I did a first test of my x-ed cascoded BOSOZ with CCS based on the the design of Metalman and the PCBs from twisted pear audio.
Some changes I made to the orignial cuircit:
o The voltage references are normal zener diodes and have 47uF across them. Zener current is about 5mA.
o Just one reference voltage for both cascode Fets (thought this is better in terms of noise when using balanced out).
o Gain reduced to 2.2 by feedback resistors R18/R19(22K).
o 30k pot at the unbal input and use the balanced outputs and just one PSU for both channels.
o un-used neg input is grounded.
o The resistors in the RC network in the PSU are reduced (4R7 in PSU and 10R on board (R3)) to get 60V with the transformer I already have.
0 Bias current is 90mA per channel.
Anyway, all in all nice sounding amp but the noise could be lower and it is slightly humming. I can't hear it at my listening postition but it is noticeable if you are closer than 0.5m (2 feet) to the (low sensitivity) loudspeaker. The hiss does not depend on input pot position.
As this was just a test rig with probalby non optimal ground scheme I am not too worried about the hum (yet).
But the hiss could be lower as I am planing to go active in the long term. I know I will attenuate the noise if I used a bal pot at the output but I think the better approach is to minimise the intrinsic hiss.
Now a couple of questions:
Is the input/feedback network important for the noise perfomance?
If so, could it be reduced to let's say 3k9/1K e.g. (gain is then again 3.9)?
Would a better (higher values for RC filters) or double PSU help in terms of hiss? It will reduce hum but I am not so sure about hiss.
Any comments are welcome.
Thanks
Joe
Some changes I made to the orignial cuircit:
o The voltage references are normal zener diodes and have 47uF across them. Zener current is about 5mA.
o Just one reference voltage for both cascode Fets (thought this is better in terms of noise when using balanced out).
o Gain reduced to 2.2 by feedback resistors R18/R19(22K).
o 30k pot at the unbal input and use the balanced outputs and just one PSU for both channels.
o un-used neg input is grounded.
o The resistors in the RC network in the PSU are reduced (4R7 in PSU and 10R on board (R3)) to get 60V with the transformer I already have.
0 Bias current is 90mA per channel.
Anyway, all in all nice sounding amp but the noise could be lower and it is slightly humming. I can't hear it at my listening postition but it is noticeable if you are closer than 0.5m (2 feet) to the (low sensitivity) loudspeaker. The hiss does not depend on input pot position.
As this was just a test rig with probalby non optimal ground scheme I am not too worried about the hum (yet).
But the hiss could be lower as I am planing to go active in the long term. I know I will attenuate the noise if I used a bal pot at the output but I think the better approach is to minimise the intrinsic hiss.
Now a couple of questions:
Is the input/feedback network important for the noise perfomance?
If so, could it be reduced to let's say 3k9/1K e.g. (gain is then again 3.9)?
Would a better (higher values for RC filters) or double PSU help in terms of hiss? It will reduce hum but I am not so sure about hiss.
Any comments are welcome.
Thanks
Joe
Attachments
It might be zener noise, or it might be noisy Mosfets. The
IRF610's and such are not the quietest parts in the world.
As to the zeners, placing a cap directly across them is not
as effective as you might imagine. You can get lower noise
from them by filtering the voltage they generate by a simple
RC circuit before giving the voltage to the Gates.
😎
IRF610's and such are not the quietest parts in the world.
As to the zeners, placing a cap directly across them is not
as effective as you might imagine. You can get lower noise
from them by filtering the voltage they generate by a simple
RC circuit before giving the voltage to the Gates.
😎
Joe,
The hiss you are experiencing is something I have on my original. I've played with filtering the reference voltage zener diodes, but this didn't impact the level of hiss. I concluded (though I haven't proved this to be the case) that the noise is MOSFET hiss. This is the reason that I use a balanced volume pot on the outputs, as this attenuates the noise well below the audible level.
The only catch with putting the volume pot on the outputs is you want a relatively low impedance pot. 10k-ohms works well, but I found 5k-ohms seemed to be about ideal.
Cheers,
Terry
The hiss you are experiencing is something I have on my original. I've played with filtering the reference voltage zener diodes, but this didn't impact the level of hiss. I concluded (though I haven't proved this to be the case) that the noise is MOSFET hiss. This is the reason that I use a balanced volume pot on the outputs, as this attenuates the noise well below the audible level.
The only catch with putting the volume pot on the outputs is you want a relatively low impedance pot. 10k-ohms works well, but I found 5k-ohms seemed to be about ideal.
Cheers,
Terry
Thanks Nelson, Metalman,
Will try to measure the noise level with my sound card and try better zener filtering. In the long term a balanced pot at the output seems the way to go.
Volume control at the input was just more handy as I already had the parts lying around.
Metalman,
Did you use a 5K pot or a stepped attenuator?
How "useful" is the characteristic in real life when using a standard (log or lin) pot?
All the best
joe
One more, will JFETs help? (with adjusted voltages and surrents of course)
Will try to measure the noise level with my sound card and try better zener filtering. In the long term a balanced pot at the output seems the way to go.
Volume control at the input was just more handy as I already had the parts lying around.
Metalman,
Did you use a 5K pot or a stepped attenuator?
How "useful" is the characteristic in real life when using a standard (log or lin) pot?
All the best
joe
One more, will JFETs help? (with adjusted voltages and surrents of course)
You need 5K Log on the output to get rid of the hiss, if you want to maintain the original mosfets.
If you feew adventurous, you could go for a cascoded jfet solution. Those are sure silent.
Another way of getting rid of the zener noise, is to simply use LED's instead for the voltage ref., like for instance seen in the Pumpkin by Zen Mod.
Magura 🙂
If you feew adventurous, you could go for a cascoded jfet solution. Those are sure silent.
Another way of getting rid of the zener noise, is to simply use LED's instead for the voltage ref., like for instance seen in the Pumpkin by Zen Mod.
Magura 🙂
- Status
- Not open for further replies.