I've always felt that we could learn much about high performance audio design from the design of recording studio consoles.
One of my audio heros, Douglas Self, has been involved in the design of such.
I am hoping that there will be people reading this who have experience with recording consoles.
Let me give some examples of the things that I believe could be
informative.
1. Grounding scheme. This is complex, and must be well done.
2. Power supplies and voltages. Such consoles have outboard power supplies? I like this. I have read that they can go up to as much as 2KW. Are they linear or SMPS? What frequency of SMPS?
What voltages? Just two, plus and minus? Another for digital? The 48v fantom voltate. I have often thought that there could be slight design performance advantages for having two pairs of plus and minus voltages. The lower one would be for outputs of op amp circuits. The higer pair could be for the inputs. I believe there could be distortion improvements for line level. I also believe that you could maximize noise performance for some low level circuits be eliminating active current sources on the inputs.
I have not fully checked any of these out. But how do these consoles do it?
3. Overall architecture. How do you get the functionality without too many opamp stages?
4. Discrete vs monolithic? Jensen makes a discrete opamp module which runs on +/- 24 volts. The design of monolithic opamps really is an art. They have all sorts of proprietery distortion correction circuitry.
5. Inverting mode. Of course using inverting mode is a way to get low distortion. But where you have a level control, it usually has to feed a non-inverting mode opamp. Look at how they architect these consoles.
6. JFET inputs. These can be lower noise for high impedance. Are they used? Are they used anywhere else, like for their distortion characteristics?
7. Diffpair mirroring, meaning use two, one NPN and one PNP, to get even order distortion?
8. Circuits without diffpair inputs? These could lower noise for
single ended source. It could work when you don't have to have response to DC, and so you could have an op-amp with low freq gain in the feedback path, to give you your good DC offset, feeding back into a common emitter.
9. Moving magnet and moving coil phono inputs.
10. All the types of tone controls and dynamics management, and microphone preamp limiting.
11. Then all the new ADC issues too.
12. Thermal, heat sinks, LM394, biased how high? How much low z drive on outputs?
Anyone have knowledge of such consoles? Any good links?
cross posted at Audio Explortions, 10 groups. Come join in
http://groups.google.com/group/audioex_preamp_source?hl=en
http://groups.google.com/group/audioex_introduction_links?hl=en
One of my audio heros, Douglas Self, has been involved in the design of such.
I am hoping that there will be people reading this who have experience with recording consoles.
Let me give some examples of the things that I believe could be
informative.
1. Grounding scheme. This is complex, and must be well done.
2. Power supplies and voltages. Such consoles have outboard power supplies? I like this. I have read that they can go up to as much as 2KW. Are they linear or SMPS? What frequency of SMPS?
What voltages? Just two, plus and minus? Another for digital? The 48v fantom voltate. I have often thought that there could be slight design performance advantages for having two pairs of plus and minus voltages. The lower one would be for outputs of op amp circuits. The higer pair could be for the inputs. I believe there could be distortion improvements for line level. I also believe that you could maximize noise performance for some low level circuits be eliminating active current sources on the inputs.
I have not fully checked any of these out. But how do these consoles do it?
3. Overall architecture. How do you get the functionality without too many opamp stages?
4. Discrete vs monolithic? Jensen makes a discrete opamp module which runs on +/- 24 volts. The design of monolithic opamps really is an art. They have all sorts of proprietery distortion correction circuitry.
5. Inverting mode. Of course using inverting mode is a way to get low distortion. But where you have a level control, it usually has to feed a non-inverting mode opamp. Look at how they architect these consoles.
6. JFET inputs. These can be lower noise for high impedance. Are they used? Are they used anywhere else, like for their distortion characteristics?
7. Diffpair mirroring, meaning use two, one NPN and one PNP, to get even order distortion?
8. Circuits without diffpair inputs? These could lower noise for
single ended source. It could work when you don't have to have response to DC, and so you could have an op-amp with low freq gain in the feedback path, to give you your good DC offset, feeding back into a common emitter.
9. Moving magnet and moving coil phono inputs.
10. All the types of tone controls and dynamics management, and microphone preamp limiting.
11. Then all the new ADC issues too.
12. Thermal, heat sinks, LM394, biased how high? How much low z drive on outputs?
Anyone have knowledge of such consoles? Any good links?
cross posted at Audio Explortions, 10 groups. Come join in
http://groups.google.com/group/audioex_preamp_source?hl=en
http://groups.google.com/group/audioex_introduction_links?hl=en
The lowest distortions are obtained from OPamp like circuits. So it works out that it is mainly the distortion of the input diff pair.
But this is determined by many things, including loading and gain of the second stage.
The noise of such circuits is also very low, but there may be ways to go a bit lower.
