XPA-1L Gen2 and 4 Ohm speakers in Class A

[jdaddabbo]

Well, first off I would LOVE to be WRONG on both accounts: XPA-1L and XPA-1. However, I received my information directly from Lonnie himself. For which I only "questioned" what I had been hearing as for the Class A outputs of these two amps after reading a paper put out on the topic of Amplifiers "Bias" into Class A. OK, so if I am "right" it was well worth asking again. And if I am "wrong", then no real harm from having gotten this re-confirmed, plus I could pass the word on to several others that believe as I do.

[jdaddabbo]

I wonder if things have simply changed since back when I got my answer... this was pre-Gen2 (so pre-XPA-1 Gen2 with its 60W Class A, as well pre-XPA-1L Gen2). Wouldn't that be wonderful. Not a better time to be wrong/out-of-date if you ask me.

[Gary Cook]

High School level science, Ohms laws apply, there's no magic at work.
The power P in watts (W) is equal to the voltage V in volts (V) times the current I in amps (A)
The power P in watts (W) is equal to the squared voltage V in volts (V) divided by the resistance in Ohms(Ω)

Typically a Class AB amplifier will supply double the power (watts) into half the speaker impedance (ohms). Note the published spec for an XPA-1L;
Power output (all channels driven):
250 watts @ 8 ohm (0.1% THD)
500 watts @ 4 ohm (0.1% THD)

But the Class A spec is different;
35 watts @ 8 ohm; Class A operation
Note: In Class A mode, under normal conditions, the XPA-1L operates in pure Class A mode at all power levels up to 35 watts, above this level it will automatically transition into Class A/B mode.

Normally we would expect the XPA-1L to put out 70 watts into 4 ohms. But in Class A it supplies 35 watts into 4 ohms and 35 watts into 8 ohms, this can be achieved by varying the voltage/current depending on the load (impedance). I suspect that the 35 watt limit is ultimately determined by the heatsink capacity, hence why Lonnie included load (impedance) dependence to vary the voltage/current.

Cheers
Gary

[jdaddabbo]

OK, so I found the response that I got from Emotiva and it dates back not only pre-Gen2 anything, but also pre-the actual release of the XPA-1L Gen1. I was asking/clarifying that the XPA-1L Gen1 would in fact NOT double down in Class A as it was doing in Class A/B. I got an Answer of NO it would not be able to due to the obvious 4 times the current which would be needed. I then proceeded to Ask what I would be getting in Class A at 4 ohms. He indicated that he would need to ask Lonnie. The response I then got (supposedly from Lonnie) was that due to design limitations the power output in Class A at 4 ohms would be HALF. Sooooo, I took the published 35W and divided by 2. However, reading what you all are stating now... I believe that what was intended to be communicated to me (from Lonnie) was that it would only be HALF, as in NOT doubled-down and therefore only 35W ☺. This is wonderful news for this means that I have an actual 60W Class A into 4 ohms with my XPA-1s! Very good news indeed. Thank you so much for working with me on this. I very much appreciate it!!!

[jdaddabbo]

You all could have just gotten mad at me and called me names (or worse ?), but instead none of you did (which I too often see on Boards when someone merely speaks that which he "knows" to be True). Thanks again!

[garbulky]

Mar 3, 2015 18:58:19 GMT -5 jdaddabbo said:

You all could have just gotten mad at me and called me names (or worse ?), but instead none of you did (which I too often see on Boards when someone merely speaks that which he "knows" to be True). Thanks again!

ya big ninny!!

[deltadube]

Mar 3, 2015 18:58:19 GMT -5 jdaddabbo said:

You all could have just gotten mad at me and called me names (or worse ?), but instead none of you did (which I too often see on Boards when someone merely speaks that which he "knows" to be True). Thanks again!

communication break down!!!!!!

[guzz46]

According to this review, the XPA-1 gen 2 transitions to Class B at 60 Watts into 8 Ohms and 30 Watts into 4 Ohms

www.hometheaterhifi.com/power-amplifiers/power-amplifiers-reviews/emotiva-xpa-1-gen-2-monoblock-power-amplifier-review/page-4-on-the-bench.html

[jdaddabbo]

Well, I don't feel so bad now Question is, where did he get his info from? It doesn't appear to be a measurement, but rather just a statement.

[KeithL]

Everybody's right

I hate to drift off into philosophy.... but ; if I set the switch on my XPA-1L to Class A and leave it there, and then feed in a signal so as to produce a 10 watt output, the XPA-1L "is running in Class A"; if I then increase the signal level so the output is now 100 watts, the XPA-1L "is now running in Class A/B"; even though no switch or relay has changed position, and none of the steady state characteristics of the XPA-1L has changed; what has changed is merely the relationship between the output power and the bias current.

