Hmmm,
According to the sim ( i hate these things ) increasing the LTP degen resistors to 220R to give me a volt drop of .66V causes the distortion to increase by a factor of 20!! Ouch! Also adding a capacitor between the emmiters of the LTP dosent seem to help either.
Is this going to be the case in the real world too? I have no accurate means of measuring real world distortion so this is why i use the sim to try and calculate it for me.
Anyway for now i'm just glad the oscilation has gone. It was such an obvious cause i would have never found it without somebody else stating the obvious LOL
Bedtime for me... i'll sleep on things for a while
LaterZ
Leigh
ZZZZzzzzzzzzzz
According to the sim ( i hate these things ) increasing the LTP degen resistors to 220R to give me a volt drop of .66V causes the distortion to increase by a factor of 20!! Ouch! Also adding a capacitor between the emmiters of the LTP dosent seem to help either.
Is this going to be the case in the real world too? I have no accurate means of measuring real world distortion so this is why i use the sim to try and calculate it for me.
Anyway for now i'm just glad the oscilation has gone. It was such an obvious cause i would have never found it without somebody else stating the obvious LOL
Bedtime for me... i'll sleep on things for a while
LaterZ
Leigh
ZZZZzzzzzzzzzz
You have to take your pick Nitrate . . . a power oscillator or a functioning amplifier.
The distoriton will go up because you are decreasing the open loop gain.
Right now, with the 2.2 Ohm resistors, the loop gain is so high that the poles formed by the otput stage added to the Cdom pole lie well below the unity gain cross over frequency - thats a recipe for oscillation. You need to cross over at less than 40db/decade. to do this, reduce the open loop gain - increasing th e degen resistors is a convenient and trusted way of doing this.
Rather focus on getting the open loop linearity as high as possible than trying massive amounts of loop gain to get low closed loop distortion.
For people that use GNF, there is a trade off between open loop gain and stability with this classic type of compensation.
BTW - if you reduce the value of Cdom your distorion should also decrease - but be sure to check for stability. This is an iterative process.
What is the idle current of the output stage? Try running it at 100mA, and if heatsinks allow, 150mA.
The distoriton will go up because you are decreasing the open loop gain.
Right now, with the 2.2 Ohm resistors, the loop gain is so high that the poles formed by the otput stage added to the Cdom pole lie well below the unity gain cross over frequency - thats a recipe for oscillation. You need to cross over at less than 40db/decade. to do this, reduce the open loop gain - increasing th e degen resistors is a convenient and trusted way of doing this.
Rather focus on getting the open loop linearity as high as possible than trying massive amounts of loop gain to get low closed loop distortion.
For people that use GNF, there is a trade off between open loop gain and stability with this classic type of compensation.
BTW - if you reduce the value of Cdom your distorion should also decrease - but be sure to check for stability. This is an iterative process.
What is the idle current of the output stage? Try running it at 100mA, and if heatsinks allow, 150mA.
Hi,
My idle current is 2000mA ( 2A )
So in the real world assuming good or perfect components distortion would increase due to the gain decreasing? In that case would it be of any benifit to use degeneration in the LTP and then boost the VAS gain by using a darlington arrangement to reclaim some gain and reduce distortion or would that be robbing peter to give to paul??
Leigh
My idle current is 2000mA ( 2A )
So in the real world assuming good or perfect components distortion would increase due to the gain decreasing? In that case would it be of any benifit to use degeneration in the LTP and then boost the VAS gain by using a darlington arrangement to reclaim some gain and reduce distortion or would that be robbing peter to give to paul??
Leigh
Arrgh!
Just when i thought i had the oscillation licked it came back. I tested the amp with capacitance on the output to check for stability and it started to oscillate again with as little as 470pF of capacitance. Even attaching an open ended 30ft length of speaker cable causes oscillation. It goes away with a speaker attached to the end tho..
Will speeding up the output drivers help with this problem?? I chose the tip42's for their linearity and i have lots lying around as i have used them as output devices in other amps but they are rarther slow.
I really don't wanna slow down the LTP much, i'd rarther attack the drivers or VAS.
Any ideas??
