Lumba Ogir said:
I disagree absolutelly with this statement because i am based on an application note published in an old book of RCA Solid State during 1980. Unfortunatelly these old and precious books are not published in the web. So you must be patient untill i find a way to transfer these notes in my PC and after in a post reply.
Fotios
Hi all
I used to think that the CFP was great to control quiescent current.
It does that, but I've been completely put off using CFP's now.
If any of you have simulated a CFP - just look at the current spike in the driver transistors immediately after turning on!
It seems that the output transistors don't see the right turn-on point in a CFP. Then, they turn on too late. So the driver has to compensate with a large spike until the output catches up. This is also, probably, caused in part by the feedback path around the CFP which borders on oscillation. It will affect quasi-complementary amps as well as a CFP pair. Perhaps this is one more reason why transistor amps had strange sound quality at times???
(simulations were for 20 kHz output signal near max. output power).
If anyone else sees such effects I'd like to hear.
Meanwhile I'm sticking with fully complementary conventional Darlington type (but not darlington transistors) outputs.
cheers
John
I used to think that the CFP was great to control quiescent current.
It does that, but I've been completely put off using CFP's now.
If any of you have simulated a CFP - just look at the current spike in the driver transistors immediately after turning on!
It seems that the output transistors don't see the right turn-on point in a CFP. Then, they turn on too late. So the driver has to compensate with a large spike until the output catches up. This is also, probably, caused in part by the feedback path around the CFP which borders on oscillation. It will affect quasi-complementary amps as well as a CFP pair. Perhaps this is one more reason why transistor amps had strange sound quality at times???
(simulations were for 20 kHz output signal near max. output power).
If anyone else sees such effects I'd like to hear.
Meanwhile I'm sticking with fully complementary conventional Darlington type (but not darlington transistors) outputs.
cheers
John
Re: Re: Re: 60 watt troubles
Andrew, as i have seen in the ESP web page about this amplifier, there is a note for the appropriate iddle current according to supply level:
1) +/-35V----> 20mA
2) +/-40V----> 53mA
3) +/-45V----> 78mA
Also the author says that a 100mA iddle current it is the better according to his experiments.
With the refference supply of +/-35V on the schematic, we can achieve an output power of about 70W/8Ù. That means a peak current of 4,2A which can be sustained from the 2955-3055.
Let us to say that i want to build a such amplifier with 150W/8Ù output. This means a supply of +/-52V at least. Also from the SOA given for 2955-3055 the max current it is 1,5A as much per device in such supply level. The current for a 8Ù load it is 6,12Apeak. To this, for safety we need at least 4 pairs of output transistors. For a 4Ù load we need 6 pairs because the absence of a VI limitter circuit.
According to the iddle current equivalents for each supply level given above, how much must be the total or the /per device iddle current for a supply level of +/-52V?
In my smallest project of +/-60V supply with EF output consists from 4 pairs of output devices, an iddle current of 25mA per device it is enough to eliminate the cross distortion and any further increase it is useless. Thus the total iddle current needed it is 100mA per rail.
This is a statement, because it has examined up to death in the workbench and for two years in real world operation - not with dummy loads but with speakers many times paralleled down to 2Ù - without any false.
I repeat my question; an amplifier such mine with EF output, with a proportional CFP output how much iddle current needs to not present x-over distortion?
Fotios
AndrewT said:
Andrew, as i have seen in the ESP web page about this amplifier, there is a note for the appropriate iddle current according to supply level:
1) +/-35V----> 20mA
2) +/-40V----> 53mA
3) +/-45V----> 78mA
Also the author says that a 100mA iddle current it is the better according to his experiments.
With the refference supply of +/-35V on the schematic, we can achieve an output power of about 70W/8Ù. That means a peak current of 4,2A which can be sustained from the 2955-3055.
Let us to say that i want to build a such amplifier with 150W/8Ù output. This means a supply of +/-52V at least. Also from the SOA given for 2955-3055 the max current it is 1,5A as much per device in such supply level. The current for a 8Ù load it is 6,12Apeak. To this, for safety we need at least 4 pairs of output transistors. For a 4Ù load we need 6 pairs because the absence of a VI limitter circuit.
According to the iddle current equivalents for each supply level given above, how much must be the total or the /per device iddle current for a supply level of +/-52V?
In my smallest project of +/-60V supply with EF output consists from 4 pairs of output devices, an iddle current of 25mA per device it is enough to eliminate the cross distortion and any further increase it is useless. Thus the total iddle current needed it is 100mA per rail.
