That's really interesting reverse engineering how that output stage on the QSC works. Certainly not a typical style of output stage.
Re: Reply
1) Class H is in no way similar to class AB as AB is something
to avoid at all costs. It is a pure class B design at the constant
supply level.
2) Class H (or sonically better Class G according to D. Self)
increases efficiency not only a little bit but dramatically.
3) As I described already the transistor is NEVER exposed to
+-140V pp but to +140, -47 or vice versa -> 187V.
4) It is better to think before giving bad counsel to beginners
just to show what you know.
Uli
Workhorse said:
1) Class H is in no way similar to class AB as AB is something
to avoid at all costs. It is a pure class B design at the constant
supply level.
2) Class H (or sonically better Class G according to D. Self)
increases efficiency not only a little bit but dramatically.
3) As I described already the transistor is NEVER exposed to
+-140V pp but to +140, -47 or vice versa -> 187V.
4) It is better to think before giving bad counsel to beginners
just to show what you know.
Uli
Well, I hate to be so sharp cut, but they are situations like this that remind me to be wary of information taken for granted and to test everything carefully against standards. 🙂
The good comes even in this though, it has given me interest in designing a class H amplifier for personal use. 😉
The good comes even in this though, it has given me interest in designing a class H amplifier for personal use. 😉
Re: Re: Reply
hi ULI HAPPY NEW YEAR !
If a Class-H amplifier during its high voltage rail in operation and the output swing is just near the clip point, so can we say Still its efficiency is greater than Class-B of same voltage rail .
Regards,
Kanwar
uli said:
1) Class H is in no way similar to class AB as AB is something
to avoid at all costs. It is a pure class B design at the constant
supply level.
2) Class H (or sonically better Class G according to D. Self)
increases efficiency not only a little bit but dramatically.
3) As I described already the transistor is NEVER exposed to
+-140V pp but to +140, -47 or vice versa -> 187V.
4) It is better to think before giving bad counsel to beginners
just to show what you know.
Uli
hi ULI HAPPY NEW YEAR !
If a Class-H amplifier during its high voltage rail in operation and the output swing is just near the clip point, so can we say Still its efficiency is greater than Class-B of same voltage rail .
Regards,
Kanwar
Duo said:
Hi Duo,
this particular QSC design is something special in 2 ways:
1) As it seems to be design philosophy of QSC the output nodes
are reversed. This means that you have a floating high current
psu and the output of the active stage is connected to gnd.
The output itself is formed by the centertap of the mains XFMR.
This leads to a simpler drive circuit for the switching fets in the
power rails as the input stage is in turn referenced to gnd and
thus the more the output is driven the lower is the difference
between gnd and the powerrails.
2) All powerdevices are driven to, or rather beyond their limits.
I would not clone this design as the Toshibas are driven out of
spec!
This design should be good for 2000W RMS into 2 Ohms.
So you got 6 NPN devices to deliver e.g. 30A peak with a Vce of
about 35 Volts. This means 35V/5A per device. This is IMHO
hard on the limit of the SOA of those devices.
I would prefer Class G for my personal use, I believe they sound
better.
Uli
Remember that in class AB circuits each half of output devices only conducts half of the time and that music signals are not continuous peak-to-peak sinewaves
In practice this allows to expose output devices to instantaneous dissipations substantially higher than their DC SOA without failures, since that only happens during a small fraction of the time
As an example, I have a single pair of ancient TIP35C/TIP36C with +-41.5V regulated rails driving a Rdc=3.4 ohms bass driver [part of an active 3-way system] mounted in a hybrid horn/reflex loaded enclosure [quite reactive load], and I haven't been still able to blow them altough I like to play just at clipping threshold most of the time and the heatsink easily exceeds 50ºC 😀
In practice this allows to expose output devices to instantaneous dissipations substantially higher than their DC SOA without failures, since that only happens during a small fraction of the time
As an example, I have a single pair of ancient TIP35C/TIP36C with +-41.5V regulated rails driving a Rdc=3.4 ohms bass driver [part of an active 3-way system] mounted in a hybrid horn/reflex loaded enclosure [quite reactive load], and I haven't been still able to blow them altough I like to play just at clipping threshold most of the time and the heatsink easily exceeds 50ºC 😀
Eva said:
Hi Eva,
in Class AB one half conducts more than half of the time.
What you mean is Class B where each half conducts exactly
180° or 50% of the time!
Its a common mistake to believe that class B needs no bias.
This is class C.
The SOA area in this case (QSC) has nothing to do with the
waveform of music and the device does not care either.
Driving a BJT beyond its specs leads to desaster, music or
sine wave.
When you drive this amp to the 3rd rail with about 35V below
the rail leads to a Vce of 35V, music or not!
Even Assuming that this highes rail drops to say 110V leads
to 75V at the output. This in turn leads to 37.5 A current.