Modern opamps have very very good specs, but not necessarily the best possible.
Still I wonder if some people are going discrete, thinking they are doing better, maybe based on simulations, but not really being better.
There are quite a number of circuit issues to consider.
I'm hoping there are some who have experience with analog recording consoles, maybe even schematics.
But this is determined by many things, including loading and gain of the second stage.
The noise of such circuits is also very low, but there may be ways to go a bit lower.
Modern opamps have very very good specs, but not necessarily the best possible.
Still I wonder if some people are going discrete, thinking they are doing better, maybe based on simulations, but not really being better.
There are quite a number of circuit issues to consider.
I'm hoping there are some who have experience with analog recording consoles, maybe even schematics.
<<
There is also the consideration that what can be done with chips won't necessarily apply to discretes, or designs at different voltage/current levels.
>>
Totally true. But it also cuts the other way. When is it best to use chips vs discretes, and at what voltage/current levels.
When you are willing to go to great expense, and to disipate lots of power, what will give you the lowest noise and lowest distortion.
I'm hoping someone might have links to web sites, or even manuals or schematics.
I have read of high performance recording consoles taking 2kVA for 32 channels.
Douglas Self has been involved in recording console design, and has engaged in this quest to get the best SNR and lowest distortion in the final product.
<<
Out of, I would guess, two dozen consoles I've had open, only one had anything other than opamps all the way down the line, and that had Jfets on the input only, and 741s everywhere else.
>>
I am surprised about the 741s.
I would wonder the date.
Jensen still offers discrete OP AMP modules which they say perform better than monolithics. I don't know if they still beat the newest monolithics.
Underlying my interest in this thread is, when can discretes still beat monolithics. It can't just be simulation results either.
JFETS? Do you mean monolithic opamps, or discretes? On what inputs? Usually it would only be for high impedance that you would bet better noise. Like High Z microphones, moving magnet phono, or elect. guitars.
Where these JFETs diff pairs, or single ended.
Using JFETs at line level for square law distortion is a controversial idea that many writters, like Doug Self, dismiss.
There is also the consideration that what can be done with chips won't necessarily apply to discretes, or designs at different voltage/current levels.
>>
Totally true. But it also cuts the other way. When is it best to use chips vs discretes, and at what voltage/current levels.
When you are willing to go to great expense, and to disipate lots of power, what will give you the lowest noise and lowest distortion.
I'm hoping someone might have links to web sites, or even manuals or schematics.
I have read of high performance recording consoles taking 2kVA for 32 channels.
Douglas Self has been involved in recording console design, and has engaged in this quest to get the best SNR and lowest distortion in the final product.
<<
Out of, I would guess, two dozen consoles I've had open, only one had anything other than opamps all the way down the line, and that had Jfets on the input only, and 741s everywhere else.
>>
I am surprised about the 741s.
I would wonder the date.
Jensen still offers discrete OP AMP modules which they say perform better than monolithics. I don't know if they still beat the newest monolithics.
Underlying my interest in this thread is, when can discretes still beat monolithics. It can't just be simulation results either.
JFETS? Do you mean monolithic opamps, or discretes? On what inputs? Usually it would only be for high impedance that you would bet better noise. Like High Z microphones, moving magnet phono, or elect. guitars.
Where these JFETs diff pairs, or single ended.
Using JFETs at line level for square law distortion is a controversial idea that many writters, like Doug Self, dismiss.
The Jensen discrete OP-AMPs that I know of are the same ones from 20 years ago. They run on +/- 24 volts, instead of +/-15
They use big LM394 diff pairs.
You wouldn't want to rebias the outputs higher because they are thermally designed to work the way they are.
You could design something new, with bigger output devices and bigger heat sinks.
There are monolithic class A audio op amps on the market. Their signal drive capability is not that great.
Fact is going class A on the outputs of such audio gain stages does not do that much to futher lower distortion.
Opamp distortion is already very low, but not zero. Mostly it is the distortion of the input diffpair.
You can write out the closed loop equations and see this.
In something like a recording console, the biggest challenge is to get the SNR that they want. So the tendency will be to favor larger signal swings.
While distortion is not a real problem, its not zero. If you push signal swings enough, to get the greater SNR, there will be some distortion.
The output of an opamp is the difference between the inverting and noninverting inputs. So how much distortion is there? How linear is the differencing part of the circuit??
Well, it depends on how much the current through the diffpair changes over the signal swing.
Say you have a 10 volt signal swing on the noninverting input.
The opamps high gain will dictate that the inverting input also follows this 10 volt signal swing.
That means that the common emitter node of the diffpair will experience such a 10volt swing.
Here is where the noninverting mode distortion comes in.
