• 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

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,716
Likes
38,889
Location
Gold Coast, Queensland, Australia
I feel like I'm talking to a wall. :facepalm:

I don't care one bit about a particular digital source limitation in this discussion. Not one iota. It has nothing to do with this thread. Is it not clear?

This thread is about amplifiers and what happens to them in reactive conditions. It is about waveform integrity in the face of opposition. You chose yet again go on the defensive, boots and all, for Class D.

Nobody was attacking Class D or your amplifiers specifically but now it ends up another sh#t-show. It doesn't need to.
 
  • Like
Reactions: pma

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,321
Location
Albany Western Australia
You are not listening and taking the information on board John.

No I made a simple point about bandwidth limiting which is relevant to ALL amplifiers, and this misguided idea you have that looking like a square wave is so very important, or that otherwise it indicates instability or poor quality.

It was actually you who went on the defensive (attack) and totally missing the context of my post. Its actually your response to a simple technical point that turned it into into shit show.
 
Last edited:

BaaM

Member
Joined
Nov 7, 2019
Messages
58
Likes
96
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,602
Likes
10,771
Location
Prague
The proposed load is a challenging load for most amps.
Not a realistic load.

The proposed load is so simple that it may be easily modified. Just take MicroCap or LTSpice and make consequently R1, R2 or C1 as a parameter and modify impedance as you like.
Step 2 - put the load into simulator and again play with parameters, especially with C1. You may get something like this

amp_sq_sim_R-RC_Cparam.png

You may find range of capacitances for which the amp is stable, and you may find another range where it oscillates like a hell.

Step 3 - tuned amplifier to be verified in a real life test with the complex load with parameters acc to simulation.

Last note - it does not make much sense to argue with input square short rise time, because input signal is not necessarily the only starter of instability. It may be output coupled EMI as well or mains line coupled interference. Square test with steep edges verifies if the amp is stable or not, especially when working into various complex impedance loads.

BTW, I had a preamp that time after time had suddenly higher distortion up to audible level. Long time I was not able to tell what is happening. GHz spectrum analyzer gave me the reply.

1575988873071.png


250MHz oscillations started usually with an impulse into output, like RCA cable connect. Now tell how you find it with the AP.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
16,052
Likes
36,427
Location
The Neitherlands
Do you feel one can only call a design succesfull if it passes this particular load ?
Is the load representative of the vast majority of speakers out there and/or worst case ?

Of course one can play with the various values one can even add some and find out with which values an amp really becomes unstable but will this say something about the same amplifier in practice with other amps ?

The only thing I am really weary about is those very high capacitance LS cables that are supposed to have 8 Ohm 'impedance' and require an extra inductor at the output and speakers with really low impedances.

btw... I don't have an AP. When I measure amps I use a 1GHz TDS680B scope but don't do much with speaker amps.

I do like to use realistic loads, though admit that it is interesting and fun to torture amps but don't want to blow them up while trying to determine its limits.
I do prefer bandwidth limited designs. 1MHz is more than sufficient for an amplifier.
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,895
Likes
16,715
Location
Monument, CO
Seems like the same old pissing contest all over again.

Many, probably most, ESLs exhibit <1 ohm (sometimes far less) impedance at 20 kHz. I don't think throwing out an entire class of speakers as "pathological" is reasonable. A number of other speakers have low-impedance HF dips, such as ribbon tweeters, that are in the 1-3 ohm range. In the real world the power is low enough that most amplifiers can handle them without issue. I don't think anyone is talking about slapping a 0.5-ohm 60(plus)-degree load at 1 kHz on the amp.

Amplifiers must have much greater than 20 kHz bandwidth to achieve the kind of flat in-band response we are seeing so applying a fast edge to check stability does not seem unreasonable to me. Run the numbers for first- or second-order response having 0.1 or 0.5 dB flatness to 20 kHz (about 125 kHz for 0.1 dB and first-order response). I always put an audio amplifier on a wideband 'scope or spectrum analyzer to verify it wasn't oscillating at some very high frequency (now we'd have to ignore the SMPS switching spurs unless they were unbounded, natch). Again, my source was typically a 1 kHz square wave with 1 us or so edge (transition, rise/fall time) so I could look out to 10~20 kHz and with a little bit of HF energy to "tickle" the amp. Having the wider frequency source just ensured the designer did his job at input/output filtering and the feedback loops didn't do anything funky.

