I'm still puzzled...
My conclusions(?) sofar:
Why do anyone use quasi designs(exept for cheapskate designs)?
Why use mosfets in audio, when BJT's are so linear and easy to work with?
-maybe because someone just can do so...
-maybe because they sounds good?
Arne K
My conclusions(?) sofar:
Why do anyone use quasi designs(exept for cheapskate designs)?
Why use mosfets in audio, when BJT's are so linear and easy to work with?
-maybe because someone just can do so...
-maybe because they sounds good?
Arne K
Re: I'm still puzzled...
Hello Arne..
You Conclusions so far.....OK...great...
If Quasi complementary amp is designed with Symmetric drives , then it surpasses the complementary....especially in N-Channel Mosfets...though Complementary designs in BJT are much better than Quasi BJT designs...But in Mosfet N-channel Quasi wins....when both High output power and great performance is required....[mine views were only in terms of mosfet amps with vertical mosfets]
BJT were very easy to work with when small power requirement less than 1 KW is required only....
Mosfet amps can have Slew rate in excess over 100V/uS easily and thus cause negligeble phase shift at higher audio frequencies...
regards,
K a n w a r
Cobra2 said:
Hello Arne..
You Conclusions so far.....OK...great...
If Quasi complementary amp is designed with Symmetric drives , then it surpasses the complementary....especially in N-Channel Mosfets...though Complementary designs in BJT are much better than Quasi BJT designs...But in Mosfet N-channel Quasi wins....when both High output power and great performance is required....[mine views were only in terms of mosfet amps with vertical mosfets]
BJT were very easy to work with when small power requirement less than 1 KW is required only....
Mosfet amps can have Slew rate in excess over 100V/uS easily and thus cause negligeble phase shift at higher audio frequencies...
regards,
K a n w a r
BJT is cost effective up to about 5KW.
The MJ21195/96 has the full 250W SOA rating up to 80V (it is a 250V device).
If the amp is driving a 90* phase angle then the devices will have to take the rail voltage until your protection kicks in.
The Crown VZ5000 runs on ±80V. In parallel bridge it does 5KW into 1 ohm, in series bridge it does 5KW into 4 ohms.
It has a total of 32 outputs. Originally it used the MJ15024/25, only rated at 160W SOA at 80V, now they use the newer MJ21195/96 as they are the same cost. In small quantity they are less than $2.50 each.
From Digi-Key in the same quantity, the IRFP260 is about the same cost as the MJ parts. Even though it is rated at higher power, because of the difference in maximum junction temperatures it is not as heavy of a part in actual use.
IXYS does have a very interesting 400W power J-fet available. I am inquiring as to cost and availability.
The MJ21195/96 has the full 250W SOA rating up to 80V (it is a 250V device).
If the amp is driving a 90* phase angle then the devices will have to take the rail voltage until your protection kicks in.
The Crown VZ5000 runs on ±80V. In parallel bridge it does 5KW into 1 ohm, in series bridge it does 5KW into 4 ohms.
It has a total of 32 outputs. Originally it used the MJ15024/25, only rated at 160W SOA at 80V, now they use the newer MJ21195/96 as they are the same cost. In small quantity they are less than $2.50 each.
From Digi-Key in the same quantity, the IRFP260 is about the same cost as the MJ parts. Even though it is rated at higher power, because of the difference in maximum junction temperatures it is not as heavy of a part in actual use.
IXYS does have a very interesting 400W power J-fet available. I am inquiring as to cost and availability.
djk said:
Hi DJK,
QSC uses 16 X IXYS 74N20 Mosfets to produce 9KW of Audio power.....in PL9.0 amplifiers...
K a n w a r
moer-
2) Fets have much higher distortion than BJts and so masses of global feedback must be applied. We have built tens of thousands of quasi FET amps for the car stereo industry (Cheap to make as the drive ccts are simple and the FETs are cheap) but they sound terrible.
----------------------------------
Many great Designers i dont wish to mention their view (not mine)
is that fets are more linear than bjt, i dont agree entirely with your post, and i dont agree with your remark that fets sound terrible, i know dear Kanwar has been saying some things about
quasis that ive found amusing to say the least
2) Fets have much higher distortion than BJts and so masses of global feedback must be applied. We have built tens of thousands of quasi FET amps for the car stereo industry (Cheap to make as the drive ccts are simple and the FETs are cheap) but they sound terrible.