If the common emitter current moves, at all, there will be a slight change in transistor Vbe.
This is a nonlinearity, in the differencing.
It is the same issue that occurs in an emitter follower gain stage. There is distortion introduced in such, because Vbe changes.
If you have an emitter follower driving a current source, instead of a resistive load, the distortion is less, but still not zero.
What about in the inverting mode? When then your differencing is via two resistors making a current summation. So this differencing is resolved, before it ever hits the transistor diffpair.
Then common emitter node really doesn't move at all.
You can see in text books or on data sheets, the noninverting mode distortion is low. But inverting mode is about another decade lower.
Mikeks, who's picture is that you posted?
ZMB
They use big LM394 diff pairs.
You wouldn't want to rebias the outputs higher because they are thermally designed to work the way they are.
You could design something new, with bigger output devices and bigger heat sinks.
There are monolithic class A audio op amps on the market. Their signal drive capability is not that great.
Fact is going class A on the outputs of such audio gain stages does not do that much to futher lower distortion.
Opamp distortion is already very low, but not zero. Mostly it is the distortion of the input diffpair.
You can write out the closed loop equations and see this.
In something like a recording console, the biggest challenge is to get the SNR that they want. So the tendency will be to favor larger signal swings.
While distortion is not a real problem, its not zero. If you push signal swings enough, to get the greater SNR, there will be some distortion.
The output of an opamp is the difference between the inverting and noninverting inputs. So how much distortion is there? How linear is the differencing part of the circuit??
Well, it depends on how much the current through the diffpair changes over the signal swing.
Say you have a 10 volt signal swing on the noninverting input.
The opamps high gain will dictate that the inverting input also follows this 10 volt signal swing.
That means that the common emitter node of the diffpair will experience such a 10volt swing.
Here is where the noninverting mode distortion comes in.
If the common emitter current moves, at all, there will be a slight change in transistor Vbe.
This is a nonlinearity, in the differencing.
It is the same issue that occurs in an emitter follower gain stage. There is distortion introduced in such, because Vbe changes.
If you have an emitter follower driving a current source, instead of a resistive load, the distortion is less, but still not zero.
What about in the inverting mode? When then your differencing is via two resistors making a current summation. So this differencing is resolved, before it ever hits the transistor diffpair.
Then common emitter node really doesn't move at all.
You can see in text books or on data sheets, the noninverting mode distortion is low. But inverting mode is about another decade lower.
Mikeks, who's picture is that you posted?
ZMB
Re: Re: learn from recording consoles?
True but wrong - or incomplete
noninverting amp transfer function has a common mode gain term that "escapes" the negative feedback loop's correction
input referred dVos with Vin_cm is in series with the input V, the loop transmission dosen't "see" it
fet input op amps show a nonlinear C to the substrate that increases distortion with the fets moving with Vin in noninverting configuration - nonlinear input current flowing in differing source and feedback Z produces a differential V that is largely insensitive to the loop transmission
Jim Williams: "always invert"
my sims of the JE990 show it to have particularly poor cm and psrr, the input diff pair for all of its being a sexy "super match" type isn't being ran with matching Ic
mikeks said:
True but wrong - or incomplete
noninverting amp transfer function has a common mode gain term that "escapes" the negative feedback loop's correction
input referred dVos with Vin_cm is in series with the input V, the loop transmission dosen't "see" it
fet input op amps show a nonlinear C to the substrate that increases distortion with the fets moving with Vin in noninverting configuration - nonlinear input current flowing in differing source and feedback Z produces a differential V that is largely insensitive to the loop transmission
Jim Williams: "always invert"
my sims of the JE990 show it to have particularly poor cm and psrr, the input diff pair for all of its being a sexy "super match" type isn't being ran with matching Ic
I would love to see your schematics!
I learn tremendously from studying schematics!
( You pictures back on the transformerless thread were fantastic! )
Please do post them. Also, I would love to really study them. To get them full sized, could you email them to me too?
brian.zenmaster@gmail.com
Audio Explorations, thread reopened!
http://groups.google.com/group/audioex_power_supplies/browse_thread/thread/dfd47387a97664c5?hl=en
I learn tremendously from studying schematics!
( You pictures back on the transformerless thread were fantastic! )
Please do post them. Also, I would love to really study them. To get them full sized, could you email them to me too?
brian.zenmaster@gmail.com
Audio Explorations, thread reopened!
http://groups.google.com/group/audioex_power_supplies/browse_thread/thread/dfd47387a97664c5?hl=en
Good consoles have very few active elements per signal path.
Most critical part of each console is a mic amp that may be substituted by an external one if needed. It have to have output to drive 600 Ohm input with +4 dB level plus headroom, usually +18 dB.