The arguments against such testing seem to be primarily (a) no digital source has greater bandwidth than 20 kHz and (b) no amplifier has been found (in how long?) that has stability issues so why bother. For the former, analog sources such as LPs have well over 20 kHz bandwidth, and not all digital sources are CDs, so there is certainly content out there with >20 kHz bandwidth. For the latter, if we stop testing for it, we'll never know if its a problem unless something bad happens, and I prefer to keep the manufacturers honest. A pathological example would be the rise in "dead" diseases because people quit getting vaccinations for them.

I have no desire to pummel an amplifier with a full-scale 20 kHz or whatever square wave but checking with a lower-frequency version at a lower level (where bandwidth is typically broader) into some sort of fairly common load seems reasonable to me. I understand Amir does not have the time and interest for such testing, and these days it seems unlikely to uncover issues, but I am very surprised at the vitriol the mere idea generates. Probably difference in backgrounds; my career has virtually always focused on worst-case conditions and performance at extremes to ensure the utmost in stability and reliability in normal operation.

As such, I'll try to stay out of this mess, since my background isn't really relevant (at least not since my audio design experience as a DIY guy many years ago).
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,511
Likes
25,348
Location
Alfred, NY
Many, probably most, ESLs exhibit <1 ohm (sometimes far less) impedance at 20 kHz. I don't think throwing out an entire class of speakers as "pathological" is reasonable.

We aren't throwing out "an entire class of speakers." Well engineered ESLs (like Quads or Acoustats, probably the most popular ESL brands ever) don't show that behavior.

FWIW, my first exposure to a class D amp was at Daan V's apartment, with Quads being driven by one of Bruno's early models. Sounded terrific.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,895
Likes
16,715
Location
Monument, CO
Engineering critiques aside, the example was a current midrange model Martin Logan ESL, probably the most popular current ESL in the US (no idea about other markets). No, I don't know the numbers, but advertising and posts on various audio fora indicate a lot of Martin Logan speakers are in use, with Quad and others hardly ever mentioned. Maybe those other owners (Quad, Acoustat - defunct now, Sanders, Soundlabs, etc.) are in the majority and simply don't post? Whatever, I have no desire to continue to roll in the mud on this issue.

AGAIN, for me this has nothing to do with class D, or A, AB(1/2), C, E, F, G, H, I, just basic amplifier testing.
 

ajawamnet

Active Member
Joined
Aug 9, 2019
Messages
288
Likes
460
The proposed load is so simple that it may be easily modified. Just take MicroCap or LTSpice and make consequently R1, R2 or C1 as a parameter and modify impedance as you like.
Step 2 - put the load into simulator and again play with parameters, especially with C1. You may get something like this

View attachment 42039
You may find range of capacitances for which the amp is stable, and you may find another range where it oscillates like a hell.

Step 3 - tuned amplifier to be verified in a real life test with the complex load with parameters acc to simulation.

Last note - it does not make much sense to argue with input square short rise time, because input signal is not necessarily the only starter of instability. It may be output coupled EMI as well or mains line coupled interference. Square test with steep edges verifies if the amp is stable or not, especially when working into various complex impedance loads.

BTW, I had a preamp that time after time had suddenly higher distortion up to audible level. Long time I was not able to tell what is happening. GHz spectrum analyzer gave me the reply.

View attachment 42043

250MHz oscillations started usually with an impulse into output, like RCA cable connect. Now tell how you find it with the AP.