This is a statement, because it has examined up to death in the workbench and for two years in real world operation - not with dummy loads but with speakers many times paralleled down to 2Ù - without any false.
I repeat my question; an amplifier such mine with EF output, with a proportional CFP output how much iddle current needs to not present x-over distortion?
Fotios
john_ellis said:
So me also. Never i made these tests you reffer, but i am in position to suppose your results simply by looking the schematic drawing of the CFP arrangement except in the ESP website also in the D. Self book. Moreover, each output transistor it is pre-polarized also by the b-e resistor of 220Ù. Very hot state.
Regards
Fotios
Personally, I feel that the triple EF output stage (as used by Leach) works best... under most conditions the predriver and driver transistors dont fully switch off as they do on a regular EF.
I've found them less prone to oscillation and they give excellent sound quality. The key is to use good predriver transistors such as those that make a good VAS - high voltage, low capacitance, good fT of around 100-150MHz. My personal favorite is KSA1381/KSC3503 from Fairchild.
I've found them less prone to oscillation and they give excellent sound quality. The key is to use good predriver transistors such as those that make a good VAS - high voltage, low capacitance, good fT of around 100-150MHz. My personal favorite is KSA1381/KSC3503 from Fairchild.
i totally disagree
EWorkshop1708 To reduce oscillation with CFP, try two things...................
1. Increase zobel cap to 0.2uf
2. Add a small cap across B-C of drivers (100pf-1nf)
this is out if the question for me .......simular tests have been performed by others and rod about zombel and miller caps in the vas or driver stage ......
both changes will have serious effect on the performance of the amplifier .....
eworkshop !!!! you dont treat ocilation with cortizon ....there is no point on trying to supress it !!!!! you simply have to find where is it coming from.....
regards sakis
EWorkshop1708 To reduce oscillation with CFP, try two things...................
1. Increase zobel cap to 0.2uf
2. Add a small cap across B-C of drivers (100pf-1nf)
this is out if the question for me .......simular tests have been performed by others and rod about zombel and miller caps in the vas or driver stage ......
both changes will have serious effect on the performance of the amplifier .....
eworkshop !!!! you dont treat ocilation with cortizon ....there is no point on trying to supress it !!!!! you simply have to find where is it coming from.....
regards sakis
Hi,
the CFP has a powerful intrinsic linearizing mechanism without a single additional part. The driver`s characteristics dominates the stage controlling all parameters including DC stability (separate heat sinks needed) and providing low output impedance. Low bias voltages improve linearity.
The CFP is far superior to the common EF (especially when the instability issue solved without much too deteriorating compensation), but the implementation may be troublesome. The devices should be chosen carefully, speed is essential (the miserable bandwidth of bipolar power transistors prevented its usage under a long time).
The turn-off problems of bipolar output transistors are well-known causing unpleasant crossover distortion. Lower value current source resistors allow faster discharge of the bases.
47Ohm is the highest value I could accept here also for minimum standing current.
the CFP has a powerful intrinsic linearizing mechanism without a single additional part. The driver`s characteristics dominates the stage controlling all parameters including DC stability (separate heat sinks needed) and providing low output impedance. Low bias voltages improve linearity.
The CFP is far superior to the common EF (especially when the instability issue solved without much too deteriorating compensation), but the implementation may be troublesome. The devices should be chosen carefully, speed is essential (the miserable bandwidth of bipolar power transistors prevented its usage under a long time).
The turn-off problems of bipolar output transistors are well-known causing unpleasant crossover distortion. Lower value current source resistors allow faster discharge of the bases.
47Ohm is the highest value I could accept here also for minimum standing current.
Attachments
lucky guy
this guy ____rtill____ the original and first post he will get all the theory and so many opinions for free !!!!!! ha ha ha
welcome to diy !!!!!! all the best is here ......
one of the boys work for me proudly talked to me about this http://www.physorg.com/news126282133.html
this is a way to compress even 1000 times more compressed from mp3....
i think about this and i say to my shelf ..... world will get back to vinyl soon when quality of sound is the target .....
same with P3 ..... simple things tend to play better ......
one thing i forgot to say in my previous post about P3 compaire and tests is power ....
since i work with this amp for more than 3 month and listen to it daily still i dont know how much power will it make ..... didnt had the time for that ....also there was never a question about power to me ...... at least for this amp
this guy ____rtill____ the original and first post he will get all the theory and so many opinions for free !!!!!! ha ha ha
welcome to diy !!!!!! all the best is here ......
one of the boys work for me proudly talked to me about this http://www.physorg.com/news126282133.html
this is a way to compress even 1000 times more compressed from mp3....
i think about this and i say to my shelf ..... world will get back to vinyl soon when quality of sound is the target .....
same with P3 ..... simple things tend to play better ......
one thing i forgot to say in my previous post about P3 compaire and tests is power ....
since i work with this amp for more than 3 month and listen to it daily still i dont know how much power will it make ..... didnt had the time for that ....also there was never a question about power to me ...... at least for this amp
photo of the stereo pcb
working ocilation free ....
