GP_phan, whether or not an amp is ostensibly "class A" is probably the last thing you should be worrying about. True Class A operation is extremely inefficient, and while it works well -- it's expensive because of its inefficiency. Which means that the term "class A" becomes a very abused one in audio nomenclature -- kinda like "digital". That is, it seems like everybody wants to say that their amp is CLASS A in big type, and then put in little type how it's not really, actually a true class A amp.
But specifically for your situation. The MC2105 is a very finely built amplifier and will last many years, but it's Achilles heel is not that it's Class B . . . but it has a quasi-complementary (all-NPN) output stage. The reason for this is that when these amplifiers were designed (in the late-1960s), complementary NPN/PNP power transistor pairs simply didn't exist.
What this means is that in your amplifier, the basic linearity of the output stage around the crossover area is quite poor. In addition, the driver stage is really primitive, with simple resistor networks setting the current through the input differential-amp. I'm not faulting McIntosh for these design choices -- they were simply doing the best they could with the materials and experience they had.
I think a good upgrade for you might be to get a later McIntosh amp, from at least the MC2255-era (early-1980s) or later. By this time, they were using fully-complementary output stages, input stages with current-sources and current-mirrors, and voltage amplifiers with an active load -- meaning they're an order of magnitude more linear. And if you like your 2105, you'll LOVE a 2255.
The C28 is a mixed bag -- its circuitry is very much like a tube preamp, but with transistors . . . so it's fully Class A, single-ended operation. Its weaknesses are that the high impedances used (like a tube preamp) make it a bit noisy, and it's pretty complicated and failure-prone in interconnection and grounding. Specific common failures are dried-out main filter caps (causing oscillation), and that four-section volume control gets noisy and has channel-balance problems when it gets older. You might get it serviced, but if yours seems to be working well, then I'd probably upgrade your amp first, and keep the C28 for a little while - and eventually upgrade the C28 with a late-model unit like a C36 or C38. |
Kirkus, I'm having a very hard understanding why an all-NPN output stage is very non-linear at the x-over point. Please explain. Thanks. In a typical all-NPN output stage, the drivers are complementary (NPN/PNP), but the outputs are not (NPN/NPN). In order to make the bias voltages work out, the transistors are connected in such a way so that for the positive pair, there are actually two local feedback loops - one around the driver, and one around the output transistor. But for the negative side, the driver and output are in a single local feedback loop together. This means that in the crossover region, there is an abrupt change in the static gain as the current is transferred from one half of the output stage to the other. If you ever play around with adjusting the bias on such an output stage (I've actually modded many MC2105s to have adjustable bias), it can be easily seen that as the operating point is shifted, the crossover distortion never goes away, it simply changes shape a bit. There doesn't seem to be anything primitive about this input stage except that it is (very) low gain since resistors are used as the load to input differential pair. The following voltage amp is a transresistance amplifier (current in, voltage out) - meaning that the signal doesn't appear as a voltage on the collector of the diff-amp transistor that drives it . . . rather, it's a current. So in the diff-amp, it's the transconductance (set by the transistors' beta and their emitter resistors) that determines its open-loop gain. And, how do "input stages with current-sources and current-mirrors, and voltage amplifiers with an active load" make the power amp more "linear"? Using a current source to supply the diff-amp tail current effectively "makes the tail longer", and eliminates the variation in tail current with common-mode voltage. Also, the performance of any diff-amp is VERY dependent on the static balance of current between the two transistors (or tubes) . . . and with resistors setting the current, this critical balance is affected by tolerances in the (carbon!) resistors, and variations in beta and Vbe in the (old!) input transistors themselves, and also the line voltage when using an unregulated supply (as is almost always the case). Using an active current mirror, instead of the resistors, forces the static currents to remain balanced regardless of these variations. As far as the active load on the voltage amp - the improvement is the same as for any traditional single-ended collector/plate-loaded voltage amp -- that is, by keeping the quescient current constant across the required voltage swing, the amplifier is more linear. And here is one place where increasing the impedance on the collector does indeed increase open-loop gain, which also reduces distortion. All of this makes the amplifier much more linear in the traditional sense -- its transfer function. And if you ever compare a MC2105 to a MC2255 (an old amp, but which has these refinements) on the test bench - they are worlds apart in the amount of distortion they produce . . . especially in the higher harmonics. |
This may be a more general answer than you're looking for, but I like audio components that both sound good and measure well . . . and do so consistently, under a wide variety of conditions, for many years, without requiring much maintainance or repair. I generally feel that measurements are a good indication as to whether the engineering and execution are competent, and good sound is the indication that the basic design (and measurement methodology) is also competent.