The difference between operating in Class A and operating in Class A/B is the amount of bias current being drawn by the output stage. The industry standard way of rating Class A amplifiers is to base the rating on this value - and to do the calculations based on the "normal speaker impedance". In the case of the XPA-1L, the amplifier is "35 watts per channel of Class A" - which is based on the power its output stage dissipates in Class A, and is independent of the impedance of the speaker you have connected (thus the same "Class A rating" into 4 or 8 ohms). Technically speaking, since the bias current is independent of the speaker impedance, the amount of bias applied to the XPA-1L in Class A defines it as "operating in 35 watts of Class A mode per channel", and this amount of bias current allows the XPA-1L to deliver 35 watts per channel to a set of 8 ohm speakers while operating in pure Class A mode, but will only allow it to deliver 17.5 watts per channel to a set of 4 ohm speakers while operating in pure Class A mode.

There are two more things you need to be aware of to put this into proper perspective:

1) Speaker impedance ratings are at best approximations - a "typical 8 ohm speaker" may well vary between as little as two or three ohms and as much as 50 ohms or more, depending on what frequency you measure it at (halve those numbers for a 4 ohm speaker). This means that your "4 ohm speaker" may well be 4 ohms at 50 Hz, 13 ohms at 100 Hz, 3 ohms at 1 kHz, and 2 ohms at 20 kHz. (So doing math treating it like a 4 ohm resistor, which is the same at all frequencies, is sort of silly anyway.)

2) Because the XPA-1L is always Class A/B, there is no point at which either set of output devices stops conducting, and so no crossover notch distortion - ever. In "regular Class A/B mode" the bias current, which is the least amount of current each set of output devices conducts at any point in the signal cycle, is still non-zero, but is relatively low, and so may be reached with relatively low signal levels. This eliminates the crossover notch, but might still theoretically produce a tiny amount of distortion when that point is crossed.  In "Class A mode" the bias current, which is the least amount of current each set of output devices conducts at any point in the signal cycle, is much higher, which makes it very UNLIKELY to occur anywhere except during a signal peak or a very loud spot in the music, and which makes it much less likely to be audible.

As a side note, it's becoming clear to me that a lot of people still don't understand the whole point of Class A, and what it actually means.

In a Class B amplifier, at any given instant, only the output devices connected to one of the supply rails conduct current to the speaker (either + or - but not both). Since this avoids having both devices conduct at once, which would be wasteful, Class B amplifiers are very efficient. However, since this implies a point (at zero volts) where one transistor "switches off" and the other "switches on", it results in a small discontinuity in the waveform - which is generally referred to as "the crossover notch". Since this notch occurs at the point where the signal is at a minimum, and is more or less independent of signal level, it tends to be audibly unpleasant sounding. (This is why you don't see many Class B amps lately.)

With a pure full Class A amplifier, in direct contrast, when the amp is delivering zero output power, the output devices connected to both supply rails conduct a significant amount of current. Then, as the signal swings positive or negative, the transistors connected to one rail conduct more and those connected to the other rail conduct less. What makes it Class A is the fact that, as the current conducted by one set of output devices increases, the amount conducted by the other set decreases by an equal amount, and vice versa. This avoids the rough spot where either set of devices switches on or off, and you either end up with "no transition" or "one big smooth transition" - depending on how you choose to look at it. In either case, the nonlinearities in the output devices tend to cancel out, so you get a very clean signal.

In a normal Class A/B amplifier, a small amount of bias current continuously flows through both sets of output devices. As long as the current required to produce the output signal remains less than this bias current, the amplifier IS operating in Class A. Once the current required to produce the output signal exceeds this bias current for part of its cycle, one of the sets of output devices continues to conduct more current, while the other set continues to conduct only its bias current for that part of the cycle. (The bias current sets a minimum amount of current that either set of output devices will ever conduct, and this current prevents either set of output devices from ever switching all the way off, and so prevents that nasty crossover notch.)

NOTE THAT WHETHER THE AMPLIFIER IS "IN CLASS A" OR "IN CLASS A/B" IS SIMPLY A MATTER OF WHETHER THE SIGNAL LEVEL EXCEEDS THE BIAS CURRENT AT A GIVEN INSTANT. AND SO DEPENDS ON SIGNAL LEVEL. There is no actual switch involved, and the amount of bias current also doesn't change. It is strictly the RELATIONSHIP between the bias level and the signal level that determines the "operating class" of the amplifier.