----------------------------------
Many great Designers i dont wish to mention their view (not mine)
is that fets are more linear than bjt, i dont agree entirely with your post, and i dont agree with your remark that fets sound terrible, i know dear Kanwar has been saying some things about
quasis that ive found amusing to say the least
I can only find the price for the IXYS 58N20 ($5.83 @ 100pc) and the 80N20 ($8.39 @ 100pc).
Even though they use half the number, there looks to be no huge cost savings. About $12.64 per KW on the FET (using the average cost of the above IXYS parts), about $16 per KW on the BJT.
Even though they use half the number, there looks to be no huge cost savings. About $12.64 per KW on the FET (using the average cost of the above IXYS parts), about $16 per KW on the BJT.
Re: Quasi vs Fully com
Hi MOER,
1)..Yes the Upper Mosfet is Common Drain and Lower Mosfet is Common Source amp.....only when implemented idividually, for example a simple CFP at lower side would virtually turns it into a Unity gain too.....
2)Not Correct...It depends upon how you design the Driver Stage
and establish symmetric drive proportions for both halves must be met, other wise non-linearities would occur....You could also build BILLIONS of QUASI-BJT amps at throw-away prices..much cheaper than Quasi-Mosfet using 2N3773 or 2N3055
3) You could only touch the Rails within 0.5V if you use Mosfets, because BJT's have typical saturation level of 1.2 Volts..so it can't be with BJT amps...Touching rails:We have 2 models of amps first one has rail loss upto 5 Volts max.....and second one's is just 0.5volt max, but the difference is apparent when drving 2 OHMS load and power output is above 1KW
Hi Mastertech,
I would really like to know the things which amused you alot about Quasi-Mosfets.......please post it....
Hi DJK,
IXYS Mosfets were very costly as compared to IRF and APT...try them...
best regards,
K a n w a r
MOER said:
Hi MOER,
1)..Yes the Upper Mosfet is Common Drain and Lower Mosfet is Common Source amp.....only when implemented idividually, for example a simple CFP at lower side would virtually turns it into a Unity gain too.....
2)Not Correct...It depends upon how you design the Driver Stage
and establish symmetric drive proportions for both halves must be met, other wise non-linearities would occur....You could also build BILLIONS of QUASI-BJT amps at throw-away prices..much cheaper than Quasi-Mosfet using 2N3773 or 2N3055
3) You could only touch the Rails within 0.5V if you use Mosfets, because BJT's have typical saturation level of 1.2 Volts..so it can't be with BJT amps...Touching rails:We have 2 models of amps first one has rail loss upto 5 Volts max.....and second one's is just 0.5volt max, but the difference is apparent when drving 2 OHMS load and power output is above 1KW
mastertech said:
Hi Mastertech,
I would really like to know the things which amused you alot about Quasi-Mosfets.......please post it....
djk said:
Hi DJK,
IXYS Mosfets were very costly as compared to IRF and APT...try them...
best regards,
K a n w a r
workhorse-
Hi Mastertech,
I would really like to know the things which amused you alot about Quasi-Mosfets.......please post it....
--------------------------------------------------------
to date i havent seen a quasi that surpasses a fully complementary design in one parameter only, sorry i cant
disclose it, if you have such a design i would like to see it
so i could run it through my pascal spice simulator to see if
it agrees or not
cheers
Hi Mastertech,
I would really like to know the things which amused you alot about Quasi-Mosfets.......please post it....
--------------------------------------------------------
to date i havent seen a quasi that surpasses a fully complementary design in one parameter only, sorry i cant
disclose it, if you have such a design i would like to see it
so i could run it through my pascal spice simulator to see if
it agrees or not
cheers
mastertech said:
What is that thing , which you can't disclose, is it a circuit or schematic of which amp..quasi or comp...
K a n w a r
you should be confident of yourself, why so curious, as i said
quasis if designed by a good designer will sound good too
good luck
P.S: you know i cant disclose cutting edge technology
quasis if designed by a good designer will sound good too
good luck
P.S: you know i cant disclose cutting edge technology
mastertech said:
Oh Mastertech,
I still cannot get you..What you want to say...Please clarify the statement a bit more...
Its true that the amp designer has its influence on its amp's design....
I dont want YOU to reveal your cutting edge technology either....
regards,
K a n w a r
amps
Dear Kanwar,
Who in his right mind would use 2N3773/2N3055/2N3443/2N6259 antiquated TO3 devices?
These are slow hogs with Fts in the KHz range. yes the Mot 2N3055 is rated at 2MHz.
I stopped using TO3 in the late 70s when high power plastics were coming out. Easy to mount, less expensive, better specs, sound better, more choices of devices.
