I think that single ended amplifiers are dominated by second harmonic.
Push Pull amplifiers can be designed to cancel the second harmonic and then the third harmonic begins to dominate the output.
With both types using the amplifiers with maximum signals that are well below clipping levels massively reduces the output distortion.
I typically aim for 10dB to 20dB of overhead to minimise distortion. This is the ratio of maximum signal to clipping level.
Not the same as power amplifiers where average level to clipping levels are important to avoid clipping rather than to avoid distortion.
Many don't agree.
eg an opamp passing a maximum CDP signal of 2.2Vac needs +-11Vdc supply rails to reach a +10dB overhead.
A +20dB overhead requires a discrete opamp on +-24Vdc supply rails.
Push Pull amplifiers can be designed to cancel the second harmonic and then the third harmonic begins to dominate the output.
With both types using the amplifiers with maximum signals that are well below clipping levels massively reduces the output distortion.
I typically aim for 10dB to 20dB of overhead to minimise distortion. This is the ratio of maximum signal to clipping level.
Not the same as power amplifiers where average level to clipping levels are important to avoid clipping rather than to avoid distortion.
Many don't agree.
eg an opamp passing a maximum CDP signal of 2.2Vac needs +-11Vdc supply rails to reach a +10dB overhead.
A +20dB overhead requires a discrete opamp on +-24Vdc supply rails.
Last edited:
Yes but i have also learned that 2nd order is nice on the ear.
I think that also the original BOZ can have some 2nd order distortion, but i am sure is very very musical.
And it is very simple also.
I understand that some of the PS noise in the original is filtered by the bias circuit instead here is not filtered
So the construction of a low noise PS is mandatory ... every noise reaching the mosfet will be then amplified by the following power amp.
Thanks, gino
I think that also the original BOZ can have some 2nd order distortion, but i am sure is very very musical.
And it is very simple also.
I understand that some of the PS noise in the original is filtered by the bias circuit instead here is not filtered
So the construction of a low noise PS is mandatory ... every noise reaching the mosfet will be then amplified by the following power amp.
Thanks, gino
Hi and thanks but i have to wait this evening.
I assume you are referring at the .asc file, the modified one.
I do not have LTSpice here. But it looks like the simulation works.
I will check this evening.
May main goals are:
1) reduce out impedance at around 500 ohm
2) reduce voltage gain at around 2 ... more than enough for a line stage
If the thump will be not dangerous for the mosfet no problem here.
Usually i turn on the preamp and after some seconds the power amp.
And viceversa when i shut the system down.
It is the normal procedure i guess.
But if this will cause damage to the mosfet i have to do something.
Problem is that the bias circuit must be re-designed.
Very difficult task for me.
If the simple voltage divider will work why not using it ?
Just two resistors ...
This evening i will check the circuit.
Thanks again, gino
Hi and thanks and let me elaborate a little.
I have noticed a strange form of attitude around.
In tube designs almost nobody care about distortion.
Sometimes it even seems that the more the better.
I see a complete different attitude between tubists and solid statists.
When solid state is involved everyone talk about distortion figures.
If we accept the idea that a little (like 0.1%) of 2nd order distortion is not bad for the overall musical outcome why not use very simple topologies also with solid state, as it is the norm with tubes ?
Then someone also put the question if the real distortion we hear is the one of the speakers and not the one of the electronics. Another interesting issue.
I had the opportunity to listen to a prototype of this BOZ circuit.
My impression was that with the right PS (and this is the tricky part) it can sound very very convincing. Much more than the average line stage.
I remember in particular a very nice 3D soundstage and also an immediacy in sound that gave a very strong live effect.
So i think that its potential is very very high.
The PS is fundamental. The better the PS the better the sound.
Do you think that the .asc circuit can work ?
Thanks again, gino
The BOZ was my first DIY preamp. I found my build to be a bit to zingy in its presentation - to lively. Mosfets in simple circuits do tend to have a Mosfet sound (not present in more complex mosfet circuits). I moved onto valve preamps soon after and have never looked back, Allen Wrights SLCF was far more neutral to my own ears. However my latest project will be a Jfet - Mosfet SLCF and I will soon find out if Fets can match the neutrality of well implemented valve circuits.
I suspect that the gate capacitance is the source of the Mosfet sound, and the fact that it varies with signal. I would therefore always strive to use the Mosfet with the absolute lowest capacitance available.