How were you coupling that Agilent to the circuit? H/E probe, antenna, or direct measurement? Getting harmonic spikes up to 750MHz (not sure what's going on at 950 - looks like an intentional radiator nearby with modulation) from audio gear is quite impressive. Typically in the preamps we design, we limit that using various filters and GBW of the opamps we use is a bit more limited. To get any considerable power at that frequency usually requires a current feedback amp.

I had to design a pulse amp and used the THS3122 CFA
http://www.ti.com/general/docs/suppproductinfo.tsp?distId=10&gotoUrl=http://www.ti.com/lit/gpn/ths3125

The bandwidth was around 125MHz - and yea it was driving square waves (needed a 20ns rise time max) into a 50 ohm load thru shit 22ga (non-coax) wire. I mentioned having to do this to Gary Stanfill - the former pres. of engineering at Vega Wireless (the original wireless mic company; I work with him at C-AT) and all he stated was "Good luck with that!"

So here's a plot of a Fourier series at 250 terms - from a simulation I had to do:

fourier-250n.jpg



Dr. Eric Bogatin, in an article found in the Dec 2004 issue of PCD&M (Rise Times and Bandwidths), states that the relation of a signal’s rise time to the circuit’s required bandwidth is roughly:

RT= 0.35/BW

Where:
RT is the rise time in seconds
BW is the sin/cos bandwidth

Recall what I was trying to do was a rise time 50,000 times that (see bottom dimension)... The 20nS performance dictated that a very high slew rate amplifier (~500V/µs) with an absolute minimum GBW > 17.5 MHz

National Instrument’s website, states that the components of a system should contain a minimum of 5 times the -3dB bandwidth calculated for a 2% increase in rise time. 87.5MHz...

The THS3122 is:
– 128-MHz , –3-dB BW (R L = 50 Ω, R F = 470 Ω)
– 1550-V/µs Slew Rate (G = 2, R L = 50Ω )
– 26-V PP Output Voltage, R L = 50 Ω

Gets hot as hell and requires some silly bypassing on the bipolar power rails to get it to work...

BTW: here's an actual Smith Chart of the load I ran this mess into:

vnplot.jpg


This was not fun...
 

JimB

Addicted to Fun and Learning
Forum Donor
Joined
Sep 19, 2019
Messages
731
Likes
493
Location
California
below the 10kHz squarewave of the 1ET400A

index.php


I see peaking but no Gibbs stuff. Mostly both in class-D and AB / A I only see either peaking or smoothing of squarewaves or oscillations and not the Gibbs stuff as amplifiers usually do not filter as steep.
8 ohm load?
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,716
Likes
38,889
Location
Gold Coast, Queensland, Australia
I've had many commercial amplifiers which were unstable on my bench. Some inexpensive, some expensive tweeter-cooking, high performance, ultra low distortion amplifiers from the past where hunting down the problems proved to be fruitless as the designs themselves were flawed. ie you had to slow them down to stop oscillations/instability or change output devices. Much time was spent scratching my head trying to work out exactly what the designers had in mind. Usually it was pressure to have better numbers on their brochure than their competition at the expense of ultimate stability.

It was the direct result of the specmanship games of the 70s and 80s, where bandwidths were pushed and THD numbers were the headline acts. Now we are in the same situation, with power outputs and THD specs (how many zeroes have we got today) as key selling points. Trouble is, they are quoting one without respect to the other- a clear violation of FTC amplifier rules (which incidentally are still in force) Feedback is wound right up again, compensation networks are numerous, an onboard PIC/uP monitors and controls everything, with tentacles into parts of the amplifier in order keep the thing stable. All talk of difficult/reactive loads is banished as it doesn't fit the core message, which is "trust us, we know what we are doing".

My old analogue oscillator sits on my bench as it has done for over 35 years. It's on all day until I turn off the power. Pretty much the first test on amplifiers and preamplifiers is feeding sines and squares through them at various frequencies to determine channel related imbalances/responses, low end coupling cap/feedback network issues, tone control functionality and high frequency stability/performance. Virtually every functional part of a typical integrated amplifier can be quickly and visually confirmed with just an analog oscillator and a 'scope. The sine/square switch is instant.

