Hey everyone!
I heard about the class H amplifier the other day, but i haven't been able to dig up any technical notes on how they work. Cany anyone point me in the right direction, or give me a short run-down? Thanks!
I heard about the class H amplifier the other day, but i haven't been able to dig up any technical notes on how they work. Cany anyone point me in the right direction, or give me a short run-down? Thanks!
class H = 2 or more fixed voltage rails on output devices to lower dissipation of them.
Have one: MA 5002 from Soundcraftsmen: 2 x 250 W with compared to Pass minuscule heat-sinks.
Kind of trick Bob Carver also uses except the higher frequency transformers Bob uses to lower the weight.
Have one: MA 5002 from Soundcraftsmen: 2 x 250 W with compared to Pass minuscule heat-sinks.
Kind of trick Bob Carver also uses except the higher frequency transformers Bob uses to lower the weight.
It's kind of an adaptation of A/B
One pair of transistors (NPN/PNP or Pmos/Nmos) runs voltages above/below say +/- 15V -> amp limit, while another pair of transistors runs the voltages of 0 -> +/- 15V. It's kind of like stacking two A/B stages on top of each other. In effect, it has the "quality" of A/B at low power (voltage) levels, with slightly more "crossover" distortion due to two additional crossover points. The benefits (my guess) are that you can use better quality, lower power transistors instead of cheap high-voltage & high-power.
For the record, Technics (Panasonic) receivers use this topology, labeling it "Class H+" - I have an older Technics 100W/ch stereo receiver... and that label began my pursuit on trying to find out what the heck Class H was. Overall, I think its a pretty neat concept, provided you can actually manage to bias all the Xisters correctly... (evil laugh) MUWWAAAA HA HA HA HAAAA!
Hope this Helps,
One pair of transistors (NPN/PNP or Pmos/Nmos) runs voltages above/below say +/- 15V -> amp limit, while another pair of transistors runs the voltages of 0 -> +/- 15V. It's kind of like stacking two A/B stages on top of each other. In effect, it has the "quality" of A/B at low power (voltage) levels, with slightly more "crossover" distortion due to two additional crossover points. The benefits (my guess) are that you can use better quality, lower power transistors instead of cheap high-voltage & high-power.
For the record, Technics (Panasonic) receivers use this topology, labeling it "Class H+" - I have an older Technics 100W/ch stereo receiver... and that label began my pursuit on trying to find out what the heck Class H was. Overall, I think its a pretty neat concept, provided you can actually manage to bias all the Xisters correctly... (evil laugh) MUWWAAAA HA HA HA HAAAA!
Hope this Helps,
Not sure the definitions are absolute.
G would have two pairs or rails, for example +/-20 volts and +/- 50 volts.
H has at least one more pair for example +/-25v, +/-50v and +/-75v.
The Carver cube of ca. 1980 used three rails but he never gave it a "G" or "H" designation. He had some other funny stuff going on inside the box as well.
Unless I'm missing something, the difference between G and H isn't one of basic principle.
These have had a reputation for being noisy. However, if you can get your hands on the dec '01, Jan & Feb '02 of the British mag "Electronics World" there is an article by Doug Self on this topic where he reports a G-class design that is very close to the best B amps. Basicly he used Spice analysis to identify the noise sources and devised ways around them. The final tweak was to bias the inner transistors (the ones using the 25v rails) so that they are in class A mode. He calls the result "A+C".
Back to Carver. The current Sunfire scheme is supossedly "none of the above". Appearently there some digital processing the holds the rails at 6v over the output voltage, thus the rails move with the output level. This is supossed to be so efficient that no heatsinks are needed.
I'm sure someone will say I've got it all wrong, but that's best understanding that I have.
G would have two pairs or rails, for example +/-20 volts and +/- 50 volts.
H has at least one more pair for example +/-25v, +/-50v and +/-75v.
The Carver cube of ca. 1980 used three rails but he never gave it a "G" or "H" designation. He had some other funny stuff going on inside the box as well.
Unless I'm missing something, the difference between G and H isn't one of basic principle.