Leigh
Just when i thought i had the oscillation licked it came back. I tested the amp with capacitance on the output to check for stability and it started to oscillate again with as little as 470pF of capacitance. Even attaching an open ended 30ft length of speaker cable causes oscillation. It goes away with a speaker attached to the end tho..
Will speeding up the output drivers help with this problem?? I chose the tip42's for their linearity and i have lots lying around as i have used them as output devices in other amps but they are rarther slow.
I really don't wanna slow down the LTP much, i'd rarther attack the drivers or VAS.
Any ideas??
Leigh
Don't get hung up on trying to get the absilute maximum gain. In a practical amp, you have to make some very heavy tradeoffs. Reduce OL gain until th e amp is STABLE and produces a good square wave response (no overshoot/ringing etc). After this you can look at increasing OL gain again. If you just try to hack your way through this phase of th e design you will end up chasing your tail.
Have you increased the LTP degen resistors to 220 Ohm?
Have you got a coil on the output (2-5uH) with a 2 Ohm resistor in parallel. Without this, you absolutely do run the risk of oscillation.
Install a zobel network (10 Ohm in series with 0.1uF high voltage cap. Cconnect the Zobel from the junction of the output devices emitter degeneration resistors and the other side to the power supply caps 0V (take care of grounding here - be careful where you return the Zobel ground connection).
If you still have oscillation, try a 50pf capacitor from the Vas collector to the base of the LTP inverting input (this is the side where the feedback resistor connects).
What frequency is the oscillation BTW?
Have you increased the LTP degen resistors to 220 Ohm?
Have you got a coil on the output (2-5uH) with a 2 Ohm resistor in parallel. Without this, you absolutely do run the risk of oscillation.
Install a zobel network (10 Ohm in series with 0.1uF high voltage cap. Cconnect the Zobel from the junction of the output devices emitter degeneration resistors and the other side to the power supply caps 0V (take care of grounding here - be careful where you return the Zobel ground connection).
If you still have oscillation, try a 50pf capacitor from the Vas collector to the base of the LTP inverting input (this is the side where the feedback resistor connects).
What frequency is the oscillation BTW?
Hi,
The oscillation is at approx 3.3MHz!! It really should not be there although it is coincidently the same freq as the drivers Ft. The amp has an inductor of approx 2uH and a 3.3R resistor in parallel with it, although adding or removing the resistor currently has no effect. The oscillation freq is rock steady, almost as if it was designed to be a transmitter even when you crank the power rails up and down although it suddenly increases in amplitude when the power rails get to about 10V or more.
I know the 2A is high idle current for the two devices, after the design is stable i will be using anougher pair for longevity of the amp as each tranny currently dissipates 70W each. The second pair will be wired exactly the same as the first except both pairs will aquire 2.2R to their bases. That should do the trick i think.
Anyway i'll plod on and have a go at reducing open loop gain to see if that helps matters. Thanks for your help so far i really do appreciate it.
Leigh
P.S i have not fitted a Zobel network yet.
P.P.S Am i interpreting the myth of CDOM caps correctly here? As far as i can gather the higher the frequency of the signal the more of the signal is shunted through the Cdom cap accross to the output side of the VAS, bypassing the VAS gain device thus lowering the effective Hfe in direct relation to frequency until we hit unity gain or lower according to how capable the LTP setup is of manipulating the VAS output section ( impedence verses impedence ) or am i completley on the wrong track here??
The oscillation is at approx 3.3MHz!! It really should not be there although it is coincidently the same freq as the drivers Ft. The amp has an inductor of approx 2uH and a 3.3R resistor in parallel with it, although adding or removing the resistor currently has no effect. The oscillation freq is rock steady, almost as if it was designed to be a transmitter even when you crank the power rails up and down although it suddenly increases in amplitude when the power rails get to about 10V or more.
I know the 2A is high idle current for the two devices, after the design is stable i will be using anougher pair for longevity of the amp as each tranny currently dissipates 70W each. The second pair will be wired exactly the same as the first except both pairs will aquire 2.2R to their bases. That should do the trick i think.
Anyway i'll plod on and have a go at reducing open loop gain to see if that helps matters. Thanks for your help so far i really do appreciate it.
Leigh
P.S i have not fitted a Zobel network yet.
