Hi,
for unity gain the 5534 needs about 22pF as Ccomp.
for gains greater than 5, Ccomp=opencircuit.
Is there a way to determine the value of Ccomp for gains between 1 and 5?
Is there any advantage in reducing Ccomp to the lowest value that consistently achieves stability?
for unity gain the 5534 needs about 22pF as Ccomp.
for gains greater than 5, Ccomp=opencircuit.
Is there a way to determine the value of Ccomp for gains between 1 and 5?
Is there any advantage in reducing Ccomp to the lowest value that consistently achieves stability?
Intermediate gains require an intermediate value, I would try 10 or 12pF for x2 and 4p7 or 6p8 for x3
If you overcompensate with 22pF for x2 or x3 you waste bandwidth
If you overcompensate with 22pF for x2 or x3 you waste bandwidth
Hi all,
David seems to be indicaing that there is an advantage in keeping the Ccomp low.
His suggested values also fit a model:- Ccomp~=22pF/gain.
Any other comments?
David seems to be indicaing that there is an advantage in keeping the Ccomp low.
His suggested values also fit a model:- Ccomp~=22pF/gain.
Any other comments?
I do not think there is some formula for best Ccomp value.
Because different circuits, different loads and each individual NE5534 can have slight variations.
To set the lowest possible capacitance can only be done by using oscilloscope.
As you can see NE5534 is already compensated internally.
With 12 pF capacitor, that will be in parallel with external.
Across pins 5 and 8.
NE5534 does not need any external comp for gain >= 5
NE5534 will be stable at unity gain = 1, with 22 pF
Gain, Int, Ext cap
x5 = 12pF
x3 = 12pF + ??
x1 = 12pF + 22pF
Because different circuits, different loads and each individual NE5534 can have slight variations.
To set the lowest possible capacitance can only be done by using oscilloscope.
As you can see NE5534 is already compensated internally.
With 12 pF capacitor, that will be in parallel with external.
Across pins 5 and 8.
NE5534 does not need any external comp for gain >= 5
NE5534 will be stable at unity gain = 1, with 22 pF
Gain, Int, Ext cap
x5 = 12pF
x3 = 12pF + ??
x1 = 12pF + 22pF
Attachments
Hi,
I am begining to understand the workings of 5534 as well as opamps generally.
My current power amp uses 5534 gain of x2 with Ccomp=22pF as a pre conditioning stage (cap coupled at in & out) before the main 3stage amp.
I am pondering further modifications.
I am begining to understand the workings of 5534 as well as opamps generally.
My current power amp uses 5534 gain of x2 with Ccomp=22pF as a pre conditioning stage (cap coupled at in & out) before the main 3stage amp.
I am pondering further modifications.
If you use 22 pF you won't have to think. It just works. If you want to tune the value put in square wave and observe when you'll get a good output signal.
Is this part of an overall feedback loop?
If it is then disturbing the opamp bandwidth will have consequences.
If it is then disturbing the opamp bandwidth will have consequences.
AndrewT said:
Yes...you will have more bandwidth and more feedback in the high frequencies , with the consequent reduction in distortion...
The NE5534 is a great Op-Amp...😉
Hi all,
thanks so far.
the times 2 gain at the front end is inside it's own feed back loop and completely separate from the global loop around the 3stage amp that follows. There is DC block & RF filter between the two gain stages as well as DC block at the input.
I am almost set up to post the schematic but the 1.7Mb 🙁 pic does not fit my floppy from work to home where I can compress it for posting. Maybe it would be easier to draw in schematic software
thanks so far.
the times 2 gain at the front end is inside it's own feed back loop and completely separate from the global loop around the 3stage amp that follows. There is DC block & RF filter between the two gain stages as well as DC block at the input.
I am almost set up to post the schematic but the 1.7Mb 🙁 pic does not fit my floppy from work to home where I can compress it for posting. Maybe it would be easier to draw in schematic software
AndrewT said:
And why not to use the op amp at a gain of 5 or 6 (without compensation ) and reducing the gain a little of the next ( I presume discrete ) amp ?
Re: Re: NE5534 compensation ?