An externally hosted image should be here but it was not working when we last tested it.
working ocilation free ....
and the photo of complete device
rear view
rear view
An externally hosted image should be here but it was not working when we last tested it.
I hate fixing power amps because the circuit is a negative feedback loop.
I have seen me replace every transistor only to find the last one was the faulty one ! Altho could have been more than one faulty.
One thing I do with my designs is add a microcontroller to monitor the output and disconnect the speakers via a relay if a fault condition is found. i.e. >+-20 volts for more than half a second means DC is being output from a probably faulty output transistor.
The micro also allows a 3 second anti thump delay before switching on the relay.
This circuit has saved me a fortune in speakers !
I would go for replacing all the transistors first and then replace any resistors on the outputs. Check for short circuit capacitors.
I have seen me replace every transistor only to find the last one was the faulty one ! Altho could have been more than one faulty.
One thing I do with my designs is add a microcontroller to monitor the output and disconnect the speakers via a relay if a fault condition is found. i.e. >+-20 volts for more than half a second means DC is being output from a probably faulty output transistor.
The micro also allows a 3 second anti thump delay before switching on the relay.
This circuit has saved me a fortune in speakers !
I would go for replacing all the transistors first and then replace any resistors on the outputs. Check for short circuit capacitors.
Re: i totally disagree
I appreciate your response, but I curiously have some questions, and statements.....
A lot of sources of oscillation come from noisy electronics around the amp, and I can't help that much with my computer for example.
I've noticed that caps in these places reduce external noise and RF pickup. My last subwoofer amp used fast 170Mhz transistors for the frontend, 35mhz ones for the drivers & output. Don't you have to slow the transistors down a bit closer to the audio range? I am aware that you can kill the treble performance by going too much on these caps, so the object is to keep em small.
Also, how does the larger zobel cap 0.2uf hurt the amp? On my subwoofer amp, I used to get a hum with 0.1uf only when my fingers touched the output positive speaker terminal. Going to 0.2uf stopped that completely. Besides going to 0.2uf, after trying several 0.2s, I used a better 0.22uf 250V cap from a computer PSU
NO oscillation.
I can drive 4 ohms with 10000 uf (speaker or resistor) and get good treble and lower. Any amp I build I would hope not have a problem driving a small 0.2uf.
Go ahead and respond. I am aware there's also other frequency compensation methods. We all can learn something, that's why this forum is awesome.
sakis said:
I appreciate your response, but I curiously have some questions, and statements.....
A lot of sources of oscillation come from noisy electronics around the amp, and I can't help that much with my computer for example.
I've noticed that caps in these places reduce external noise and RF pickup. My last subwoofer amp used fast 170Mhz transistors for the frontend, 35mhz ones for the drivers & output. Don't you have to slow the transistors down a bit closer to the audio range? I am aware that you can kill the treble performance by going too much on these caps, so the object is to keep em small.
Also, how does the larger zobel cap 0.2uf hurt the amp? On my subwoofer amp, I used to get a hum with 0.1uf only when my fingers touched the output positive speaker terminal. Going to 0.2uf stopped that completely. Besides going to 0.2uf, after trying several 0.2s, I used a better 0.22uf 250V cap from a computer PSU
NO oscillation.I can drive 4 ohms with 10000 uf (speaker or resistor) and get good treble and lower. Any amp I build I would hope not have a problem driving a small 0.2uf.
Go ahead and respond. I am aware there's also other frequency compensation methods. We all can learn something, that's why this forum is awesome.
i will answer with a very
small example
once i was working on a mosfet amplifier board that was given to me by a "friend" that had vas amplifier mje340 and also on the pcb a mounting hall .... so i used a very small heat sink one M3 screw and attached MJE340 to the board .....
that prooved major mistake since the screw ir shelf created some capacitance while laying very close to some of the traces of the pcb ....please notice i said laying not touching resulting huge ocilation
at the time no miller cap was placed at all in the vas amlifier ....
so what i mean is that simply you dont place a 10mf ( !!!!! ) miller cap to the vas stage to avoid this ......
