XPA-1L Class A vs Class AB

[audiosanity]

Oct 6, 2013 12:20:03 GMT -5 leonski said:

Audio,
The EMO won't run any HOTTER...measured with a thermometer...probably.  You can only run transistors so-hot before they cook themselves into junk.
What you WILL get is much more HEAT.

You are right. I suppose they set a target max temperature and design a heat sink with a capacity commensurate with the dissipated power.

[leonski]

I think that's right.   The calculation includes number of output devices and bias current along with the characteristics of the transistor.  
Heatsink costs and lots of heatsink gets heavy and bulky for shipping purposes.  
No matter what.....I believe you have a maximum temp after which NO semiconductor will work properly.   Just for an example, I installed 
at FAN CONTROL program on my iMac the DAY I brought it home.  I try to keep it internally cool to maximize life.    Cooking a CPU or 
shortening its life with excess heat is an expensive proposition.

Keep it cool!

[dac0964]

I made a $10 cooling solution (DIY) with a computer fan (60 CFM) and an old AC/DC adapater for my 1Ls. I measured that without any fan (my amps are seating on top of an entertainment center) the temp just outside of the air vents was 116F. Very hot to touch. Then I put one fan (to suck out the hot air from the middle vents) and after 30 minutes the temp went down to around 98F (almost 18 degrees lower). I then added a second fan on top of the vents on the right and it went down another 10 degrees.

I also did not notice any SQ difference between A and A/B although I listened only for a few minutes in A/B. My speakers' sensitivity is 90db. However, like the others, I'm very happy with the work these amps are doing on my speakers.

[powxpa]

How do xpa 1ls in AB mode compare to xpa 100s in term of SQ and heat?

[leonski]

It is possible that you are effecting bias current with forced cooling.   
I would, if I owned the amps, measure idle current / power using a Kill-A-Watt plug in meter.  
Do so with and without forced cooling.  It is possible the amp knows....via thermaster, perhaps, the temp of some point or even a heatsink.
If the heatsink is too cool, bias goes UP to generate more heat to compensate.  

Heatsink temp is a good indicator of bias current.   Pass adjusts their amps for a certain temp rise under bench conditions.  Even removing the top cover
makes a difference and effects bias current.   At least ONE Pass owner reports the amp sounds better without the top cover...due in no small part to the
increase in bias current. 

[dac0964]

I can understand what you mean by affecting the bias current. But the sound (SQ) was the same, at least to my ear, with or without the cooling fans. I don't have a KAW meter to test it. Anyway, I'm planning to just use one fan to not affect the bias current that much. I just wanted to experiment and see the effect with using 2 fans. Either way, the amp still felt warm unlike a traditional A/B amp (like my UPA-200) that was always cool to the touch.

[leonski]

I wouldn't expect nite/day differences using the AB/A switch.  
ME?  I'd use the amp in 'A' only for music listening when I just sit and listen.  Otherwise?  AB it is.  

The meter I mention, the Kill-A-Watt, is Very Inexpensive.  I paid about 20$ at the local emporium.  They shouldn't be above 25$, even today.    I think it should be part of EVERYONEs kit.

The question is....how loud do you listen, how big is your listening space and how sensitive are your speakers?  A big room with insensitive speakers and for those who like it LOUD, will get you out of 'A' in a big hurry.  Real sensitive speakers....say over 93db/watt/meter in a small room at modest levels will use FAR less power.  

IMO a proper amp design will make 'telling' the class of operation nearly impossible.   Me?  Since I want to eventually experiment with a Line Level crossover.....which goes between a preamp and the amp, instead of the speaker level crossover, going to a set of 4 of the 1Ls makes sense in the long run.   And I suspect, given my listening habits and even WITH low sensitivity speakers, I may be able to stay pretty much at the 'A' end of the scale and not tap what would be nearly 1000 watts per speaker AB. 

The only way to tell if my specuation is right....about bias current and temp, would be to either MEASURE or have someone from the design team chime in and confirm or call me nuts.

[saru]

Sorry to revive an old thread, but has this issue been settled? I was just curious what the exactly the point of the A-to-A/B toggle is if the amps run in Class A for the first 35W anyway. If left in Class A mode, what happens if the power draw spikes above 35W? Do the speakers clip or something? Furthermore, the consensus seems to be that the amps sound virtually identical in either setting, so SQ doesn't appear to be a factor here. Is the only real difference that the amps put out less heat in A/B mode, or is there something that I'm missing? Seems to me that there shouldn't be a toggle, and that the "Class A for up to 35W, Class A/B for higher loads" should be sole operating status. Thoughts?