With compensation 0.0015% dist and without 0.0006%.
Page 6.54 in the book "Opamp applications" edited by Walt Jung
Please put things into perspective.Tube_Dude said:
With compensation 0.0015% dist and without 0.0006%.
Page 6.54 in the book "Opamp applications" edited by Walt Jung
Re: Re: Re: NE5534 compensation ?
0,0015% / 0,0006 is 2.5 times less distortion, for free ... 😉
peranders said:
0,0015% / 0,0006 is 2.5 times less distortion, for free ... 😉
Peranders & Tube dude,
keep talking, don't fall out over me.
Tube,
gain of five or slightly more and then include it inside the global NFB loop using W.Jung topology (composite line driver 1996 & buffers 1998) is something I have in mind for later.
Just now reducing the Ccomp from 22pF to 10pF might be where I go first.
Just to confirm something I said in another thread: inject the 10kHz square wave AFTER the first DC block and RF filter.
Examine what happens on the oscilloscope into 8R and 8R//47nF upto 2200nF or is 10kHz sqr into cap loaded output asking too much? or just use sinewave into 8R//Cap? and save the sqr for 8R?
keep talking, don't fall out over me.
Tube,
gain of five or slightly more and then include it inside the global NFB loop using W.Jung topology (composite line driver 1996 & buffers 1998) is something I have in mind for later.
Just now reducing the Ccomp from 22pF to 10pF might be where I go first.
Just to confirm something I said in another thread: inject the 10kHz square wave AFTER the first DC block and RF filter.
Examine what happens on the oscilloscope into 8R and 8R//47nF upto 2200nF or is 10kHz sqr into cap loaded output asking too much? or just use sinewave into 8R//Cap? and save the sqr for 8R?
AndrewT said:
I think is asking too much , 2.220 nF is 2.2 uF and this capacitance , has ~1,5 Ohms impedance at 50KHz and ~ 0,75 Ohms at 100 KHz.
As the 10KHz square wave have many high order harmonics the test is to punishing .
More, if the open loop output impedance is not low enough , the added phase shift induced by the capacitance at the output , will introduce another pole and the amp can become Joan D'Arc and "burn into flames"...
Anyway , usually the only capacitance that the speaker "see" is the speaker cable . because the dynamic speaker is a inductive load and even the capacity of the electrostatic speakers become inductive in the high frequencies , because the input transfomer.
Only a electrostatic speaker direct feed (without the input transformer) will be a purely capacitive load,I'm afraid..😉
If the 2nd and 3rd stage are independent then you are only interested in the output of the 2nd stage. Apply a square wave and take a good look at the settling time.
I believe in keeping signal levels between stages as high as possible, it reduces effects of noise and distortion on the ground.
If you can run your NE5534 x5 then the compensation tracks can be cut and that one entry for stray pick up gone.
When I worked at McMichaels making broadcast monitors we had picture disturbance that turned out to be a TI 741 in a low frequency circuit that was picking up a magnetic field. TI used ferrous lead frames. Opamps can do strange things.
I believe in keeping signal levels between stages as high as possible, it reduces effects of noise and distortion on the ground.
If you can run your NE5534 x5 then the compensation tracks can be cut and that one entry for stray pick up gone.
When I worked at McMichaels making broadcast monitors we had picture disturbance that turned out to be a TI 741 in a low frequency circuit that was picking up a magnetic field. TI used ferrous lead frames. Opamps can do strange things.
davidsrsb said:
Maybe that's the reason why resistors with ferrous leads and end caps , don't sound so good...
Hi David,
can you explain this
Are you saying inject the test signal into the VAS base?
and examine the collector output?
reminder: DC block, opamp gain times 2, DC block, LTP, VAS, EFdriver, Source Followers NFB returns to LTP inverting input.
can you explain this
Are you saying inject the test signal into the VAS base?
and examine the collector output?
reminder: DC block, opamp gain times 2, DC block, LTP, VAS, EFdriver, Source Followers NFB returns to LTP inverting input.
davidsrsb said:
Not enterily true. For slow IC's it doesn't matter and NE5534 is "slow".
EDIT: Andrew!
- Status
- Not open for further replies.