P.P.S Am i interpreting the myth of CDOM caps correctly here? As far as i can gather the higher the frequency of the signal the more of the signal is shunted through the Cdom cap accross to the output side of the VAS, bypassing the VAS gain device thus lowering the effective Hfe in direct relation to frequency until we hit unity gain or lower according to how capable the LTP setup is of manipulating the VAS output section ( impedence verses impedence ) or am i completley on the wrong track here??
Your described behaviour nicely fits oscillations caused by missing zobel...
nitrate,
the oscillation basically is not GNF related. The present phase compensation is very deteriorating, consider the solution suggested by Bonsai in post #27 and omit C4 and C1. For Q9 select a high speed video device with small internal capacitances. Q12 is critical, it should have the same properties as Q9, operating at its maximum speed, increased bias current and compatible coupling circuit (low phase shift needed). In order to considerably improve overall performance, lower the open loop gain.
the oscillation basically is not GNF related. The present phase compensation is very deteriorating, consider the solution suggested by Bonsai in post #27 and omit C4 and C1. For Q9 select a high speed video device with small internal capacitances. Q12 is critical, it should have the same properties as Q9, operating at its maximum speed, increased bias current and compatible coupling circuit (low phase shift needed). In order to considerably improve overall performance, lower the open loop gain.
O.K I've managed to tune out the oscillations. I now have a resistor accross the output filter to reduce its Q. This has a value of 2.2R. Also as suggested i've added a Zobel network from just befor the output flter to ground or Zero volts. The network ended up consisting of a 5R resistor and 47nF poly cap. I've tested for oscillations on the output at cap values of 33p, 120p, 470p, 1n, 10, 100n, 1u, 4.7u. All values probuced no signs of oscillation the only visible disturbences on the scope being 500uV ripple from the PSU rectifiers. I'm not going to open the champers yet though as as per usual you fix one thing then somthing else goes wrong as a result LOL.
I'll post updated shematic next post
Leigh
I'll post updated shematic next post
Leigh
Reviving POA 6600 A
I All
Here I am again searching for specialised help !
I am rebuilding a power amp (DENON POA 6600 A) and believe I am almost done.
However I am getting some trouble getting the the bias simetrical in the pre drivers bases.
This is a push pull amp that relies on a optical ship to set bias.
On the top pre driver I get 1.67v and the lower predriver base I get -1.79v.
On the drivers bases I get 1.04v (top) and -1.19v (bottom driver).
The final stage is composed of four 2SC3856 and four A1492.
I should be able to get +-0.5v on the output transistors bases but I can only get 0.48v on the 2SC3856 bases and -0.62v on the bases of the A1492.
I have not soldered the output transistors yet so the readings are made in the drivers and pre drivers only (without output stage).
I did not solder the output transistors because I am afraid these voltage differences might provoque an oscilation and burn the outputs.
I really need some help and advise.
Best Regards
Ricardo
I All
Here I am again searching for specialised help !
I am rebuilding a power amp (DENON POA 6600 A) and believe I am almost done.
However I am getting some trouble getting the the bias simetrical in the pre drivers bases.
This is a push pull amp that relies on a optical ship to set bias.
On the top pre driver I get 1.67v and the lower predriver base I get -1.79v.
On the drivers bases I get 1.04v (top) and -1.19v (bottom driver).
The final stage is composed of four 2SC3856 and four A1492.
I should be able to get +-0.5v on the output transistors bases but I can only get 0.48v on the 2SC3856 bases and -0.62v on the bases of the A1492.
I have not soldered the output transistors yet so the readings are made in the drivers and pre drivers only (without output stage).
I did not solder the output transistors because I am afraid these voltage differences might provoque an oscilation and burn the outputs.
I really need some help and advise.
Best Regards
Ricardo
RCruz,
I dont know how dennon have arranged things but surley the bias will not be correct until the output devices are fitted anyway as the whole circuit will adjust itself according to what is being sensed at the output??? I think if your sure everything else is o.k you need to fit the outputs and give it a try
Good luck,
Leigh
I dont know how dennon have arranged things but surley the bias will not be correct until the output devices are fitted anyway as the whole circuit will adjust itself according to what is being sensed at the output??? I think if your sure everything else is o.k you need to fit the outputs and give it a try
Good luck,
Leigh
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.