[boomzilla]

You'll have to listen for yourself. No, this has not "been settled." Some claim that the Class-A mode sounds better - I, personally, think that the Class A-B mode sounds more dynamic. If you're in Class A, I think you clip if you exceed 35 WPC. In Class AB, the amp transitions automatically from A to AB modes. I'm not sure where the transition occurs in AB mode, but I suspect it'll be at less than 35 watts.

And as always, I could be wrong...

Cheers - Boom

[leonski]

Apr 14, 2017 9:15:06 GMT -5 saru said:

Sorry to revive an old thread, but has this issue been settled? I was just curious what the exactly the point of the A-to-A/B toggle is if the amps run in Class A for the first 35W anyway. If left in Class A mode, what happens if the power draw spikes above 35W? Do the speakers clip or something? Furthermore, the consensus seems to be that the amps sound virtually identical in either setting, so SQ doesn't appear to be a factor here. Is the only real difference that the amps put out less heat in A/B mode, or is there something that I'm missing? Seems to me that there shouldn't be a toggle, and that the "Class A for up to 35W, Class A/B for higher loads" should be sole operating status. Thoughts?

The transition is NOT a 'brick wall', but rather more of a ramp.   The transition will be more of degree than a 'switch flip' in condition.  

Substantial power spike / demand will be in A/B, most likely.   Unless you run some very high sensitivity speakers.

[saru]

Oh, I just realized we are talking about two different things I don't mean the point at which the class switches from A to A/B when in A/B-mode (though that is interesting in of itself)... I meant it more like "why have this feature at all?". I'm having trouble envisioning a reason for running the amp in A-mode when A/B-mode does all that plus allows you to run higher power loads than 35WPC. And heating your house in the winter is not a valid reason .

[garbulky]

Apr 14, 2017 9:15:06 GMT -5 saru said:

Sorry to revive an old thread, but has this issue been settled? I was just curious what the exactly the point of the A-to-A/B toggle is if the amps run in Class A for the first 35W anyway. If left in Class A mode, what happens if the power draw spikes above 35W? Do the speakers clip or something? Furthermore, the consensus seems to be that the amps sound virtually identical in either setting, so SQ doesn't appear to be a factor here. Is the only real difference that the amps put out less heat in A/B mode, or is there something that I'm missing? Seems to me that there shouldn't be a toggle, and that the "Class A for up to 35W, Class A/B for higher loads" should be sole operating status. Thoughts?

Several misconceptions here.
First....Class AB is not 35 watts class A. It's much less than that.
The class A mode is 35 watts of class A power @ 8 ohms (which btw is 17.5 watts @ 4 ohms).

After getting past 35 watts there is no clipping.
What Emotiva has done is adjust the bias. So when the power draw passes 35 watts it switches automatically in to class AB mode and delivers its full rated power which i think is 250 watts for the XPA-1 L .
In class AB mode - there is a small amount of class A power too - much less than 35 watts - and when it exceeds the bias of that small amount, it switches in to class B mode.

You have to realize that with class B mode there is crossover notch distortion in every single electric wave. You can't get away from that. However improvements in class AB has made that cross over notch distortion much much lower to the point that many people claim it's inaudible. But it IS there. So we can't wish it away by trivializing it. It's there. It's small, but there.

Class A mode does away with cross over notch distortion present in every single sine wave. So if you are listening to 1 khz, you are getting 1000 of those distortions every second. In class A mode you are getting none of that.

Now this is likely somewhat exagerrated compared to well designed class AB amps but it illustrates cross over notch distortion



So that's the point of class A mode.

Class A mode is not necessary for good sound. However it also doesn't hurt good sound and some people (me included) feel there is a difference. It is very subtle.
But class A mode brings these negatives - lots of heat, lots of weight, usually in the form of heatsinks, and lots of power consumption.
Companies like Krell have humungous class A amps and they are terrifically heavy.
Audio GD's flagship amps are also heavily biased class A.

[saru]

Thanks for the information, garbulky. Very interesting stuff! So, let me see if I have this right:

Class A mode on the XPA-1L (The front toggle switch is flipped to the LEFT, toward "A"):

• 35 watts into an 8ohm load, you get Class A power as you described
• For loads higher than 35 watts...  Dunno what happens here.  From everything that I've read, the 1L only switches to A/B power when it's physically in A/B mode (see below).  Boomzilla says it should result in clipping, if I'm reading his post correctly.
• Runs really hot, even in idle.  I've got mine in A-mode right now and they are very toasty to the touch.  Glad I have them out in the open!