When you set the switch on the XPA-1L into Class A mode, it will be drawing 35 watts per channel of Class A bias (which, again, is the traditional way of rating Class A amplifiers). This will enable it to deliver up to 35 watts of power into an 8 ohm load while still operating "fully in Class A mode" and will enable it to deliver about 17.5 watts of power into a 4 ohm load while still operating "fully in Class A mode".

There is also an important distinction going on here that most people seem to miss. If you have an XPA-1L, with it's switch set to "Class A", connected to 8 ohm speakers, running at 120 watts per channel, it can be said "to be operating in Class A/B", because it has reached a point where, for part of the signal's cycle, the current through one set of output devices continues to increase, while the current through the other set has stopped decreasing, and remains at the minimum set by the bias current. HOWEVER, it can also be equally accurately said to STILL be operating in Class A mode up to 35 watts. (You could say that "all of the signal that's still below the 35 watt point is still being amplified in Class A - even though the part that exceeds 35 watts is now being handled in Class A/B".) Since distortion that occurs near the zero point of small signals is more audible, by raising the bias point in Class A mode, we've lowered the audible distortion at points where the signal is small - EVEN WHEN THE AVERAGE OR PEAK SIGNAL LEVEL IS HIGH ENOUGH THAT THE XPA-1L IS OPERATING IN CLASS A/B MODE SOME OF THE TIME.

We do apologize if this is confusing, but we're doing our best to avoid that...

We'd really honestly rather you based your judgment on how the thing SOUNDS than on any of this theoretical stuff

Mar 3, 2015 15:23:03 GMT -5 geebo said:

Would you mind letting us in on how you know this to be the case when Keith and I believe Lonnie also has said 35 watts into 4 or 8 ohms. Are they not familiar with their product's design parameters?

[monkumonku]

Question: what makes class A better than class A/B? Does it sound better? Is there an audible difference? If you can't tell when an amp switches from A to A/B then what difference does it make to have a class A amp?

Is the average person going to hear any difference between the two classes (this asked, realizing that the Lounge is 97% golden ears so that may be an irrelevant question)?

[audiosyndrome]

Keith- As is your usual, excellent description of Class A, Class A/B.

Russ

[KeithL]

Without getting ridiculously technical, there are two main ideas at work:

1) Transistors tend to be more linear in the middle of their operating ranges than at the ends... and they tend to be worst at the very ends of their range.

So if, for example, the signal requires that the output current of a certain transistor "go up one amp".....

* In a Class B amplifier, that transistor would go from 0 amps to 1 amp (and there would be a lot of distortion right where it turns on at 0 amps).

* In a Class A/B amplifier, that transistor would go from 0.1 amps to 1.1 amps (we've avoided that nasty spot around zero, but we're still changing the current over a factor of 10:1).

* In a Class A amplifier, that transistor would be going from, perhaps, 3.0 amps to 4.0 amps (which not only avoids that zero point, but is proportionally an even smaller percentage of its overall operating range).

2) If both transistors are changing at the same time, and the changes are symmetrical, then the distortion will also be symmetrical and, if so, we can get it to cancel out.

Now let's use the same example, except that we'll raise the output current to go from -1 amps to +1 amps....

In our example above:

* For Class B, we start with -1 amps going through the -rail transistor, and nothing going through the +rail transistor.
As we raise the voltage the -rail transistor conducts less and less, and eventually shuts off when we reach 0 amps (the +rail transistor stays at 0 amps).
Finally, as we keep raising the voltage, the +rail transistor starts conducting more and more, until it reaches +1 amp (but now the -rail transistor stays off).
Both transistors never conduct at the same time.

* Now, in our Class A amplifier, we may start with both the -rail and +rail transistors conducting 3 amps each.
As we raise the voltage, the current in the +rail transistor goes up to 3.5 amps, while the current in the -rail transistor drops to 2.5 amps.
Now, as we lower the voltage back to zero, we return to each passing 3 amps.
Finally, as we lower the voltage still more, eventually the -rail transistor ends up conducting 3.5 amps, while the +rail transistor conducts 2.5 amps.

Notice how, in the Class B amp, there was pretty much never a time when both transistors were on.

However, in our Class A amp, they were always both on, and, even better, they were "moving in opposite directions".
Therefore, we can expect that, assuming the transistors are "similar", the distortion produced by each will be "equal to and opposite of the other".
Further, because of the way things are connected, we can expect these "equal and opposite distortions" to cancel each other out.

In the Class A/B amp, which I didn't detail, the transistors are only both operating (and in opposite) over the region where the bias current causes both to conduct.
So, in the Class A/B design, this cancellation effect occurs at the critical point where the signal is tiny, but, in the Class A design, it occurs over the entire range of the signal.
(And, in our high bias Class A/B designs - like the XPA-1L - we have extended the range over which this occurs to be much larger than with a "normal" Class A/B design.)