Please name one really good high end solid state amplifier which uses Mosfets....... and please do not say Halcro.
Mosfets do have higher intrinsic distortion than BJts.
Mosfets have SOA curves just like BJts
Mosfets have thermal problems yes not the same as BJts. Try biasing a mosfet amplifier at say 10mA per fet and see how it sounds.
The quasi mosfet amps I designed used symetrical drive to the outputs and no CFP can make a common source behave like a source follower.
This is why the Futterman tube amps sounded bad, the bottom tube had gain no matter how hard they tried.
This is why the older quasi amps from Crown (DC300 etc), Phase Linear, BGW and other from the 70's era sounded so bad. the fact that they had poor drive circuits was one issue, but the quasi output topology is prone to common mode conduction as well.
Doulas Self has a good chapter on the pros and cons of the quasi output stage in his book.
Biasing is more tempremental than a fully comp design. The bottom trannie's Vbe is not in the loop.
Biasing of quasi FET amps is not that easy and I have found that they do not tolerate any over biasing at all. They tend to run away.
I am sure for PA applications that quasi designs are OK either BJt or FET.
have fun all
Dear Kanwar,
Who in his right mind would use 2N3773/2N3055/2N3443/2N6259 antiquated TO3 devices?
These are slow hogs with Fts in the KHz range. yes the Mot 2N3055 is rated at 2MHz.
I stopped using TO3 in the late 70s when high power plastics were coming out. Easy to mount, less expensive, better specs, sound better, more choices of devices.
Please name one really good high end solid state amplifier which uses Mosfets....... and please do not say Halcro.
Mosfets do have higher intrinsic distortion than BJts.
Mosfets have SOA curves just like BJts
Mosfets have thermal problems yes not the same as BJts. Try biasing a mosfet amplifier at say 10mA per fet and see how it sounds.
The quasi mosfet amps I designed used symetrical drive to the outputs and no CFP can make a common source behave like a source follower.
This is why the Futterman tube amps sounded bad, the bottom tube had gain no matter how hard they tried.
This is why the older quasi amps from Crown (DC300 etc), Phase Linear, BGW and other from the 70's era sounded so bad. the fact that they had poor drive circuits was one issue, but the quasi output topology is prone to common mode conduction as well.
Doulas Self has a good chapter on the pros and cons of the quasi output stage in his book.
Biasing is more tempremental than a fully comp design. The bottom trannie's Vbe is not in the loop.
Biasing of quasi FET amps is not that easy and I have found that they do not tolerate any over biasing at all. They tend to run away.
I am sure for PA applications that quasi designs are OK either BJt or FET.
have fun all
Re: amps
Who in his right mind would use 2N3773/2N3055/2N3443/2N6259 antiquated TO3 devices?
These are slow hogs with Fts in the KHz range. yes the Mot 2N3055 is rated at 2MHz.I stopped using TO3 in the late 70s when high power plastics were coming out. Easy to mount, less expensive, better specs, sound better, more choices of devices.
HAHAHAHAAAA....I thing a person with mental runaway would definately use them.....
Please name one really good high end solid state amplifier which uses Mosfets....... and please do not say Halcro.
Nelson Pass uses Mosfets in Pass Amps..
ECLER ..Professional amps uses Mosfets only N-channels...
HAFLER amps....both vertical and Lateral mosfets...
MTX audio....Car amps...
Mosfets do have higher intrinsic distortion than BJts.
YES
Mosfets have SOA curves just like BJts
NO, Mosfets Have Much Greater SOA than BJT's
Mosfets have thermal problems yes not the same as BJts.
BJT suffer from Second Breakdown Voltage.....
Try biasing a mosfet amplifier at say 10mA per fet and see how it sounds.
I think do same with BJT's and here it...as well
The quasi mosfet amps I designed used symetrical drive to the outputs and no CFP can make a common source behave like a source follower.
Performance is based on the designing as well....
The quasi output topology is prone to common mode conduction as well.
So as do Complementary pairs also
Biasing is more tempremental than a fully comp design. The bottom trannie's Vbe is not in the loop.
Again influenced by design...
Biasing of quasi FET amps is not that easy and I have found that they do not tolerate any over biasing at all. They tend to run away.
Only If Temp compensation isn't well implemented...
have fun all
Cheers to you too...
[/B][/QUOTE]
Who in his right mind would use 2N3773/2N3055/2N3443/2N6259 antiquated TO3 devices?
These are slow hogs with Fts in the KHz range. yes the Mot 2N3055 is rated at 2MHz.I stopped using TO3 in the late 70s when high power plastics were coming out. Easy to mount, less expensive, better specs, sound better, more choices of devices.
