Hi Chris,anatech said:
Imagine If you need to design a 3KW amp and for switching section how many of several MJ15024 / MJ15025 are needed, but if you go for mosfet only single IRFP260N would do the job much easily
Gold_xyz said:so the fig.3 is when the voltage lifter work in switching mode.
when the voltage overcomes the threshold the lift switch on
+/-Vss supply. I do this mode call H-class.
and in fig.4 is when the voltage lifter work in class C.
when the voltage overcomes the threshold the ago a linear
regulation from Vs up to Vss. I do this mode call G-class.
on many sites it seems to say the contrary one..
see also http://www.qscaudio.com/support/library/papers/amptalk.pdf
bye 🙂
Good illustrative description for the ones under the confusion.....
ZGarage said:
If you look at the waveform of that amp[SoundCraftsmen], it clearly says its a Class-H amp , not a Class-G or Class-TD Rail Tracking amp, because THe Amplitude of Step Voltage is constant, but the width changes accordingly, which clearly depicts the PWM using Switches....
But the literature is little misunterstood, waveform spell the Truth very Well...
K a n w a r
Hi Kanwar,
However, for industrial servos your point is well taken. A Carver amp, though powerful, is not an industrial appliance. My point was that the commutators were not limited by transistor switching speed at all, but rather some other constraint. Could be it sounded better with full range music the way they did it with time constants on the commutators. Your designs come at these problems from a different angle.
Have you tried the "down converter" idea? This might prove to be a better way of executing the same idea.
-Chris
I'm not going to bother imagining that. We can't draw that much power from our AC supply in normal 120 VAC outlets. 😉 My believe is that the power is better split up. So you would end up with several amps and speakers. I would also drive the mids and highs from their own source. When you design for such high powers, current becomes a very real problem. The losses become too high and efficiency drops to unacceptably low levels. That and loudspeaker drivers are more efficient with lower voice coil temperatures, another loss.
However, for industrial servos your point is well taken. A Carver amp, though powerful, is not an industrial appliance. My point was that the commutators were not limited by transistor switching speed at all, but rather some other constraint. Could be it sounded better with full range music the way they did it with time constants on the commutators. Your designs come at these problems from a different angle.
Have you tried the "down converter" idea? This might prove to be a better way of executing the same idea.
-Chris
Around the efficiency of class H
With sine wave (at max level) the normal AB-class have an
efficiency of Pi/4 and the max dissipation of end stage is on
a square wave with Vs/2 (but with efficiency of 50%)
In a single step class H amplifier the efficiency is determinate
in what voltage it switch the supply.
how i can calculate this ?
bye 🙂
With sine wave (at max level) the normal AB-class have an
efficiency of Pi/4 and the max dissipation of end stage is on
a square wave with Vs/2 (but with efficiency of 50%)
In a single step class H amplifier the efficiency is determinate
in what voltage it switch the supply.
how i can calculate this ?
bye 🙂
This came up before and I still think some of you are confusing the two. It makes no sense to me to have Class G and Class H basically look the same on the scope. Workhorse, your scope of Class H looks like Class G, jump rails as needed, 2 tier. Now if Class H is a rail tracker, it makes total sense. Class G is a good idea, change rails to reduce power dissipation unless needed. Someone probably said, well 3 rails is more efficient than 2 and heck 4 is even better so finally they said, why not an infinite amount of rails and hence Class H, the more improved Class G but different enough to be a separate class. Class TD doesn't even make sense. Class T is nothing but Class D but Tripath wanted to get attention and so call it Class T. Marketing class not a real class. So what would TD be? Class D supplying the voltage to a Class D? I think we call this SMPS with a Class D amplifier.
Audiolabs definition is correct. Class G is a rail switcher/changer so different rail levels as needed. Class H is a rail tracker, infinite tier Class G.
-SL
Audiolabs definition is correct. Class G is a rail switcher/changer so different rail levels as needed. Class H is a rail tracker, infinite tier Class G.
-SL
The Efficiency of Class-H with sine wave signal is exactly same if you drive the amp to its max voltage swing near clipping, Its efficiency is only considered when playing music, because music has certain Peak swing areas but most of the content is very much below the peak swing area[as much as 2 to 5 times dependng upon musical content]Also if you fed an input signal using sinewave and set the output voltage to the Half of max rail voltage yes then it has much higher effiecieny than other Class-Ab amp with same magnitude of rails voltage and output voltage equals 1/2 of rail voltage...as Class-H cuts the waste heat arisen from the much greater voltage drop accross class-AB..using rail switching...Gold_xyz said:Around the efficiency of class H
With sine wave (at max level) the normal AB-class have an
efficiency of Pi/4 and the max dissipation of end stage is on
a square wave with Vs/2 (but with efficiency of 50%)
In a single step class H amplifier the efficiency is determinate
in what voltage it switch the supply.
how i can calculate this ?
bye 🙂
Using Typical musical program content
Class-AB is only 40 to 50 % efficient
Class-H is 65 to 75 % efficient
SpittinLLama said:
Look at the waveforms posted by GOLD_XYZ they clearly illustrate the difference of Tier Switching ENVELOPE in Class-H and Class-G, if you still donot understand then its your own mental wisdom....
There is huge difference which could be seen....
Class-G has Linear Envelope whereas Class-H has Square Packet Envelope and also Class-G requires same number of Transistors in both Upper and Lower Tiers, on the otherhand Class-H uses only one set of transistors in lower tier and upper tier is managed by only one Mosfet typically...Both topologies have huge difference in working and efficiency...
Class-H lifts the amp rail in Switching phenomena, Class-G Lifts it linearly as demanded by the signal.....