These have had a reputation for being noisy. However, if you can get your hands on the dec '01, Jan & Feb '02 of the British mag "Electronics World" there is an article by Doug Self on this topic where he reports a G-class design that is very close to the best B amps. Basicly he used Spice analysis to identify the noise sources and devised ways around them. The final tweak was to bias the inner transistors (the ones using the 25v rails) so that they are in class A mode. He calls the result "A+C".
Back to Carver. The current Sunfire scheme is supossedly "none of the above". Appearently there some digital processing the holds the rails at 6v over the output voltage, thus the rails move with the output level. This is supossed to be so efficient that no heatsinks are needed.
I'm sure someone will say I've got it all wrong, but that's best understanding that I have.
Drafance007
My Friend recently did buy amplifier in clase H. A agree on comeents of all people before me. I am giving a link to schematic of that amp so you can practicly see how class H lokes like. it is PHONIC AMP P3 3000
Here is the link http://www.phonic.com/tech_bulletin.htm
or http://www.phonic.com/txt/p3_3000b_2000b_new1.pdf
My Friend recently did buy amplifier in clase H. A agree on comeents of all people before me. I am giving a link to schematic of that amp so you can practicly see how class H lokes like. it is PHONIC AMP P3 3000
Here is the link http://www.phonic.com/tech_bulletin.htm
or http://www.phonic.com/txt/p3_3000b_2000b_new1.pdf
sam9 said:
Indeed, there does appear to be some confusion regarding Classes G and H on the internet. My understanding is that Class G is a typical Class AB amp with multiple rail voltages. The rails are switched depending on the output level required. Class H takes a typical class A or AB design and modulates the rail voltages to follow the input.... Much like the "classless" Sunfire amp....
Mark Broker
Doug Self just confuses it with his design. (I don't mean this as criticism.) In his, the inner transistors (those powereed by the 25v rail) are in class A mode so when the output signal is near the crossover point there is no crossover noise generated. This otherwise fits the definition of class G as "A/B with multiple rails".
He does one more cute thing (at least); the transistors that drive in inner out put transistors get their voltage from the higher/outer rail thus eliminating rail switching of the drivers as a noise source.
He does one more cute thing (at least); the transistors that drive in inner out put transistors get their voltage from the higher/outer rail thus eliminating rail switching of the drivers as a noise source.
Tiered output stages.
There are basically two ways to drive them: linear and switched. Switched geneally means more noise. The Douglas Self design is linear.
If you delete the tier and the diode from the Douglas Self design you now have the cascode output stage used in many Nelson Pass designs.
The new high voltage Schottky diodes are an order of magnitude quieter than the 'fast' diodes normally used in class G/H circuits. Global feedback aggravates any residual diode noise due to loop settling time.
The Stasis feedback scheme with its common emitter output stage not in the loop would be the ideal candidate to cascode and use with the tiered supply.
The use of FETs for the cascode would simplify the design and lower cost. Mouser sells the IRF 640/9640 for less than $1 in quantity.
The $6 Pass A40 board
http://www.audioxpress.com/bksprods/pcbs/nelsonpass.htm
can be used for the front end of the Nakamichi Stasis design. The output stage is simple enough to do point-to-point, or perf board could be used. The ESP 3A board could also be used with a couple of cuts.
There are basically two ways to drive them: linear and switched. Switched geneally means more noise. The Douglas Self design is linear.
If you delete the tier and the diode from the Douglas Self design you now have the cascode output stage used in many Nelson Pass designs.
The new high voltage Schottky diodes are an order of magnitude quieter than the 'fast' diodes normally used in class G/H circuits. Global feedback aggravates any residual diode noise due to loop settling time.
The Stasis feedback scheme with its common emitter output stage not in the loop would be the ideal candidate to cascode and use with the tiered supply.
The use of FETs for the cascode would simplify the design and lower cost. Mouser sells the IRF 640/9640 for less than $1 in quantity.
The $6 Pass A40 board
http://www.audioxpress.com/bksprods/pcbs/nelsonpass.htm
can be used for the front end of the Nakamichi Stasis design. The output stage is simple enough to do point-to-point, or perf board could be used. The ESP 3A board could also be used with a couple of cuts.
- Status
- Not open for further replies.