37.5 divided by 6 is 6.25 A per device peak current.
Not even the 2SC3264 is able to deliver that safely!(218.75W)
It is astonishing how fast a BJT dies if driven beyond SOA.
This worst case scenario happens sometimes.
When I was in the PA business some years ago, QSC was not
among the most reliable amps.
Uli
:
But DC SOA only makes sense for DC operation. You should look at pulsed SOA
Another example are switching transistors, these are exposed to dissipations as high as 2KW for a single TO-220 device during 200nS at turn-off, and this is perfectly within SOA
Remember that the percentage of time the third floor supply is active in QSC amps is very small, and even smaller when plaing music
Also I'm not aware of such QSC reliability problems. As a reference I had a friend that owned a local music group with 4 QSC EX4000 and 4 QSC EX1600. The EX4000 were driving 4 ohms on both channels and the EX1600 were driving 4 in one and 8 in another. He bough all the amplifiers and loudspeakers second-hand and after 4 years he sold all them when he left the live-audio business. There were no failures in the amplifiers at all, altough he destroyed several LF, MF and HF drivers due to massive overpowering
The EX1600 uses +-54V and +-108V supply rails and mounts three pairs of 2SA1302 and 2SC3281. The EX4000 uses +-46V, +-94V and +-142V supply rails and mounts six pairs of 2SA1302 and 2SC3281
These amplifiers are probably still playing in some live-audio performance somewhere this night with no failures yet, altough they are probably 7 or 8 years old and QSC has discontinued EX series long ago
Also, note that the heatsinks of these amplifiers were only warm when playing at clipping threshold and driving 4 ohms reactive loads [4 EX4000 were driving a total of 16 horn-loaded bass-bins]
How class-H affects amplifier efficiency when playing music signals [and not sine waves] is not allways well understood, in practice the lower supply rails are active more than 90% of the time
Another example are switching transistors, these are exposed to dissipations as high as 2KW for a single TO-220 device during 200nS at turn-off, and this is perfectly within SOA
Remember that the percentage of time the third floor supply is active in QSC amps is very small, and even smaller when plaing music
Also I'm not aware of such QSC reliability problems. As a reference I had a friend that owned a local music group with 4 QSC EX4000 and 4 QSC EX1600. The EX4000 were driving 4 ohms on both channels and the EX1600 were driving 4 in one and 8 in another. He bough all the amplifiers and loudspeakers second-hand and after 4 years he sold all them when he left the live-audio business. There were no failures in the amplifiers at all, altough he destroyed several LF, MF and HF drivers due to massive overpowering
The EX1600 uses +-54V and +-108V supply rails and mounts three pairs of 2SA1302 and 2SC3281. The EX4000 uses +-46V, +-94V and +-142V supply rails and mounts six pairs of 2SA1302 and 2SC3281
These amplifiers are probably still playing in some live-audio performance somewhere this night with no failures yet, altough they are probably 7 or 8 years old and QSC has discontinued EX series long ago
Also, note that the heatsinks of these amplifiers were only warm when playing at clipping threshold and driving 4 ohms reactive loads [4 EX4000 were driving a total of 16 horn-loaded bass-bins]
How class-H affects amplifier efficiency when playing music signals [and not sine waves] is not allways well understood, in practice the lower supply rails are active more than 90% of the time
Hi Eva,
yes I know that all, there is a reason why choosing the -10dB
point as the first switching point. But when I worked for the
biggest Austrian ( 😉 ) PA company we threw all QSCs out after
blowing some. Now they own Crown Macrotech and are happy
with them.
My point is, with that huge price tag and size why dont they
put 8 pairs in?
Uli
yes I know that all, there is a reason why choosing the -10dB
point as the first switching point. But when I worked for the
biggest Austrian ( 😉 ) PA company we threw all QSCs out after
blowing some. Now they own Crown Macrotech and are happy
with them.
My point is, with that huge price tag and size why dont they
put 8 pairs in?
Uli
Duo said:
I just got 25 each of the 4281 and 4302. I figured since they are 350V, 15A, I could use them in just about any amplifier I build.
I'm thinking of even using a pair of them to make a variable split power supply for testing amps, or whatever, and want to make it +/- 40V, 1 or 2A. Just not sure how many pairs would be good for that.
Reply
hi ULI,
The fact is that QSC simply wants to maximize its profit in every aspect, whether it is a simple question of paralleling devices or not.
In our case we use 16 IRFP250N devices per channel to get 1000 WRMS . This is ofcourse an overmargin but it simply makes the product more reliable in harsh conditions.
regards,
kanwar
uli said:
hi ULI,
The fact is that QSC simply wants to maximize its profit in every aspect, whether it is a simple question of paralleling devices or not.