you simply remove the screw that holds down the mje 340 replacing it with aplastic one or even better you "lift " the mje 340 standing up and away from any of the traces ....
all this of course was bad pcb design but even the "friend " who gave it ot me didnt know that ....
you may see the vas mje 340 standing up in this picture
coments please
small example
once i was working on a mosfet amplifier board that was given to me by a "friend" that had vas amplifier mje340 and also on the pcb a mounting hall .... so i used a very small heat sink one M3 screw and attached MJE340 to the board .....
that prooved major mistake since the screw ir shelf created some capacitance while laying very close to some of the traces of the pcb ....please notice i said laying not touching resulting huge ocilation
at the time no miller cap was placed at all in the vas amlifier ....
so what i mean is that simply you dont place a 10mf ( !!!!! ) miller cap to the vas stage to avoid this ......
you simply remove the screw that holds down the mje 340 replacing it with aplastic one or even better you "lift " the mje 340 standing up and away from any of the traces ....
all this of course was bad pcb design but even the "friend " who gave it ot me didnt know that ....
you may see the vas mje 340 standing up in this picture
coments please
An externally hosted image should be here but it was not working when we last tested it.
to make the post complete
and be some help to our friend who started the thread ...
to my opinion if you have a bad designed pcb or a terrible ground rooting or non existing by pass caps and suff like that ...then of course you might have a device that ocilating .....
solution to that will not be to add a huge miller (s) cap to vas or driver stage or a big zombel ..... ( i havent seen 0.2mf to any comercial amplifier ever in my life )
solution will be to redesign the pcb even in a veroboard .... add miller cap to vas stage but only up to needed ( scope required ) and also very close to pins ( also importand ) and thats it !!!!!
and be some help to our friend who started the thread ...
to my opinion if you have a bad designed pcb or a terrible ground rooting or non existing by pass caps and suff like that ...then of course you might have a device that ocilating .....
solution to that will not be to add a huge miller (s) cap to vas or driver stage or a big zombel ..... ( i havent seen 0.2mf to any comercial amplifier ever in my life )
solution will be to redesign the pcb even in a veroboard .... add miller cap to vas stage but only up to needed ( scope required ) and also very close to pins ( also importand ) and thats it !!!!!
Hi Jaycee
I agree that the triple EF output stage (really this probably is a "common collector" rather than emitter follower- emitter followers assume low impedance drive. When the drive impedance is high, the transistors behave more like CE) works well.
Simulations of a simple complementary design using a symmetrical input pair,VAS stage and NPN/NPN/NPN outputs (2 power devices in parallel) with complementary PNP stage excellent performance into 2 or 4 ohms...
No problem with loop oscillation paths within the output stages, which CFP's have a tendency to generate.
cheers
John
I agree that the triple EF output stage (really this probably is a "common collector" rather than emitter follower- emitter followers assume low impedance drive. When the drive impedance is high, the transistors behave more like CE) works well.
Simulations of a simple complementary design using a symmetrical input pair,VAS stage and NPN/NPN/NPN outputs (2 power devices in parallel) with complementary PNP stage excellent performance into 2 or 4 ohms...
No problem with loop oscillation paths within the output stages, which CFP's have a tendency to generate.
cheers
John
hi
I would not recommend this amp. as "hifi" but as a good standard workhorse it is OK.
Mods from the Rod Elliott include:
six diodes for some degree of protection
- two in the VAS
-two in each output half
output current limited to approx. 4A each, can reduce emitter resistors a little if needed.
One of the two 470 ohm base resistors in the bias regulator is a variable.
SHould not need a sprog-stopper capacitor with CC driver/outputs.
I recommend 2N3055/MJ2955 instead of MJE versions - the TO-3's have higher SOA (2.9A @40V, instead of ~2A/35V for MJE3055T's. Not sure if MJE3055 (90W) available any more.
You could use two MJE3055T's in parallel instead of a TO-3 can type...
cheers
John
I would not recommend this amp. as "hifi" but as a good standard workhorse it is OK.
Mods from the Rod Elliott include:
six diodes for some degree of protection
- two in the VAS
-two in each output half
output current limited to approx. 4A each, can reduce emitter resistors a little if needed.
One of the two 470 ohm base resistors in the bias regulator is a variable.
SHould not need a sprog-stopper capacitor with CC driver/outputs.
I recommend 2N3055/MJ2955 instead of MJE versions - the TO-3's have higher SOA (2.9A @40V, instead of ~2A/35V for MJE3055T's. Not sure if MJE3055 (90W) available any more.
You could use two MJE3055T's in parallel instead of a TO-3 can type...
cheers
John
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.