Shoog
I suspect that the gate capacitance is the source of the Mosfet sound, and the fact that it varies with signal. I would therefore always strive to use the Mosfet with the absolute lowest capacitance available.
Shoog
Hi i think that a lot can be related to the execution
Layout, parts selection, power supply quality ... there are many variables that can impact the sound quality.
As reported in the article the noise coming from the PS mixes at the output with the signal.
So may be the "zingy" character could have been due to some diodes switching noise in the PS.
I understand some specific diodes are immune to this.
I am particularly intrigued by this project. I heard a very nice 3D soundstage for instance and much less flat sound that from the average solid state line preamp (i have some commercial units and they are quite flatter than the BOZ).
I have also a general comment.
When i look for instance a tube line stage i see a very simple almost textbook schematic.
Instead i have here now a solid state line preamp that i would like to recapp ... and there are tenths of active devices in each channel.
This i do not understand.
Trivially speaking i would use a component within its limits of course and in its best working conditions where it is more linear.
I have modified a little the circuit and simulate. Apart some little 2nd order distortion quite acceptable it looks fine.
I think that the real secret is in the power supply execution.
That will make the sound in the end. Maybe i am wrong.
Thanks again, gino
Layout, parts selection, power supply quality ... there are many variables that can impact the sound quality.
As reported in the article the noise coming from the PS mixes at the output with the signal.
So may be the "zingy" character could have been due to some diodes switching noise in the PS.
I understand some specific diodes are immune to this.
I am particularly intrigued by this project. I heard a very nice 3D soundstage for instance and much less flat sound that from the average solid state line preamp (i have some commercial units and they are quite flatter than the BOZ).
I have also a general comment.
When i look for instance a tube line stage i see a very simple almost textbook schematic.
Instead i have here now a solid state line preamp that i would like to recapp ... and there are tenths of active devices in each channel.
This i do not understand.
Trivially speaking i would use a component within its limits of course and in its best working conditions where it is more linear.
I have modified a little the circuit and simulate. Apart some little 2nd order distortion quite acceptable it looks fine.
I think that the real secret is in the power supply execution.
That will make the sound in the end. Maybe i am wrong.
Thanks again, gino
SS devices have extremely high gains and are generally very unlinear. Typically small signal transistors have gains of 100x when we are only looking for an overall gain of 2-5x. This means that it is very difficult to translate the very simple circuits seen in valve preamps into an equivalent SS preamp. Generally the only way to cure the excessive gain and the unlinearity is to apply large amounts of negative feedback which will produce that flat sound which is typical of most SS preamps. Coupled to this is the relative cost of transistors (cents compared to dollars for valves) and the temptation is always there to solve all problems with an extra gain stage and then linearize it with with more feedback - its a habit which is difficult to cure.
The BOZ has relatively little global feedback so it retains the air and space associated with zero feedback valve preamps. However the none linear gate capacitance will have an effect and the only way to solve that is to up the feedback with a reduction in the qualities we like.
Shoog
The BOZ has relatively little global feedback so it retains the air and space associated with zero feedback valve preamps. However the none linear gate capacitance will have an effect and the only way to solve that is to up the feedback with a reduction in the qualities we like.
Shoog
Last edited:
Hi and thanks a lot fo the very valuable advice
So your opinion is after the changes above mentioned to keep every other things equal ?
I think it is a very reasonable approach because in this way some filtering effect is mantained (like R103-C102).
Thanks a lot again. Kind regards, gino
Hi and thanks for the helpful reply. This is indeed the neverending debate going on in the audio circles.
This is where the perversion is. Because it can lead to long transistors rows like in my present preamp.
I am sure that even staying with solid state a better and simpler preamp exists ... so this is wrong.
For instance i liked better the sound of the BOZ in comparison to some much more complex ss preamp i listened to ... this is wrong design for me.
I do not understand well which performance of the preamp the non linear gate capacitance will impact.
Could you tell me a little more ?
Thanks again, gino
Member
Joined 2009
Paid Member
The overall gain at higher frequencies will vary with both signal level and frequency. Depending on the circuit this maybe in the audio band and will represent a form of distortion which will not be readily measurable at static test signal levels. Remember this capacitance is not what is quoted on the datasheet but is multiplied by the stage gain which can make it very high indeed - as you can imagine some of the switcing FET's have gate capacitances in the nanoFarad range which when multiplied by the stage gain starts to look very significant indeed for your high frequency performance.
