1. Please take 30 seconds to register your free account to remove most ads, post topics, make friends, earn reward points at our store, and more!  

MOSFET vs. Class D

Discussion in 'Amps and Cabs [BG]' started by terbay, May 5, 2005.


  1. 44me

    44me

    Jun 17, 2002
    Bedford, NH USA
    A conventional supply is...
    Line frequency (50/60Hz) transformer, rectifier, & filter capacitor(s) – develops an unregulated supply voltage, typically used to supply amplifier power stage

    A Linear Supply is...
    Same as above with addition of a linear regulator – functions like a linear power amplifier set to deliver a fixed, DC output voltage, typically used to supply preamp and driver circuits

    A switching supply is....
    A power supply that eliminates the line frequency transformer and develops high voltage supply (typically 325V) by directly rectifying and filtering AC input voltage. Then uses switches (typically MOSFET’s) to chop 325V at high frequency (typically 100kHz – 400kHz) to drive a high frequency transformer, rectifier, and inductor/capacitor filter. Output voltage is almost always sensed and fed back to circuit that chops the high voltage (called a modulator) to adjust the pulse width (PWM) to keep the output voltage constant.

    ‘Linear supply’ is sometimes used to indicate line frequency transformer, rectifier, and filter cap, but to a power supply designer, a linear regulator is also implied, which is not typically used to supply power amps.

    - John
     
  2. 44me

    44me

    Jun 17, 2002
    Bedford, NH USA
    Jerrold, all good points – I completely agree. It all goes back the fact that switching power supply designs are much more difficult to implement. Regarding regulation, you’re right, well regulated isn’t always best. Just ask any guitar player. Tube rectifiers degrade power supply performance but yield a sonic improvement to overdriven tones.

    - John
     
  3. Bob Lee (QSC)

    Bob Lee (QSC) In case you missed it, I work for QSC Audio! Commercial User

    Jul 3, 2001
    Costa Mesa, Calif.
    Technical Communications Developer, QSC Audio
    That's the thing. Look at the power lost in the transformer of a conventional (which is non-linear, BTW) supply versus the power lost in the transformer of a good switch-mode supply. They're quite different. It takes a relatively large core and relatively large number of wire turns for a mains-frequency transformer to pass a certain amount of power. Those wires in the primary and secondary windings, then, have to be fairly long, and unless transformer size isn't a consideration, they must be of a small enough gauge to fit. The high-frequency transformer in the SMPS, though, not only uses a smaller core, but many fewer turns of wire. The result is that the wire can be significantly larger in cross-sectional area and much, much shorter. This reduces the resistance of the windings, and therefore the I^2R losses, by several orders of magnitude. Thus, the potential bottleneck that you rightly warn of is actually not a bottleneck at all, but in fact a lower-resistance channel for current if it's done right.
     
  4. Jerrold Tiers

    Jerrold Tiers

    Nov 14, 2003
    St Louis
    Quite true...the bottleneck may not be resistance in an SMPS, so much as it is instantaneous max current limits.

    There are a number of components inserted in the path which are not in a "conventional" * power supply. Some have "absolute" current limits, more "real" than those a rectifier has.
    Protective circuits must be set to some limit, and therein may lie the bottleneck.

    However, various "resistive" effects occur at high frequencies which are negligible at low frequencies. So resistance still plays a part at high powers. This is especially true in units where the magnetic components may be designed for less than a 100% duty cycle (quite common).

    It is worth also noting that some types of SMPS have what may be accurately called a "current slew rate". The output current cannot change faster than a certain rate. Care must be taken by the designer to be sure that won't be a problem for the audio load. Audio amps have a number of characteristics which are a disadvantage to an SMPS.

    *
    Ok, OK, "linear" is misleading, wrong, whatever....but it is general usage, much as "rms power"....another wrong description
     
  5. MJ5150

    MJ5150 Moderator Staff Member Supporting Member

    Apr 12, 2001
    Olympia, WA
    I agree with Tom....this is a most interesting conversation. I would also like to thank Bob, Jerrold, and 44me for allowing us to "listen in" on this discussion. The only missing is Gary from EA. He is smart like you guys.

