel with 2uF as their standard test load. They did all THD and square wave testing into that load and compared it to the non-reactive pure 8ohm load. It fail to see why that can't be the ASR standard as it will suitably stress amplifiers, especially when combined with 4 and 2ohm resistive loads, go down to 1ohm if you must.

I tried that for a while when testing the Hypex NC400. I got pushback from an amp designer. Basically all it does is reduce the load impedance proportional to frequency, hence increasing distortion. We could just put in a lower impedance resistive load and get the same thing.

 

[sergeauckland]

I tried that for a while when testing the Hypex NC400. I got pushback from an amp designer. Basically all it does is reduce the load impedance proportional to frequency, hence increasing distortion. We could just put in a lower impedance resistive load and get the same thing.

Agreed, but it's a good way of seeing what an amplifier does with square waves into a capacitative load. One could do much the same with an inductive load, but my understanding is that amplifiers are more affected by capacitance than inductance, who's impedance increases with frequency.

S

 

[VMAT4]

I looked at the Stereophile circuit. And, I have to wonder if the circuit needs to be so complex. Wouldn't there be a Thevenin or Norton equivalent for it?

 

[Speedskater]

I looked at the Stereophile circuit. And, I have to wonder if the circuit needs to be so complex. Wouldn't there be a Thevenin or Norton equivalent for it?

Some time ago, Arny Krueger (RIP) was a main part of a discussion in one of the forums, about why the Stereophile circuit was to simple.

 

[restorer-john]

Years ago, when magazines still employed engineers to do reviews, HiFi News used an 8ohm load in parallel with 2uF as their standard test load. They did all THD and square wave testing into that load and compared it to the non-reactive pure 8ohm load. It fail to see why that can't be the ASR standard as it will suitably stress amplifiers, especially when combined with 4 and 2ohm resistive loads, go down to 1ohm if you must. If an amp can have low THD into the resistive loads and generate decent square waves into the reactive load, there's not a lot wrong with it.

Because a single capacitive load is not enough. Sure, it is useful and gave us a few decades of pretty oscillograms in HiFi reviews. The overload recovery behaviour into capacitive loads is/was particularly interesting and that is yet another area not considered these days.

In the old days, everything was rated into a standard 8 ohms. 4 ohms is becoming the standard for headline advertised power output numbers and there's no doubt nominal rated impedances have at least halved since the 1970s.

I support building a common basic design, multi network reactive load-box anyone can build or adapt to suit their own testing requirements.

 

[restorer-john]

I think testing into 1, 2, 4, and 8 ohms is really all you need (though I'd like the addition of 16 ohms). It appears using tone burst testing will do it so very little danger to the DUT.

Basically you need to know the amps current, and voltage operating envelopes. And you don't need reactive testing to do that.

If for completeness sake one still wishes to actually test into a reactive load I think the following Power Cube results indicate it is a waste of time to test at 30 and 60 degrees. Things could be simplified by testing only at 0 and 60 degrees. Those inductive tests mainly point to amplifier instability.

These points are valid, but I can assure you there are amplifiers that will only have their flaws flushed out in extreme situations such as the inductive/capacitive tests. Coupled in the real-word, with difficult (multi element steep slope Xovers) things can get ugly, fast.

These days, as SergeA pointed out, the real investigations are simply not done in reviews anywhere. It's been missing in audio reviews for the best part of 30 years. Basically, since the subjective crowd took over. We see best case, single point, selling numbers on specs, controlled situations not disclosed, and anything that may paint the product in a more honest light is excluded.

That said, if you kill the reactive and capacitive tests, what have you got? Just benign fixed loads. Sure, we can do CEA/IHF and continuous, 1 channel driven vs 2 and even the peak current single cycle 1ohm (or 0.1 ohm) tests if you want (can't really see a value in that myself other than a big number on a spec sheet) and/or a blown output stage.

We all have load banks (well some of us) and can tack on reactive components for single point tests, but the visual and easily understood nature of the represented data points in the power cube is a compelling reason to have a 'box that does it all' don't you think? At-a-glance analysis is easier than a text box full of numbers.

 

[Blumlein 88]

These points are valid, but I can assure you there are amplifiers that will only have their flaws flushed out in extreme situations such as the inductive/capacitive tests. Coupled in the real-word, with difficult (multi element steep slope Xovers) things can get ugly, fast.

These days, as SergeA pointed out, the real investigations are simply not done in reviews anywhere. It's been missing in audio reviews for the best part of 30 years. Basically, since the subjective crowd took over. We see best case, single point, selling numbers on specs, controlled situations not disclosed, and anything that may paint the product in a more honest light is excluded.

That said, if you kill the reactive and capacitive tests, what have you got? Just benign fixed loads. Sure, we can do CEA/IHF and continuous, 1 channel driven vs 2 and even the peak current single cycle 1ohm (or 0.1 ohm) tests if you want (can't really see a value in that myself other than a big number on a spec sheet) and/or a blown output stage.

We all have load banks (well some of us) and can tack on reactive components for single point tests, but the visual and easily understood nature of the represented data points in the power cube is a compelling reason to have a 'box that does it all' don't you think? At-a-glance analysis is easier than a text box full of numbers.

Well if impedances are the same, the difference in fixed loads and reactive loads is dissipation stress on the output transistors. If they can handle it they'll show the same results more or less. The manner of testing being discussed isn't actually going to be testing this. I was expecting tests at 1 ohm to be the same as higher impedance testing. So I think that tells us enough. Rather than a box of numbers you'd get a graph at least. That isn't so hard to look at I don't think.

I certainly don't mind the reactive addition. It might not be of much real extra benefit.