Class A/B mode on the XPA-1L (The front toggle switch is flipped to the RIGHT, toward "A/B"

• Some amount less than 35 watts into an 8ohm load for Class A power; so far, no one has been able to say how much it actually is.
• For loads higher than this unknown threshold, the 1L automatically shifts to A/B power and can deliver up to 250 watts into 8ohms.
• Runs cooler, but still very warm.

[garbulky]

Apr 14, 2017 16:06:43 GMT -5 saru said:

Thanks for the information, garbulky. Very interesting stuff! So, let me see if I have this right:

Class A mode on the XPA-1L (The front toggle switch is flipped to the LEFT, toward "A"):

• 35 watts into an 8ohm load, you get Class A power as you described
• For loads higher than 35 watts...  Dunno what happens here.  From everything that I've read, the 1L only switches to A/B power when it's physically in A/B mode (see below).  Boomzilla says it should result in clipping, if I'm reading his post correctly.
• Runs really hot, even in idle.  I've got mine in A-mode right now and they are very toasty to the touch.  Glad I have them out in the open!

Class A/B mode on the XPA-1L (The front toggle switch is flipped to the RIGHT, toward "A/B"

• Some amount less than 35 watts into an 8ohm load for Class A power; so far, no one has been able to say how much it actually is.
• For loads higher than this unknown threshold, the 1L automatically shifts to A/B power and can deliver up to 250 watts into 8ohms.
• Runs cooler, but still very warm.

In class A mode: After getting past 35 watts (class a mode)there is no clipping.
it delivers its full rated power which i think is 250 watts for the XPA-1 L .
B'zilla got it slightly wrong.
The reason amps don't have a lot of class A is that it requires a beefy amp with beefy components and heatsinks able to handle the heat and current.
In class A an amp is always consuming power as its transistors are always running.
So - expensive.

[KeithL]

You're sort of getting it - but you're not entirely getting it.

First, when we talk about "transitioning into Class A/B mode" we're talking about a sort of imaginary line - like your property line.
When the amplifier is operating in Class A, the transistors on both sides of the supply rail are both conducting for the entire audio signal.
When the amplifier is operating in Class A/B, each transistor is operating for more than half of the time, but not for all of it.
(In Class B, each transistor would be operating for exactly half of the time.)

So, for an XPA-1L, with an 8 ohm load, with the switch in "Class A mode"...
... for signals below 35 watts the amp will be operating in Class A
... for signals greater than 35 watts, it will be operating in Class A/B

And, for an XPA-1L, with an 8 ohm load, with the switch in "Class A/B mode"...
... for signals below a few watts the amp will be operating in Class A
... for signals greater than a few watts, it will be operating in Class A/B

The actual number where the amp ceases to be operating in Class A mode will vary depending on exactly how warm it is...
but in Class A/B mode, it will be between one and a few watts.... and quite a bit less than 35 watts.

"Whether the amplifier is operating in Class A or Class A/B" is determined by the level of the signal relative to the amount of bias.....
And, in either switch position, from 35 watts to its maximum full power the XPA-1L is operating in Class A/B.

Also note that, while Garbulky's explanation is quite concise and mostly correct, the pictures are a bit misleading.
In order to be visible on an oscilloscope trace, you need to have around 5% distortion or more.
The numbers we're talking about have several more zeros to the left, and so wouldn't be visible at all on an oscilloscope trace.

Another minor but important detail is that, in a Class B amplifier, the crossover notch occurs when the signal is crossing 0 volts.
This means that the distortion occurs when there is absolutely the least signal there to make it less audible.
It also means that, as the signal gets very small, the distortion is proportionately larger, and so more audible.
In contrast, with an amplifier operating in Class A/B, not only are the "notches" FAR smaller, but they occur at two different points, NOT at the zero crossing.
(So that picture showing crossover notches at 0 V is correct for Class B but not for Class A/B; in a Class A/B amp the tiny notches, if at all visible, occur away from the zero crossing.)
And, as the signal gets very small, the amplifier eventually starts operating in Class A, and the notches disappear entirely.