    -Mike
     
  6. Check out the FAQ section on the EA site. It's actually an answer to a question I posed to them a while ago. I always had a confusion about what constituted a 'digital' amp. A number of people use 'Class D' and digital interchangeably. Digital is a much looser 'marketing' term. A 'digital' amp can either be an analog amp with a switching power supply (e.g., Walter Woods) or a 'class D' amp (digital) with a traditional analog power supply (e.g., iAmp800). I think the Acoustic Image guys use a class D amp and a switching power supply, but I'm not sure. Anyway, there are distinct advantages/disadvantages to each design. The analog amp/switching power supply design of the Walter results in a very small and lightweight amp (due to no analog power transformer) that is very warm sounding (due to the analog amp), but rather thin on the low end (due to the switching power supply not being able to handle the huge power requirements of loud, low notes. The class D amp with analog power supply (iAmp) results in a much larger amp due to the need for a traditional power supply, etc. that runs very cool (due to the class D design). The low end response is great due to the traditional power supply, but some people find the class D sound somewhat sterile.

    I'm not an electical engineer, so this is a layman's explanation. However, I've spent a lot of time through the years talking with Walter and Gary (from EA). So, I believe the above is basically accurate.
     
  7. Jerrold Tiers

    Jerrold Tiers

    Nov 14, 2003
    St Louis
    Walter's first "little red brick" amp (about 130W IIRC) was an SMPS with an analog amp.

    But these days, he does make Class-D amps with SMPS.

    In fact, he has a patent on a particular way of configuring the Class-D and SMPS to work together. You can look it up on the US government patent site.

    www.uspto.gov/patft/index.html

    (USPTO=US Patent and Trademark Office)
     
  8. Thanks for the updated info. I didn't realize he switched to class D amps also. Very cool. I
     
  9. 44me

    44me

    Jun 17, 2002
    Bedford, NH USA
    A few more points on this subject worth considering –

    Though switching amplifiers and switching power supplies have been around for a long time, their use in audio applications is still relatively immature compared to traditional approaches. I would be very hesitant to assign sonic characteristics to a switching amplifier or an amplifier with a switching power supply based on a small sampling of production gear. Properly (or overly) designed, one should be able to make the sonic impact of the power supply transparent. It really comes down to cost, size, and weight constraints.

    Things are more complicated with a switching amplifier, as there are many different ways to control (modulate) the output transistors. A class D (PWM) amplifier will have different characteristics than a naturally sampled amplifier. As time goes on, expect to see true digitally controlled amplifiers that use numerical methods to not only generate the modulation directly, but also correct for non-ideal output stage behavior. It’s likely that the wide variety of control methods will make it difficult to impossible to assign particular sonic characteristics to switching amplifiers as more designs see production.


    As an aside, MOSFET’s have become dominant over bipolar transistors for both performance and economic reasons. MOSFET’s are generally much more rugged devices, making protection simpler while reducing the amount of over-design required for reliable operation. MOSFET’s are also significantly better for switching applications, which is where most industrial/commercial/automotive applications use them. This has driven the development of new parts, each with better performance for less cost. Bipolars have seen much less new development, and in fact, many manufacturers have gotten out of the business. The sonic difference between bipolar’s and MOSFET’s is much more subtle than either compared with tubes. This is due to both the devices themselves as well as how they are used.

    - John
     
  10.  
  11. B String

    B String Supporting Member

    Apr 11, 2002
    Los Angeles
    Not to seem stupid, but I would ask. If my Woods or Focus
    is a bit thin or shy on the low end, can I help remedy this
    with adding a bit more bass frequency in the signal chain?
    (like an eq pedal boosting only the lowest frequencies)
     
  12. 44me

    44me

    Jun 17, 2002
    Bedford, NH USA
    Assuming it’s thin, but you haven’t run out of headroom, a little EQ could make a big difference.

    - John
     
  13. boogiebass

    boogiebass

    Aug 16, 2000

    Small, yes. Thin on the low-end, not at all, IME.
     
  14. tombowlus

    tombowlus If it sounds good, it is good Gold Supporting Member

    Apr 3, 2003
    North central Ohio
    Editor-in-Chief, Bass Gear Magazine
    I agree with boogiebass. The WWU has a full-sounding low end, though to my ears it is not quite as tight or defined as that from my iAMP 800 (which, in turn, cannot match the silky, airy highs of the WWU).
     