If speaker reviews included equivalent peak dissipation resistance (EPDR) vs frequency graphs then resistive testing could be combined with it for any given speaker for which you have the impedance curve.

https://www.stereophile.com/reference/707heavy/index.html

Some of this can also be found in Douglas Self's Amplifier book on page 209 and the next several pages.

There is an Eric Benjamin article I don't have access to about the same discussion.

http://www.aes.org/e-lib/browse.cfm?elib=6931

 

[restorer-john]

Rather than a box of numbers you'd get a graph at least.

I agree. Here's one of my boxes of numbers, just 8 and 4 ohms, continuous, both channels vs single and burst. If I added 2 and 1 ohms, lots of vintage gear wouldn't survive unless it was burst only.

 

[Blumlein 88]

I agree. Here's one of my boxes of numbers, just 8 and 4 ohms, continuous, both channels vs single and burst. If I added 2 and 1 ohms, lots of vintage gear wouldn't survive unless it was burst only.

View attachment 28588

From this it appears the toneburst method gives roughly the same results as continuous testing. Have you found that to usually be the case for most amps you've tested?

 

[dkfan9]

Rod Elliot's circuit: http://sound.whsites.net/patd.htm

I realize after reading through that this isn't exactly what's being looked for, but figured it's worth a post if just for reference.

 

[restorer-john]

From this it appears the toneburst method gives roughly the same results as continuous testing. Have you found that to usually be the case for most amps you've tested?

Sadly, no, but that said, my tests were usually one or the other in the past.

The lower powered well designed amplifiers generally have burst/continuous numbers that are close. Then as you go up in power, the gap between continuous and burst can widen significantly or not, depending on the regulation of the PSU (TXF and caps). The biggest differences of course are single channel vs both channels driven.

I'll do a bunch of tests this week, on a pile of amps here, and post the results. I'd like to know myself what the spread is between cont/burst.

 

[Blumlein 88]

Sadly, no, but that said, my tests were usually one or the other in the past.

The lower powered well designed amplifiers generally have burst/continuous numbers that are close. Then as you go up in power, the gap between continuous and burst can widen significantly or not, depending on the regulation of the PSU (TXF and caps). The biggest differences of course are single channel vs both channels driven.

I'll do a bunch of tests this week, on a pile of amps here, and post the results. I'd like to know myself what the spread is between cont/burst.

That's more or less what I expected. Look forward to seeing what you get for results.

 

[Blumlein 88]

I am certainly interested in the results of this execise. As a lifetime Quad electrostatic owner I have always cautiously stuck to Quad´s own amplifiers (a 303 with the ELS 57 and now a 606-2 with the 2805), and, to be honest, not regretted it. So I would love to see how their large current dumpers like my 606-2 or the modern Artera are testing. And I would love to know if, in fact, I have a lot more options, or not.

You have more options. The big deal with the ESL 57 is not to send it too much voltage. 33 v peak or 23 volts RMS. Quad 405's had an option to limit output voltage to 20 volts RMS for use with the ESL57. Generally an SS amp rated at 50 wpc into 8 ohms would be safe with the ESL57. The old C-J MV45 and MV50 were a really nice match with the ESL57 as was a VTL 75/75 converted to triode for half of the 75 wpc.

The 2805 has a max voltage input of 38 volts RMS. So any amp of 150 wpc at 8 ohms or less should be safe on it. Often a good 75 wpc amp is enough.

 

[Willem]

I stil own the combinaton of ELS 57 and 303 power amp. That is a very good combination.
The 2805 clearly needs more power hence my choice for the 2x140 watt 606-2. I am considering moving up to the bridged 2x260 watt QMP monoblocks. This is in a large listening room.

 

[JohnYang1997]

I see this thread have been up for quite a long time now. Why not just use real speakers. You can get a few different ones for few hundred bucks. If it's too loud when testing, why would anyone need that anyway. Or if still needs to test at high level. You can either get an isolation box (eg for electric guitar cabinet) or build one your own and let second channel play inverted signal. Thus won't be so annoying.
Randall has isolation boxes. But i think a big shipping box + inverted signal is a good option.
Thanks amirm.

 

[restorer-john]

Why not just use real speakers.

Because we are testing amplifiers, not the longevity of speakers when subjected to high powered burst signals. 'A few different' speakers will not give the range of load angles we need to characterize an amplifier's abilities, nor will it allow direct comparison with other similar, pre-existing load boxes out there.

 

[JohnYang1997]

Because we are testing amplifiers, not the longevity of speakers when subjected to high powered burst signals. 'A few different' speakers will not give the range of load angles we need to characterize an amplifier's abilities, nor will it allow direct comparison with other similar, pre-existing load boxes out there.

I'm not sure why it's needed. It's a bi directional thing. You can always create worse behaved load. So worse behaved speakers can be avoided first.
The thing is why do we need it. I presume it's to use reactive load to measure distortion, multitone distortion, imd etc at 5W 10W 100W or something like that. If it's to assess the stability of amplifier just use capacitors. 1n 2n 10n 100n. If there is any other reason, excuse my ignorance.

 

[amirm]

I see this thread have been up for quite a long time now. Why not just use real speakers. You can get a few different ones for few hundred bucks. If it's too loud when testing, why would anyone need that anyway. Or if still needs to test at high level. You can either get an isolation box (eg for electric guitar cabinet) or build one your own and let second channel play inverted signal. Thus won't be so annoying.
Randall has isolation boxes. But i think a big shipping box + inverted signal is a good option.
Thanks amirm.

That's a good idea! But those Randall boxes are expensive at $450.