To that end (to consistently sound good and measure well), I think there are some things that the best audio products have in common:
1. They are very well suited to the application
2. They make effective use of current technology and financial resources
3. They effectively solve the problems inherent in their basic design choices
4. They have quality and consistency in their execution
There are many topological choices that can fit this criteria well, both tube and transistor . . . just like there are so many different types of amplifiers that both sound good and measure well.
As far as Class A vs. Class B vs. Class AB . . . it's not that I don't think it doesn't matter, it's just that its very difficult to get a true class A design to meet criteria #1 and #2 above. For most solid-state Class AB or Class B amps, the main difficulty comes in meeting #3. For most Class AB or Class B tube amps, the difficulty comes on criteria #3 and #4.
I personally feel that these days, bipolar solid-state amplifiers show the most promise to meet all the criteria similarly well, but regardless of the basic choice of topology, it's a difficult road to get there. If I was designing a solid-state bipolar amplifier to drive a typical high-end domestic passive loudspeaker, I would choose a Class B design. If I was designing a tweeter amplifier for a high-end active loudspeaker, I would choose a class A design. If I was designing a tube amp, I would choose a Class AB design. |
I am so much at awe with what true Class A amplifiers can do to my life that I just want people to check them out. Spatine, I really don't disagree with you in any fundamental way - it's just that for me, I could say this about a number of different classifications of audio equipment, not just Class A amplifiers. And I do indeed value a rational, disciplined approach to the craft of designing and building audio equipment over a dogmatic approach . . . mainly because I consistently have positive musical experiences from systems that don't fit into any particular belief system. A well-executed Class A amp can indeed show the true potential of solid-state amplification . . . I simply feel that it's possible these days for Class B amplification to be just as good (and better in some ways) if properly applied. It's just unfortunate that the market is brimming with too many examples to the contrary. |
Kijanki, let's look at it like this . . . and this is kinda what I was getting at when I talked about an amplifier being "well suited to the application".
A realistic amount of juice necessary to drive most of the domestic audiophile loudspeakers to comfortably high domestic volumes is, say 25 volts RMS . . . (approx. 75W/8 ohms or 150W/4 Ohms). I feel that a high-quality amplifier must have a low output impedance in order to have predictable performance into the types of loudspeakers that are likely to get connected to it. If we then to honestly call it a "true Class A" amplifier, it must maintain class A operation to the lowest realistic impedance that the amplifier will see -- let's say 3 ohms.
Our class A amplifier will then operate from 20 volt rails, have a quescient current through the output stage of 13.5 amps, and thus dissipate at least 1100 watts of heat for two channels (you did want stereo, no?). If we want this amplifier to be reliable, consistent, and last many years (because it's hard to enjoy the amazing sound quality of a broken amplifier), we need to keep temperature rise to a minimum (do we include the gentleman who lives in an un-airconditioned flat in Singapore in our calculations?), and we of course don't want a fan . . . it's plain to see the bar-tab for transformer and heat-sink is getting pretty hefty.
When I say a Class B amp could be "better in some ways", I'm thinking about all the other things I could, as an amplifier designer, spend the customer's money on to improve the sound, other than a brute-force approach to linearizing the output stage. If I'm clever enough to get similar performance results from from a Class B output stage (yes I know this is a significant challange), I can definately build a better amp for a given amount of resources.