  15. I think we are probably just down to personal preference, since I thought my iAmp800 had a much more airy highs versus the Walter!!! Even EQ'ing more bass with the Walter still leaves things a little weak below 50hz or so... but that's just to my ears. I guess that's what makes the world go round, and why all these manufacturers stay in business!!!! I've enjoyed this thread and learned some new things. :)
     
  16. tombowlus

    tombowlus If it sounds good, it is good Gold Supporting Member

    Apr 3, 2003
    North central Ohio
    Editor-in-Chief, Bass Gear Magazine
    I guess to clarify my comments somewhat, I would also say that the iAMP 800 has more low end (in addition to tighter lows) compared to the WWU. But then again, the iAMP 800 has exceptionally full lows. The WWU is not deficient in the low end, to my ears, but rather the iAMP 800 is strong in that category. With the amps set flat, though, I certainly get more airy high end out of my Woods. But again, personal preference and related gear certainly makes a big difference!

    Tom.
     
  17. B String

    B String Supporting Member

    Apr 11, 2002
    Los Angeles
    Right on the old sonic button Mr. Tombo This is exactly why I
    couldn't buy either one. When I used them next to each other
    on gigs, the differences were so obvious, that I just couldn't
    pick one. The EA lows were "funkalishous". The highs on the
    Woods were a "complex breath of spring damn it". I ended up
    with a Focus and just use different preamps depending on the
    type of gig and bass. Not as good as the other two, but its
    flat, light, powerful, less expensive.
     
  18. barebones

    barebones Supporting Member

    Jan 3, 2005
    Denver, CO
    Did anyone happen to catch my previous question? I know it's not up to the tech guru talk currently taking place, but it'd be interesting to know what you all know about the major bass amp brands out there. Thanks again.

    Matt
     
  19. 44me

    44me

    Jun 17, 2002
    Bedford, NH USA
    Matt, I was hoping someone with a good amount of repair experience would comment on your question. I know a little about particular amps I’ve considered getting There’s also a few generalizations based on design practicality I can make.

    Virtually all linear bass amps will be biased class AB. Class B causes crossover distortion (a discontinuity from when one device (transistor, MOSFET or tube) turns off and another turns on as the waveform goes through zero). This is a non-musical distortion. Class A is the opposite end of the spectrum, with neither sourcing or sinking devices every completely turned off. This eliminates crossover distortion entirely, but creates a lot of heat. You’ll see it used in low power, tube guitar amps and some high end audio gear. It’s generally too inefficient for bass or PA applications. Class AB runs class A for small outputs, and becomes class B for large. The transition between operating modes is smooth, so the effect is greatly reduced crossover distortion. Class C is for radio circuits with tuned loads, and is not useful for audio at all.

    I don’t know which amplifier was the first to use MOSFET’s commercially, but MOSFET’s started to become a cost effective alternative to bipolar transistors in the mid-late 80’s.

    Eden uses bipolar transistors for their higher end gear and MOSFET’s for the Nemesis line. I’m pretty sure Thunderfunk also uses bipolar’s. Mesa Boogie uses MOSFET’s, as does Aguilar (DB750), and Ampeg (‘B’ series).

    Of course using tubes for the preamp and/or driver stages greatly alters the tone characteristics and complicates things quite a bit. This brief list of manufactures represents some very different approaches. Eden uses a tube in the preamp, Tunderfunk and Ampeg (B series) use all SS, Mesa Boogie uses tube preamp and driver stage, and Aguilar uses an all tube preamp (not sure what they use for driver).

    I think it’s safe to assume that all switching amplifiers use MOSFET’s for the output switches. Common control schemes that I’ve seen people using for audio are class D (PWM), naturally sampled, and class T (Tripath chipset). Within these broad categories there are a lot of tweaks and optimizations that a designer can utilize to alter a switching amplifier’s behavior.

    I’m pretty sure EA uses class D with a conventional supply. Focus uses both a switching supply and switching amplifier. I’m pretty sure the same is true for WW (don’t know what kind of control method either uses).

    - John
     
  20. Just a note on this - Class D amps have been integrated into low power audio chips for quite a while. Just recently I saw a Texas Instruments multichannel audio chip for home theatre apps that claims 5 ch x 30 watts and 1 ch x 50 watts. So that's over 200 Watts (probably peak) on a single chip with all the control built in. I imagine it gets pretty hot to the touch though.:eek:
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.