HAHAHAHAAAA....I thing a person with mental runaway would definately use them.....
Please name one really good high end solid state amplifier which uses Mosfets....... and please do not say Halcro.
Nelson Pass uses Mosfets in Pass Amps..
ECLER ..Professional amps uses Mosfets only N-channels...
HAFLER amps....both vertical and Lateral mosfets...
MTX audio....Car amps...
Mosfets do have higher intrinsic distortion than BJts.
YES
Mosfets have SOA curves just like BJts
NO, Mosfets Have Much Greater SOA than BJT's
Mosfets have thermal problems yes not the same as BJts.
BJT suffer from Second Breakdown Voltage.....
Try biasing a mosfet amplifier at say 10mA per fet and see how it sounds.
I think do same with BJT's and here it...as well
The quasi mosfet amps I designed used symetrical drive to the outputs and no CFP can make a common source behave like a source follower.
Performance is based on the designing as well....
The quasi output topology is prone to common mode conduction as well.
So as do Complementary pairs also
Biasing is more tempremental than a fully comp design. The bottom trannie's Vbe is not in the loop.
Again influenced by design...
Biasing of quasi FET amps is not that easy and I have found that they do not tolerate any over biasing at all. They tend to run away.
Only If Temp compensation isn't well implemented...
have fun all
Cheers to you too...
[/B][/QUOTE]
more mosfet
I have built about 10 variations on the Nelson Pass' BoSoZ, and I have also tried similar circuits with BJT. With LOTS of bias...
MosFets rule...BJT loose.
Maybe J-Fets will be even better?, still waiting for devices.
OK, it is not the most advanced design, it does have measurable distortion (barely) on my HP339A.
But it sounds GREAT!
But in another config, BJT's are great.
I just do not buy a "mosfet write-off".
Arne K
I have built about 10 variations on the Nelson Pass' BoSoZ, and I have also tried similar circuits with BJT. With LOTS of bias...
MosFets rule...BJT loose.
Maybe J-Fets will be even better?, still waiting for devices.
OK, it is not the most advanced design, it does have measurable distortion (barely) on my HP339A.
But it sounds GREAT!
But in another config, BJT's are great.
I just do not buy a "mosfet write-off".
Arne K
NVMOS Addiction
Hi ARNE,
Don't You know a Gentleman known as
Negative Channel Vertical Metal Oxide Semiconductor Field Effect Transistor a.k.a. NVMOS......If implemented properly, results in exceptionally great sounding amp with great performance...if you are unfamiliar than meet this guy and do a friendship act....he would never let you down ..trust him.
regards,
K a n w a r🙂
Hi ARNE,
Don't You know a Gentleman known as
Negative Channel Vertical Metal Oxide Semiconductor Field Effect Transistor a.k.a. NVMOS......If implemented properly, results in exceptionally great sounding amp with great performance...if you are unfamiliar than meet this guy and do a friendship act....he would never let you down ..trust him.
regards,
K a n w a r🙂
Hi everyone....,
Visit these links of a high end N-channel Mosfet amp Manufacturer using HexFet's in Quasi comp.Manner...ADCOM is the name of company....
http://www.audioadvisor.com/store/p...&product_name=GFA-5802 300wpc Power Amplifier
http://www.adcom.com/itemreviews_002.htm
http://www.adcom.com/amplifiers/gfa5802.htm
Visit these links of a high end N-channel Mosfet amp Manufacturer using HexFet's in Quasi comp.Manner...ADCOM is the name of company....
http://www.audioadvisor.com/store/p...&product_name=GFA-5802 300wpc Power Amplifier
http://www.adcom.com/itemreviews_002.htm
http://www.adcom.com/amplifiers/gfa5802.htm
Hi Kanwar
How you doing mate😉
Its true properly implemented N-channel output stage amplifiers
are very rugged, more so than BJT designs.
dispite the intrinsic low linearity of HEXFETs or NVMOS devices
very low distortion figures can be achieved and excellent performance in terms of efficency are realised.
The new n-channel design I am working on has a R2R performance of around 1 volt rail loss.
allowing well over 500 watts RMS into 4 Ohms with only +-70 volt rails
under load the supply drops to 66 volts
The sonic performance is also very good, of cause this is only my opinion.
But you will get to hear this design yourself when it is released onto the world market....
Thermal management of N-channel Mosfets is tricky, but once mastered, works very well in deed.