A great comparison is Levinson ML-2s and the ML-3. I've had both amps in my system for a bit, and enjoyed the latter much more . . . the ML-2s were like a high-maintainance chick that looked gorgeous on your arm, but didn't know how to have fun when you got home and turned the lights out. |
Actually, not a bad analogy, Kijanki. It's just a question of . . . when do you want the bad news? To deliver more than you promise is usually more exciting than the other way 'round.
But with most semiconductor amplifier output stages, the linearity problem around the crossover point in Class B is substantially less severe than the one that occurs at the transition between Class A and Class B operation on an AB amp. If class B operation is thoroughly and rigourously optimized, then its performance will be less dependent on load and output level (which I think is very important), but it doesn't give true Class A performance under any conditions. Again, the particular application makes all the difference. |
In all my previous discussion about linearity, I have been talking about the performance of JUST the output stage in isolation. And whether or not the output stage is biased as Class A, Class AB, or Class B . . . has NO effect on TIM.
The "TIM" acronym these days seems to be frequently flung about as a method to justify virtually any school of thought in amplifier design. But if we to talk about Transient Intermoduation Distortion as described by Matti Otala in his early-1970s AES paper -- the main thrust of this paper (and concept) revolves around frequency-compensation techniques. IIRC, Otala was basically proposing alternatives to the ubiquitous Miller compensation around the voltage-amplifier stage, under the supposition that lag compensation would reduce the loading of the differential amplifier under hard-slewing (transient) conditions, thus reducing a major source of nonlinearity.
While I have great respect for Otala's work, there are a few reasons that I feel the TIM concept, as he described it, is long overdue to be put to rest:
- The specific techniques he describes were based on observations in an era when power transistors were extremely slow, even compared to the small-signal stages that preceeded it - meaning that correctly-applied Miller compensation can be far less heavy-handed in the context of modern power semiconductors.
- Reducing the open-loop gain has absolutely no effect on the fundamental mechanism that causes the problem, it just forces the amplifier to work under conditions where it's difficult to occur. Kinda like strapping yourself to a sofa to avoid having foot pain that occurs when you stand up.
- A far more useful method of analysis of TIM is as a conditional reduction of large-signal open-loop gain and phase margin. This predicts the increase in distortion, and explains why a blanket reduction in open-loop gain can reduce the effect. It also gives valuable insight into how to solve the fundamental issue.
- This also lets us look at TIM distoriton for what it truly is -- a stability problem.
There are vastly more resources available today to more fully understand and predict the actual open-loop behavior of an amplifier than in Otala's day (i.e. high-bandwidth DSOs, FFT spectral analysis, and SPICE). This means that the rigourous engineer can more thoroughly investigate and anticipate all stability issues (including TIM), if he/she chooses to do so.
And if they don't, then TIM is the least of our worries. |
The term "gm doubling" as I understand it refers to the increase in gain that occurs from when one half of the output stage is conducting, to when both halves are conducting. The idea is that there's "double the transistors", so there's "double the transconductance". The static gain is of course no where near double because of the local feedback intrinsic in the output stage.
"Gm doubling" is the main problem with a Class AB amplifier, and manifests itself as a sharp nonlinearity at the point of transition between Class A and Class B operation. This can easily be seen when looking at the distortion output on a 'scope - as the signal is increased into a resistive load, the nonlinearity on the distortion residual slides, in alignment, from the tips of the waveform toward the center, as the amplifier transitions from Class A to Class B. And you're correct in the assertion that increasing bias doesn't improve the problem, it just increases the signal level at which it occurs.