Just so there's no confusion....
AT SIGNAL POWER LEVELS BELOW THE TRANSITION POINT, A CLASS A/B AMPLIFIER IS A CLASS A AMPLIFIER, AND HAS NO CROSSOVER NOTCHES....
AT SIGNAL POWER LEVELS ABOVE THE TRANSITION POINT, THE NOTCHES ARE VERY SMALL, AND OCCUR AWAY FROM THE ZERO CROSSING POINT....

Apr 14, 2017 16:06:43 GMT -5 saru said:

Thanks for the information, garbulky. Very interesting stuff! So, let me see if I have this right:

Class A mode on the XPA-1L (The front toggle switch is flipped to the LEFT, toward "A"):

• 35 watts into an 8ohm load, you get Class A power as you described
• For loads higher than 35 watts...  Dunno what happens here.  From everything that I've read, the 1L only switches to A/B power when it's physically in A/B mode (see below).  Boomzilla says it should result in clipping, if I'm reading his post correctly.
• Runs really hot, even in idle.  I've got mine in A-mode right now and they are very toasty to the touch.  Glad I have them out in the open!

Class A/B mode on the XPA-1L (The front toggle switch is flipped to the RIGHT, toward "A/B"

• Some amount less than 35 watts into an 8ohm load for Class A power; so far, no one has been able to say how much it actually is.
• For loads higher than this unknown threshold, the 1L automatically shifts to A/B power and can deliver up to 250 watts into 8ohms.
• Runs cooler, but still very warm.

[KeithL]

Easy answer.....

Most people listen to music using from a few watts to 20 or 30 watts most of the time.
Therefore, in the Class A position, most people will be listening in Class A most of the time.
(And, arguably, distortion is also more critical when you're listening quietly.)

The distortion on the XPA-1L is very very low in either mode, but it's LOWER in Class A mode.
The actual spectrum of the distortion is also a little bit different.
Some people can hear the difference, and consider it to be significant.

You're getting the same maximum power in both modes....
Class A mode just makes a lot more heat - and maybe sounds a little better - when you're listening quietly.

Apr 14, 2017 15:17:31 GMT -5 saru said:

Oh, I just realized we are talking about two different things I don't mean the point at which the class switches from A to A/B when in A/B-mode (though that is interesting in of itself)... I meant it more like "why have this feature at all?". I'm having trouble envisioning a reason for running the amp in A-mode when A/B-mode does all that plus allows you to run higher power loads than 35WPC. And heating your house in the winter is not a valid reason .

[ehmokey]

Love the technical discussion and provides me learning opportunities. Having said that, I am unable to hear a difference in either A or AB mode with my 1L's. I love how these amps turned on my small and inefficient bookshelf speakers and make my music sound so good. I do run them on class A most of the time because I can. I do turn them completely off during the week when not in use but leave them on Stand-By and switched to class A/B during the weekends to save on electricity costs. I have listened to music in A/B mode by unintentionally forgetting to slide the switch to the left and I still enjoyed music just as much.

[garbulky]

On the brief time I listened to the XPA-1 L monoblocks, I did not discern a noticeable difference - not saying it's not there just that I didn't hear it.

However, I did notice a difference on the XPA-1 gen 2. It's very subtle though. At first I wasn't sure. For instance it's not a case of - oh it's clearer or I could hear this instrument better. It took me more time. It's more a case of the soundstage being more "connected" and the sound feeling more natural and more involving. I was able to hear the room dimensions captured on the recording a bit better.
How I noticed it was when I had the switch in class AB, I wasn't as engrossed in the sound. Put it in class A mode, and I felt more engrossed. I noticed this when I would glance at where the switch was during these different reactions I was having.

So...it's definitely subtle, you can't point out the difference readily....the XPA-1 gen 2 is a good amp either way, but the class A switch is a welcome addition.

[teaman]

I have never heard the XPA-1L but I have owned both generations of the XPA-1's and I could not decipher any audible difference in Class A vs. Class A/B. The only difference was in Class A the amp got a lot warmer.

[pedrocols]

Apr 15, 2017 10:48:52 GMT -5 teaman said:

I have never heard the XPA-1L but I have owned both generations of the XPA-1's and I could not decipher any audible difference in Class A vs. Class A/B. The only difference was in Class A the amp got a lot warmer.

I wouldn't be surprised if you are using no more than 5 or 10 watts with your Klipsch speakers. Hence the amps will always operate in class A so you wouldn't really listen the amp in class A/B. Every A/B amp has class A operation even if it is 1 watt.