EQs and mixers on +4 dB work easilly.
Tascam consoles have -10 dB level instead of +4 dB, and they are more noisy.
Most critical part of each console is a mic amp that may be substituted by an external one if needed. It have to have output to drive 600 Ohm input with +4 dB level plus headroom, usually +18 dB.
EQs and mixers on +4 dB work easilly.
Tascam consoles have -10 dB level instead of +4 dB, and they are more noisy.
Going back to post 10, and about other reasons to use JFETs.
Most of the authors I admire reject the argument of JFET inputs for square law distortion.
So the other reasons are lower noise. But this only applies to high impedance inputs.
This would mean moving magnet phono, high z microphone, the internal microphone preamps in concensor mikes, and maybe electric guitar pickups.
In a recording studio this is limited. It certainly does not apply to things driven by opamp outputs.
So, other reasons for using JFETs, such as??
As far as the JE990, my perception is that there could be advantages and disadvantages of a discrete opamp.
Monolithic OPAMP circuits are extremely complex. They have very sophisticated current sourcing. Some have distortion correction circuits of a confidential nature in the emitter pull down portion of the diffpair. Some even couple the emitter node to other things, to try to do some correction.
Monolithic opamp design really has been pushed.
I would still say there could be advantages in going discrete:
1. Able to run hotter. Bigger parts, heat sinks, lower impedances, higher bias at all stages.
2. Could use higher voltages.
3. Could have decoupled power for different stages, even different voltage rails.
But does this beat monolithic opamps in the end? In what applications?
That is why I am asking about recording consoles? When price is not an object and you are willing to blow 2kw away in heat, which approaches do you use.
Inverting is clearly best. But, for a gain control that fads to zero, you really need noniverting.
I believe bandaxal tone controls and graphic eq. need noninverting too.
Salen-Key filters, the most common type, need non-inverting.
But there is a less common form of the circuit that works with inverting. It is not widely known.
If someone can post a reference, I would like it.
For phono stage, people debate inverting v noninverting. There, distortion is minimal because of the small signal. But the noise ends up colored differently. Most use noninverting.
I would like to see the architectures of recording consoles. I believe that a true high performance control preamp could be desiged with what is gleaned by studying recording consoles.
These monolithic v discrete issues are also being considered here:
http://groups.google.com/group/audioex_preamp_source?lnk=li&hl=en
Most of the authors I admire reject the argument of JFET inputs for square law distortion.
So the other reasons are lower noise. But this only applies to high impedance inputs.
This would mean moving magnet phono, high z microphone, the internal microphone preamps in concensor mikes, and maybe electric guitar pickups.
In a recording studio this is limited. It certainly does not apply to things driven by opamp outputs.
So, other reasons for using JFETs, such as??
As far as the JE990, my perception is that there could be advantages and disadvantages of a discrete opamp.
Monolithic OPAMP circuits are extremely complex. They have very sophisticated current sourcing. Some have distortion correction circuits of a confidential nature in the emitter pull down portion of the diffpair. Some even couple the emitter node to other things, to try to do some correction.
Monolithic opamp design really has been pushed.
I would still say there could be advantages in going discrete:
1. Able to run hotter. Bigger parts, heat sinks, lower impedances, higher bias at all stages.
2. Could use higher voltages.
3. Could have decoupled power for different stages, even different voltage rails.
But does this beat monolithic opamps in the end? In what applications?
That is why I am asking about recording consoles? When price is not an object and you are willing to blow 2kw away in heat, which approaches do you use.
Inverting is clearly best. But, for a gain control that fads to zero, you really need noniverting.
I believe bandaxal tone controls and graphic eq. need noninverting too.
Salen-Key filters, the most common type, need non-inverting.
But there is a less common form of the circuit that works with inverting. It is not widely known.
If someone can post a reference, I would like it.
For phono stage, people debate inverting v noninverting. There, distortion is minimal because of the small signal. But the noise ends up colored differently. Most use noninverting.
I would like to see the architectures of recording consoles. I believe that a true high performance control preamp could be desiged with what is gleaned by studying recording consoles.
These monolithic v discrete issues are also being considered here:
http://groups.google.com/group/audioex_preamp_source?lnk=li&hl=en
Waveborne, If you have any links with specs, panel layouts, or manuals, or especially schematics, I'd love to see them.
Are you talking about contemporary Tascam, or historical?
The issue is always how many controls do you need. How any circuit stages to get utility, without compromising noise and distortion.
Thats why I want to look at examples
ZMB
Are you talking about contemporary Tascam, or historical?
The issue is always how many controls do you need. How any circuit stages to get utility, without compromising noise and distortion.
Thats why I want to look at examples
ZMB
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.