Class A amplifiers don't have this problem, because "gm doubling" occurs all the time - since both halves are always conducting, there's no change in gain to the point where they're not. Class B amplifiers must walk a thin line between cutoff (underbiased) and gm-doubling (overbiased) - this is why bias tracking is so critical on class B designs. |
The magnitude of TIM is highly dependent on the open loop gain (everything else being equal) up to point where output transistors go to momentary saturation and stay there for a moment (having charge trapped at the junction). No, I think you're getting a couple of concepts confused. Saturation of the output transistors happens at clipping or at reverse bias, the latter of which being a point where the charge carriers are accelerated maximally away from the transistor junction. Whether or not this happens is indeed a function of output stage slewing, but is completely an open-loop phenomonon and occurs independently of loop gain. The amplifier need not have feedback (actually it doesn't need small-signals stages at all!) for it to occur. If for some reason the designer wishes to never reverse-bias the output transistors, this is easily acheived by making minor changes in the driver connections - and the result is a slightly slower output stage. The concept of slew rate limiting that Otala discusses in his seminal paper on TIM is related to the charging and discharging of the capacitor(s) used to set the small-signal bandwidth of the input and voltage-amp stages, and thus the open-loop response corner frequency. Since these small-signal stages are always biased Class A, their slewing performance (and ultimately that of the whole amplifier) is dependent on the quescient current flowing through them (as used to charge/discharge the capacitors), not the open-loop gain (which BTW I'm assuming means the o/l gain below the corner frequency). Otala advocated the use of capacitors in different places (lag compensation), which basically simply changes which stage in the amplifier is responsible for their charge/discharge current. Both Otala's method and the conventional approach have their pluses and minuses . . . and both approaches can be much less drastic with modern semiconductors than the ones available when he wrote the paper. My point is that TIM can be understood, analyzed, and avoided - and we don't need to go down the "is feedback good or bad?" road to do it. The latter is of course an unsolveable debate at this time (so let's not go there). The biggest point to me about THD, IMD, and TIM is not so much is not what the numbers themselves are -- but what's causing it, and what the best ways are to fix the problems. |
High slew rate input signals come back to summing junction thru negative feedback delayed because of signal path delays. For a moment amplifier has no feedback and overshoot appears at the output (or earlier dependent on design). This will happen to any amplifier if slew rate is not properly limited at the input. The only part of this I disagree with is the phrase "at the input". And yes, these are the very fundamentals of proper frequency compensation. The only nit-pick I would add is if this is indeed done "at the input", that implies a passive network before amplification . . . and while slew-rate and bandwidth are two different things, you can't limit one without the other in a passive network. Which makes it a bandwidth discussion, and again takes us right back to the fundamentals of stability in feedback amplifiers, and phase margin. Also, a Class AB amp cannot "exhibit higher order of mostly odd harmonics at very low signal levels" as compared to a Class A design . . . very simply because at "very low signal levels", it's a Class A amp. That's its raison d'etre. I could maybe see how other non-crossover nonlinearities (this "kink" you describe) may have been a slight contributor at very specific power levels in days of output devices like 2N3055/2955, but is virtually nonexistant in properly loaded constant-beta modern bipolar power transistors. But to answer Oem's question . . . I don't think that Class B is "the answer", it's one of many valid options. Why is it not used more? I don't know for sure, but I'd speculate that the main reasons are because it requires extremely accurate thermal bias compensation, and the nonlinear base currents demanded by the drivers from the voltage-amp . . . both present significant (but not insurmountable) design challanges for good results. Class AB is significantly less critical in these regards, but the trade-off is a greater variation in performance with load and signal level, both of which are dynamically working to pull the amp from Class A operation towards Class B. And this transition isn't a particularly graceful one. |
In transformer-coupled tube amps, the physical mechanisms that cause crossover distortion are very differnt from a direct-coupled bipolar transistor amplifier. In a push-pull amplifier (like both the Marantz and Mac), the concern is hysteris distortion in the output transformer as the current is transferred from one side of the primary to the other. The specific design subtleties in making a quality output transformer are way over my head . . . but suffice it to say that there's a great number of details that can be tweaked to make it work it's best in the particular circuit.