As Don mentioned, it's the vitriol that spews forth like a foul-smelling geyser when decades old established amplifier testing procedures are dismissed as being unimportant, that I do not like one bit. I'm not screwing pre-built off the shelf modules into a box, I'm chasing down mysteries with each and every item I restore/rebuild/repair. I do it because it's my passion, not to make money, as that's not really important to me.

Allowing commercially driven narratives and thought policing is not a good look. In fact, it serves to gradually and slowly poison the entire premise of ASR. In my opinion, there is already creeping evidence of it in the review threads.

Amir can't be bothered with extensive amplifier testing- that is quite clear. He sees little 'value' to the membership. Fair enough, he's only got so much time in retirement ;), but that doesn't mean we can't test and characterize amplifiers for others to see and comment on, without being shouted down.

And just remember, this entire thread is posted in the DIY forum, not the Desperate Dealers forum for anyone who hadn't noticed.
 
Last edited:

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,833
Likes
9,573
Location
Europe
[..]
I do like to use realistic loads, though admit that it is interesting and fun to torture amps but don't want to blow them up while trying to determine its limits.
I'm so happy now that I don't have to care any longer about stability problems of power amps with difficult loads.:p

You can all do as well - buy active speakers :cool: and there are no difficult loads any longer, just the bare chassis. And if a designer of an active speaker decides to use a difficult load like an electrostatic panel then it's his job to design a matching power amp.
 
Last edited:

ajawamnet

Active Member
Joined
Aug 9, 2019
Messages
288
Likes
460
I recall having a Polytone 101 bass amp in my early twenties.
s-l1600.jpg




I was in a band and had just gotten a large Randall, so at rehearsal I let the guitarist use the Polytone with one of his Rockman's (thing that Tom Scholz from Boston designed - https://en.wikipedia.org/wiki/Rockman_(amplifier) )
Amplifier4.jpg

So we're playing Rush 2112. It gets to the one lead at about 3 minutes in.

He kicks in this setting and it gets really funky sounding - actually quite close to the album. But it quickly deteriorates into some even more funky sounding stuff - then cuts out.

We stop...

So I look over at my Polytone ... I noticed an orange ring on the driver thru the grille cloth.

I was thinking, "... huh, I don't remember the dust cap on the 15" having orange glue..." and suddenly it erupts with flame through said grille cloth.

The drummer's brother goes to get a garden hose while I unplug it.

Turns out the amp didn't dig the crap coming out of that Rockman... I put one on a scope and whoa... really strange looking waveforms. The amp turned into a 110 watt oscillator and just toasted that voice coil.

Ahhh ... to be a kid again.
 

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,321
Location
Albany Western Australia
Seems like the same old pissing contest all over again.

Many, probably most, ESLs exhibit <1 ohm (sometimes far less) impedance at 20 kHz. I don't think throwing out an entire class of speakers as "pathological" is reasonable. .

No pissing contest Don, just a simple technical point which John decided to take exception to and go off on one because he read into the text what he wanted to and not what was actually there.

It was NOT saying don't test with square waves, it was NOT denying usefulness for assessing stability. It was NOT an argument against testing for stability.

It was merely a point about the effect of bandwidth limiting anf fiktering, which was relevant to all amps but specifically pertinent to @SIY who was going to be testing a class d amp on the weekend.

Are you sure that's correct about most being sub 1 ohm. It's not been on the searches I have performed.
 
Last edited:

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,321
Location
Albany Western Australia
We aren't throwing out "an entire class of speakers." Well engineered ESLs (like Quads or Acoustats, probably the most popular ESL brands ever) don't show that behavior.

FWIW, my first exposure to a class D amp was at Daan V's apartment, with Quads being driven by one of Bruno's early models. Sounded terrific.

No we're not, it's about throwing out bad engineering where it exists. Going sub 1 ohm is bad engineering.

This isn't a fundamental characteristic of ESL speakers as the Quads etc demonstrate, it's a result of design choices. Bad ones.
 