I postulate, but cannot confirm, that this is responsible for the characteristic Mosfet sound - which once you have heard you will never miss again.
The solution is to use a very low gate capacitance FET, or apply lot of feedback to mask it, or only use your FET's in source follower configuration where the gate to source capacitance is lowest.
Another approach, and what a professional designer would do, is use a beefy pre-driver with minimal gate capacitance - probably in source follower configuration, to squash the gate capacitance of the gain stage you are looking to get your gain out of. So a single simple gainstage morphs into a compound two stage design to achieve the same outcome. KISS has left the building.
Shoog
Hello !
I do not know if this topic is always followed but I wanted to share my experiences with the gate capacitance.
I made a BOZ in compliance with the original schematic but adding a potentiometer at input.
I heard the sound changed depending on the setting of the input potentiometer.
I connected the oscilloscope yesterday and I was very surprised by observing what was going on with a square 100 khz signal:
-With the potentiometer set to maximum, the square is clean, very slightly rounded.
-With the potentiometer set approximately 70% of the maximum, the square has become almost a triangle!
-With the potentiometer set approximately 35% of the maximum, the square is clean, very slightly rounded but not as good as maximum.
Is this is the gate capacitance is responsible?
I think so, yes, but I would like your opinion.
I tried a irf610, 510 and 510 has better sound but I find treble are less defined.
Gate capacitance of 510 135 pf
Gate capacitance of 610: 140 pf
I think trying the 2SK234: 90 pf
I do not know if this topic is always followed but I wanted to share my experiences with the gate capacitance.
I made a BOZ in compliance with the original schematic but adding a potentiometer at input.
I heard the sound changed depending on the setting of the input potentiometer.
I connected the oscilloscope yesterday and I was very surprised by observing what was going on with a square 100 khz signal:
-With the potentiometer set to maximum, the square is clean, very slightly rounded.
-With the potentiometer set approximately 70% of the maximum, the square has become almost a triangle!
-With the potentiometer set approximately 35% of the maximum, the square is clean, very slightly rounded but not as good as maximum.
Is this is the gate capacitance is responsible?
I think so, yes, but I would like your opinion.
I tried a irf610, 510 and 510 has better sound but I find treble are less defined.
Gate capacitance of 510 135 pf
Gate capacitance of 610: 140 pf
I think trying the 2SK234: 90 pf
a 100kHz square wave is way above audio frequencies we can hear........
use 1kHz or 10kHz waves.
the potentiometer at the input of the BOZ and the gate-capacitance create a RC- filter.........a 100k - pot has a max. output res of 25k (this is much!). with 140pF gate cap. you get around 45kHz @ -3dB.....
you can use a pot with 10k or 25k (or as cheap solution build in a res 10k from wiper to ground)
use 1kHz or 10kHz waves.
the potentiometer at the input of the BOZ and the gate-capacitance create a RC- filter.........a 100k - pot has a max. output res of 25k (this is much!). with 140pF gate cap. you get around 45kHz @ -3dB.....
you can use a pot with 10k or 25k (or as cheap solution build in a res 10k from wiper to ground)
I agree that 100 kHz is well above what the ear can hear, but I heard it. The rise time of the signal is really affected.
Even on a square wave signal at 1 KHz, that is visible on the oscilloscope.
If I decrease the input resistance, I will overload the source and affect the bandwidth.
I noticed that it also depends on the source. It's worse with my function generator (50 ohm output impedance). I have a better measure with my dac (4 x 1543 non oversampling). This is amazing.
I see no difference in the rise time regardless of the output volume.
I think we should really avoid putting a potentiometer at the input.
Do you know a simple way to reduce the gain of the preamp?
Guillaume
Even on a square wave signal at 1 KHz, that is visible on the oscilloscope.
If I decrease the input resistance, I will overload the source and affect the bandwidth.
I noticed that it also depends on the source. It's worse with my function generator (50 ohm output impedance). I have a better measure with my dac (4 x 1543 non oversampling). This is amazing.
I see no difference in the rise time regardless of the output volume.
I think we should really avoid putting a potentiometer at the input.
Do you know a simple way to reduce the gain of the preamp?
Guillaume
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.