Rail Tracker is not Class-H and Class-H is not Rail Tracker.....
Hi Chris,
The fact is that I couldnot think about Down-converter or Rail Tracking design because I already have 5KW Class-D module under development which would completely wipe out the complexities involved in Tracking Down COnverter amps with Class-D rails and Class-AB amp....Why not have alone Class-D amp, a much better Solution to go....
K a n w a r
Hi Kanwar,
I know you make good sounding amplifiers. I don't think you can equal that in class D unless they are sub amps. However, using a class D amp (basically) to supply the rails would allow you to finesse the output and save on heat. The best of both worlds.
Also, good luck on your 5KW design. I hope it's a winner for you. Are you going for sub duty or full range?
-Chris
I can think of one reason. Poor sound quality in full range. The normal output stage helps smooth out the rough spots. Mosfet or bipolar.
I know you make good sounding amplifiers. I don't think you can equal that in class D unless they are sub amps. However, using a class D amp (basically) to supply the rails would allow you to finesse the output and save on heat. The best of both worlds.
Also, good luck on your 5KW design. I hope it's a winner for you. Are you going for sub duty or full range?
-Chris
anatech said:
No, Class-D requires careful analysis to weedout the non-linearities presented by switching artifacts and it doesnot sound bad in fullrange also...Have you listened to UCD, NCD, COLDamp, TRIPATH all are Class-D and have a good repution among DIYers..
Mine is Fullrange version and it sound excellent even when at max power....
Hi Kanwar,
At this time I can do better in class AB. So that's what I use. Class H is a method of reducing wasted energy and class D would be another topic again. Your amp does sound interesting, high power too.
-Chris
At this time I can do better in class AB. So that's what I use. Class H is a method of reducing wasted energy and class D would be another topic again. Your amp does sound interesting, high power too.
-Chris
There was Time When I first developed N-channel Mosfet CLass-AB amp with 2000W/ch, and again its another different Time when i developed Class-H amp.....
Passage of Time always Demands Change....and its Phenomeanal in nature....
Passage of Time always Demands Change....and its Phenomeanal in nature....
I think there is a semantic problem and one switch between what some of us are saying is Class H and G. The figures shown by Gold_XYZ are how I understand the classes but G is the sqaure envelop and H is the linear one, not the other way around as some say. The linear envelop to me is also a rail tracker just tracks only down to the Vs supply level. This bottom level (+/-Vs in the Figs) and, therefore, how much of the rail is tracked is all design dependant. I actually made such a design as Fig 4 to test efficiency of the AB amp to see how much it improves. I cheated and used LM3886's as the supply rails driving another LM3886 but the results were pretty interesting. I used +/-15V as the Vs levels and tracked or linear envelop if you will, for levels above these rails. Fun experiment but not practical since all efficiency gain is more than lost in the linear supplies. But we also made a board using a SMPS to see how it performed and effieciency for the whole system (SMPS + AB amp) was very respectable. The LM3886 is not the best because PSRR is not great at higher frequencies but it all worked and ran way cooler.
I guess I misunderstood some of the other posts but Gold_XYZ figures show the two classes. Can argue some other time about which is G or which is H 😉
-SL
I guess I misunderstood some of the other posts but Gold_XYZ figures show the two classes. Can argue some other time about which is G or which is H 😉
-SL
SpittinLLama said:
I think you have ABSOLUTELY NO idea about the difference between Rail Tracking and Linear Envelope[Class-G] amp....May I suggest you to update your knowledge
Rail Tracking has No Tiers only one symmetric variable supply, Linear Envelope[Class-G] has 2 or 3 Tiers Supply Rails....
K a n w a r
"An oxymoron?"
I don't think so. Imagine D3 and D4 are gone. We now have a cascode output stage. With the diodes, we have a class AB amplifier that cascodes itself at higher voltages (so to speak).
I don't think so. Imagine D3 and D4 are gone. We now have a cascode output stage. With the diodes, we have a class AB amplifier that cascodes itself at higher voltages (so to speak).
Well as I do not pretend to be able to design amplifiers (SS that is...), I'm very much interested in this thread.
The picture and the few words explaining the view made a lot "clearer"
Where (wich book) can I find more info on the subject?
Zilog
The picture and the few words explaining the view made a lot "clearer"
Where (wich book) can I find more info on the subject?
Zilog
Hi djk,
Great diagram! I'm sure that will help people understand the concept better.
Hi Tarzan,
Read the Carver threads or look at some manuals. There is a lot of extra circuitry for compensation and protection in any full schematic.
-Chris
Great diagram! I'm sure that will help people understand the concept better.
Hi Tarzan,
Read the Carver threads or look at some manuals. There is a lot of extra circuitry for compensation and protection in any full schematic.
-Chris
Tarzan said:
At Douglas Self site (google it) or in the book from the same author, where this diagram is.
You will find references about the book at the site.
regards,
above efficiency switched rail
I have made some calculation with switched rail amplifier
With rail switch at Vs/2 i have around 84% efficiency 😱
I call this "rail switch" so there are not doubtful !!! 😀
the demonstration...
(I hope not to have made errors)
direct link at image...
http://img515.imageshack.us/img515/9292/efficenzajr3.gif
I have made some calculation with switched rail amplifier
An externally hosted image should be here but it was not working when we last tested it.
With rail switch at Vs/2 i have around 84% efficiency 😱
I call this "rail switch" so there are not doubtful !!! 😀
the demonstration...
(I hope not to have made errors)
An externally hosted image should be here but it was not working when we last tested it.
direct link at image...
http://img515.imageshack.us/img515/9292/efficenzajr3.gif