Last edited:

ajawamnet

Active Member
Joined
Aug 9, 2019
Messages
288
Likes
460

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,716
Likes
38,889
Location
Gold Coast, Queensland, Australia
I do recall having some Acoustat and Magnepans in the show room at Opus One... The Acoustat's would just toast the Kenwood L-07M amps. Fixed a lot of those... MKII's If I recall.

Absolutely. I've had so many of those amplifiers through my hands it's amazing. The Mk2s have 600KHz bandwidth and turn into oscillators in their old age (compensation cap issues). In fact, I just helped a friend with my last set of 6 original outputs. I sold my last pair about 5 years ago. Had 9 at one point (mix of 07s and 07mk2s).

And they are unconditionally stable with any resistive dummy load you can throw at them. Current limiting is clean. Difficult load however...

This was my last pair, I kept for a most of a decade in storage, then decided they need to go to someone who would use them. Had the original unused 12M directional (!) cables and the 12M trigger cables too.

lo7 john (1).jpeg


lo7 john (2).jpeg


Here is six of them at my place back in 2007. The pair above is the pair on the left.

1576025309111.png

The outputs in the mk1.
1576025390911.png
 
Last edited:

ajawamnet

Active Member
Joined
Aug 9, 2019
Messages
288
Likes
460
Absolutely. I've had so many of those amplifiers through my hands it's amazing. The Mk2s have 600KHz bandwidth and turn into oscillators in their old age (compensation cap issues). In fact, I just helped a friend with my last set of 6 original outputs. I sold my last pair about 5 years ago. Had 9 at one point (mix of 07s and 07mk2s).

And they are unconditionally stable with any resistive dummy load you can throw at them. Current limiting is clean. Difficult load however...
What was amazing is thru all that we had to do very little repairs to the Acoustats themselves. Not so with the damage the Plasmatronics did to the Acoustats... fixed a lot of those the few weeks we had the Plasmatronics on the floor...

The rating was 50kHz is what this state - https://www.hifiengine.com/manual_library/kenwood/l-07m.shtml
But I do recall seeing those go into the low band AM radio ...


Another interesting story I posted a while ago on Steve's forum:

"Interesting story about a McIntosh amp and the FAA. When I was a broadcast engineer in Pittsburgh , I did some work for Comsult. They had the contract for an old sister AM station of ours that got sold to a local church. The owner of Comsult told me about and issue he ran into when we were doing intermod (IM) studies for another FM he worked for. Any FM that wants to move within a certain distance of am airport has to do a thorough IM study if it's carrier is above 100MHz since it can PIM with other stations nearby and go into aircraft bands...

Intermodulation - Wikipedia

So at the time he was also working for a station in the north side of Pittsburgh, across the Allegheny river from "dahntahn" (Pgheese for "downtown"). He gets to the tower site and there's a bunch of FAA guys. One of the approaches for Greater Pitt Airport is over the city, and the pilots mentioned hearing his station on their comm gear - I think it was around around 119 MHz. The thought was that something was IM'ing with the station carrier and causing it. So they pull out their directional sniffer and lo and behold it's actually coming from the studio's down the hill. Thinking maybe something was weird with the studio to transmitter link (STL - typ ~940MHz on a paraflector) they head there. Typical STL systems : Radio Studio Transmitter Link

Nope, the signal is actually coming from the amp for the air monitor in the program director's office

There's some great info here:
Oscillation in tube amps - diyAudio

"
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,511
Likes
25,348
Location
Alfred, NY
Were those the amps that Kenwood made which used remote sensing for the feedback? It was a clever idea in theory. In theory.

I first heard Acoustats with Acoustat amps. When I got my 1+1s, I ran them with an old Adcom (worked great) and a couple of home-made tube amps (worked great at lower volumes). I did a short proof of concept for a direct drive system, but the panels came to an unfortunate end. Not from the amp, but from a child and cats. If I ever find a decent deal on replacement panels, I'll try it again now that the child is grown and the cats left with my ex.:D
 
Top Bottom