Now, all of this logic is true "if all else is equal"...... In reality, that isn't always true. For example, the two complementary transistors don't produce distortion that is exactly equal and opposite; and the extra power dissipated by a Class A design makes the entire device run hotter, which requires that you get rid of more heat, and causes the circuitry itself to generate some additional noise and distortion simply because of the heat; and your power supply is going to be working harder to supply that extra heat that the Class A design requires. Excluding those possible qualifications, the end result of the cancellation effect is that Class A amplifiers (all else being equal) produce lower distortion overall, and especially low levels of certain types of distortion (types which cancel out especially well), compared to equivalent Class A/B units. The level of distortion is lower overall, and the distortion spectrum is different.

As it turns out, the distortion of the XPA-1L is very low in either mode; so low that, arguably, most people don't hear it either way. (I can't speak to "average", but I'll fess up that I don't hear the difference. The XPA-1L happens to be one of our lowest distortion amps either way.)

Note that, since the majority of music is typically at levels between 1/10 and 1/20 of the highest peak present, and those highest peaks only occur for a small percentage of the time, and we're a lot more sensitive to distortion that occurs during quiet spots than to distortion during crashing crescendos anyway, if you put that switch in the Class A position, the XPA-1L will be operating in Class A for the vast majority of the time where you would be likely to hear the difference... even if you like to play things loud and have inefficient speakers.    

Mar 4, 2015 14:11:03 GMT -5 monkumonku said:

Question: what makes class A better than class A/B? Does it sound better? Is there an audible difference? If you can't tell when an amp switches from A to A/B then what difference does it make to have a class A amp?

Is the average person going to hear any difference between the two classes (this asked, realizing that the Lounge is 97% golden ears so that may be an irrelevant question)?

[garbulky]

So 35 watts @ 8 ohms and 17.5 watts @ 4 ohms.

[jdaddabbo]

Yes, Thank You very much Keith!!! Glad to hear that I'm not a Big Ninny after all, and that how I understood it to be is how you have explained it. As well I so much agree with your final thought, and that is to forget about ALL of the Specs, ALL about of the Design considerations, and ALL about of that which is Stated by the Manufacturer... and instead judge any Audio gear by the way it sounds. And do my XPA-1s sound GOOD!!! (into my Legacy Audio Focus).

[Jim]

Mar 4, 2015 14:11:03 GMT -5 monkumonku said:

Question: what makes class A better than class A/B? Does it sound better? Is there an audible difference? If you can't tell when an amp switches from A to A/B then what difference does it make to have a class A amp?

Is the average person going to hear any difference between the two classes (this asked, realizing that the Lounge is 97% golden ears so that may be an irrelevant question)?

97%? That's funny.

It's 100%. Duhhhh.

[Gary Cook]

Mar 4, 2015 14:11:03 GMT -5 monkumonku said:

Question: what makes class A better than class A/B? Does it sound better? Is there an audible difference? If you can't tell when an amp switches from A to A/B then what difference does it make to have a class A amp?

Is the average person going to hear any difference between the two classes (this asked, realizing that the Lounge is 97% golden ears so that may be an irrelevant question)?

Keith gave, as usual, a very detailed answer in regards to the technical differences between Class A and Class AB. I learn something every time

Having had XPA-1L's in my system on and off for a couple of months, I've noticed some tiny subtle things. Keeping in mind that I had Class A (only) amps for some years, so I'm used to listening to/for their sound. No golden ears here.

It's not an "OMG it's in Class A" revelation. So it's very easy to imagine it's not there or equally imagine it is there. I can pick it some times with cymbals, right hand piano, violins and even harp. Sometimes not, it's very music dependant and even affected/masked by loud ambient noise. The Class A sound is more realistic to me, it has a metallic edge, it's sharp, hard edged, like a real cymbal does when you hear it live.

Cheers
Gary

[joto]

Hi I am using UPA 1L with Maggie 1.7 in class A and they sound great. You can definitely hear the difference between A and A/B operation on the high end when using my VPI Scout wit Sumiko blue point evo 3.

[Canuck_fr]

You must mean XPA 1L

[hcsunshine90]

hey, no one answered my question. since the XPA-1L puts out (in A/B) 250 watts into 8 ohms and 500 watts into 4 ohms...then why when in class A when it puts out 35 watts into 8 ohms it only puts out 17.5 watts into 4 ohms and not 70 watts (into 4 ohms)? since it can put out twice the amount (in A/B) into 4 ohms as it does into 8 ohms...then why when in A does in only put out half the amount into 4 ohms as it does into 8 ohms? I don't mind lookin like a dummy here. I want to know why this is.