There's also the differnce that the final stage of most solid-state amplifiers is a current amplifier, and the output stage of a tube amp is a power amplifier (both a current and a voltage amp) . . . and tetrodes and pentodes have extra grids . . . so the number of various feedback configurations available increases quite a bit, all of which affect how the amplifier behaves around the crossover point.
True Class B tube amps do seem to be, as a group, less linear than their counterparts running class AB, the McIntosh amps being a bit of an exception to that (they're allmost class B), but the McIntosh "Unithy Coupled" circuit has much more local feedback than most designs, and very high-quality transformers. The Marantz transformers are also very high quality, but designed very differently for the "Ultralinear" circuit. The Model 2 is pretty straightforward Class AB1.
I'm a huge fan of the MC275 in any vintage . . . and the re-issues are excellent, probably better than the original . . . and they're reliable, and available new with a warranty. Also IMO one that's very fairly priced - these were all reasons that I chose it to pair with the existing Marantzes in the system you mention. I just feel that the Model 2 is a big step better than even the MC275, and would have preferred to four Model 2s . . . or even eight of them with a fully active crossover.
But that would be like opening a whole case of 1940s-era Inglenook cabernet . . . which would be awesome, and I tip my hat to those who can afford to do such things. But we didn't have that kind of budget, so I put a 1999 Caymus next to the vintage stuff we already had. |
Kijanki, we're definately running into some terminological inconsistencies in this field - but if we're talking about very low signal levels, I'd say that the amp in question must certainly be a Class B amp that for marketing reasons was being labelled as AB. If it occurred at higher power levels (say above a few watts), then I could see how a Class AB amp could exhibit this behavior.
100mA is probably about right for a single pair of bipolar transistors biased Class B across some pretty low emitter resistors, maybe 0.15 ohm? But I think of 0.22, 0.33, and 0.47 ohm as being the common values, so I'd say that most amps are biased proportonately lower for Class B.
It does seem quaint to use an all-NPN output stage these days . . . maybe for an inexpensive paging-system amplifier it might make sense, but for hi-fi, your're right. Crazy. |
You now have distortion only when the output signal is very high and you have no crossover distortion (GM doubling or whatever you like to call it - lets say transition distortion) when output signal is very low and it runs in Class A (both sides conducting). This is quite correct, but the gm-doubling transition distortion is much worse than the crossover distortion. So as to audibility . . . it all depends on the application - each road has its burdens to bear. A small absolute amount of distortion on a large signal is better than the same absolute distortion on a small signal. I've heard it asserted that crossover distortion manifests itself (it drives THD upward) more as the signal level is reduced . . . and honestly, I'm not sure whether or not it's true. It seems to make intuitive sense, but I've measured lots of amplifiers, and I'm doubtful as to whether or not the measured data supports it. Complicating the issue is that THD+N of course rises in a linear manner with a reduction of signal level . . . but when you measure just the noise, you get the same results. I think it may be that as the signal levels are reduced, the proportion of the total signal that's in the crossover region increases, but the crossover non-linearities are at the same time being spread out across a larger proportion of the waveform, making them less severe. Whether or not these opposing factors cancel each out is the question, and I certainly haven't the skill to investigate it with pure mathematics, and my current measurement equipment isn't sensitive enough to find the answer emperically. |
Spatine, I share your respect for Nelson Pass and his products, even though they're not my cup of tea. But since his amps are generally single-ended designs with a high output impedance, I would say that the "class-A-ness" isn't necessarily the main source of their sonic characteristics, and the advantages/disadvantages of these aspects of his designs require separate consideration.
That's the main reason why I chose the Levinson ML-2/ML-3 for comparison . . . where the circuit designs are much more similar, whereby the Class A/Class AB distinction is THE big difference. |
I'll second the recommendation on Self's work . . . he also has a couple of excellent books on the subject. I don't necessarily agree with all of his approaches and conclusions, but everything he presents he does with wonderful concise logic, and thoroughly cites all his sources - which seems to be rare in the audio field these days. He's definately one of the modern masters of audio amplifier design, and a delight to read as well. |