In our case we use 16 IRFP250N devices per channel to get 1000 WRMS . This is ofcourse an overmargin but it simply makes the product more reliable in harsh conditions.
regards,
kanwar
Re: change my mind
If you are one of those with "Golden" ears, also know that devices have sonic characters.... To many, the 1943/5200 sound sweet, the MJL's sound big robust and hard...... if you don't hear the difference, then you are among the lucky ones...
😀
hienrich said:
If you are one of those with "Golden" ears, also know that devices have sonic characters.... To many, the 1943/5200 sound sweet, the MJL's sound big robust and hard...... if you don't hear the difference, then you are among the lucky ones...
😀
lucky enough......
😀
K-amps: ya, actually not so lucky with me here in my place,
why? becuase fake 5200's/1943's are really very ramphant here
but very lucky! enough to acquire some Onsemi power outputs
as free of charge about two years ago. but still on the crave of having
some 5200's/1943's (originalls).
as for my experience, I have a friend which owns a PA system
they are using these devices (I mean with the originals)
and they sound great.
regards,
hienrich
😀
😀
K-amps: ya, actually not so lucky with me here in my place,
why? becuase fake 5200's/1943's are really very ramphant here
but very lucky! enough to acquire some Onsemi power outputs
as free of charge about two years ago. but still on the crave of having
some 5200's/1943's (originalls).
as for my experience, I have a friend which owns a PA system
they are using these devices (I mean with the originals)
and they sound great.
regards,
hienrich
😀
Heinrich,
Actually you can get genuine 2sc5200/a1943 from www.digikey.com and also from www.avnet.com.
Other places have a mix of fake and real...... but the above have genuine parts as they buy directly from Toshiba. Digikey has stock on both NPN and PNP while AVNET has only PNP (but at better price). Perhaps if you buy enough, it will make up for shipping from the US. 😉
Actually you can get genuine 2sc5200/a1943 from www.digikey.com and also from www.avnet.com.
Other places have a mix of fake and real...... but the above have genuine parts as they buy directly from Toshiba. Digikey has stock on both NPN and PNP while AVNET has only PNP (but at better price). Perhaps if you buy enough, it will make up for shipping from the US. 😉
😀
maybe I can get those genuine devices later, since
financial distortions are still present for the moment.
I have bought two pairs before.
and used them as drivers, driving Onsemi's Mjl21196(quasi complementary output stage)
I wonder why MJL21196's are not popular,
actually these devices are the highest in SOA if im not wrong in
Onsemi's list
It's really hard to kill these devices, since I've tried (through
forgetfulness) my amp playing without thermal tracking and fans,
driving 4 ohm loads full power
playing with in an hour or more perhaps
its heatsinks are very very hot, you can almost light up a
cigarette,
but still after putting it off for a while they're alive and
kicking.
regards,
hienrich
maybe I can get those genuine devices later, since
financial distortions are still present for the moment.
I have bought two pairs before.
and used them as drivers, driving Onsemi's Mjl21196(quasi complementary output stage)
I wonder why MJL21196's are not popular,
actually these devices are the highest in SOA if im not wrong in
Onsemi's list
It's really hard to kill these devices, since I've tried (through
forgetfulness) my amp playing without thermal tracking and fans,
driving 4 ohm loads full power
playing with in an hour or more perhaps
its heatsinks are very very hot, you can almost light up a
cigarette,
but still after putting it off for a while they're alive and
kicking.
regards,
hienrich
I have a similar problem
Hi all.
I have decided to repair and to improve an old Kenwood KA 59. It had completely lost two channels. I think to substitute power transistors, one pair 2SB 1163 - 2SD 1718 (180V 15A 150W 20Mhz) by one pair 2SC 5200 - 2SA 1943 (230V 15A 150W 30Mhz).
The power supply voltage is:
+- 66,5 V DC for 8 ohm load
or +- 46,5 V DC for 4 ohm load.
A question:
I can to use one pair transistors or two pair transistors.
Thanks.
Guillermo.
Hi all.
I have decided to repair and to improve an old Kenwood KA 59. It had completely lost two channels. I think to substitute power transistors, one pair 2SB 1163 - 2SD 1718 (180V 15A 150W 20Mhz) by one pair 2SC 5200 - 2SA 1943 (230V 15A 150W 30Mhz).
The power supply voltage is:
+- 66,5 V DC for 8 ohm load
or +- 46,5 V DC for 4 ohm load.
A question:
I can to use one pair transistors or two pair transistors.
Thanks.
Guillermo.
according to the thread
😀
hi ,
GEirin: for best results and safer operating conditions as what
have been said in the thread multiple outputs are better than a pair
specially for higher rail voltages.
good luck.
hienrich
😀
hi ,
GEirin: for best results and safer operating conditions as what
have been said in the thread multiple outputs are better than a pair
specially for higher rail voltages.
good luck.
hienrich
- Status
- Not open for further replies.