How you doing mate😉
Its true properly implemented N-channel output stage amplifiers
are very rugged, more so than BJT designs.
dispite the intrinsic low linearity of HEXFETs or NVMOS devices
very low distortion figures can be achieved and excellent performance in terms of efficency are realised.
The new n-channel design I am working on has a R2R performance of around 1 volt rail loss.
allowing well over 500 watts RMS into 4 Ohms with only +-70 volt rails
under load the supply drops to 66 volts
The sonic performance is also very good, of cause this is only my opinion.
But you will get to hear this design yourself when it is released onto the world market....
Thermal management of N-channel Mosfets is tricky, but once mastered, works very well in deed.
Hi kanwar,
I doubt that proving Hi End cred for quasi-comp is really advanced by links to Adcoms well oiled publicity machine.
These may drive nails and sound like a slice of heaven to a journo who's just received one on permanent loan, as is the practice.
These are a cut below what I'm used to blowing away.
What I'd really like to see is some science here - some FFT's of residuals. Let's see what we'd be listening to. I'm a little more demanding than an Indian Wedding party. Sorry.😀
Cheers,
greg
I doubt that proving Hi End cred for quasi-comp is really advanced by links to Adcoms well oiled publicity machine.
These may drive nails and sound like a slice of heaven to a journo who's just received one on permanent loan, as is the practice.
These are a cut below what I'm used to blowing away.
What I'd really like to see is some science here - some FFT's of residuals. Let's see what we'd be listening to. I'm a little more demanding than an Indian Wedding party. Sorry.😀
Cheers,
greg
AndrewT said:
Firstly apology for replying only now; every so often I am out of contact for a few days.
I did not check this, but would imagine that things would stay the same. If you consider each pair on its own you have a certain distortion behaviour, which would not change as you multiply pairs. It is rather like feeding so many amplifiers (plus loudspeakers) separately. The maximum output will "multiply", but the distortion will not decrease. But I will just check as soon as I am able; interesting question. (The main mechanism is that while both sides are conducting the Re's are essentially in parallel, suddenly doubling to one as class B is entered.)
It might be of interest to note what sort of advantage one can get. I have simulated using very basic transistors, viz. TIP 35 and 36, driven by BD139 and 140 in a full complementary fashion. At an output of 50W (Q. current 90 mA) and with 0R33 ohm per each side, I find the following analysis (I am writing text as tables can get upset): Harmonic 3rd = 14mV; 5th = 4.1mV; 7th = 3.9mV; 9th = 3mV; 11th = 2.4mV; 13th = 1.8mV; 15th = 1.5mV; 17th = 1.3mV; 19th = 1.1mV (figures rounded off). Under the same conditions without the Re's: Harmonic 3rd = 9.5mV; 5th = 2mV; 7th = 0.02mV, with the rest buried in the noise floor of about 0.015mV. (I stayed clear of maximum output of about 70W.) The tendencies were confirmed by spectrum analysis of the built-up model and is typical.
The penalty is of course that one must now temperature-stabilise in another way, which can be done by a sensing transistor and proper heat-sinking. This requires longer time to settle after switch-on, but then I never turn my amplifier off.
Regards
Hi Andrew T
I rushed back here, hoping that nobody had noticed that my previous reply to you was sadly incomplete!
Firstly, where is my Re: In the complementary pair it is in serie with the whole pair, i.e. from the connection of the driver E and the power C, to the output. Thus, not just in serie with the power transistor E to rail, inside the pair. So for both sides.
For power transistors in parallel (one presumes the same driver) one cannot, of course, just omit an Re and connect bases and emitters together (division of current will not be equal). Thus, to omit the Re in serie with a pair as above, one will have to insert serie Rb's, or serie Re's in the emitter only of every power transistor (sorry not to append a diagram; I hope you follow).
This forms a new analysis, which I will have to perform before being able to give you a sensible answer. Sorry about that.
I rushed back here, hoping that nobody had noticed that my previous reply to you was sadly incomplete!
Firstly, where is my Re: In the complementary pair it is in serie with the whole pair, i.e. from the connection of the driver E and the power C, to the output. Thus, not just in serie with the power transistor E to rail, inside the pair. So for both sides.
For power transistors in parallel (one presumes the same driver) one cannot, of course, just omit an Re and connect bases and emitters together (division of current will not be equal). Thus, to omit the Re in serie with a pair as above, one will have to insert serie Rb's, or serie Re's in the emitter only of every power transistor (sorry not to append a diagram; I hope you follow).
This forms a new analysis, which I will have to perform before being able to give you a sensible answer. Sorry about that.
- Status
- Not open for further replies.