Yup, agree uneqivocally Nico,
The reality is that simulation is just that - a fair stab at reality with untested models and a simplified program, a guide, and not much more.
Right now I'm doing compensation regimes for a power amp, and finding that the results of the simulations are at best a guide. The ear seems to be good at discerning tiny changes, too. Not very scientific, but the changes cause such sonic variations that telling the differences is easy.
The basic principles are elucidated by the simulator, but you still need to know precisely what is going. I'm using nested feedback, lag comp and phase lead, and the relationships are complex.
Cheers,
Hugh
The reality is that simulation is just that - a fair stab at reality with untested models and a simplified program, a guide, and not much more.
Right now I'm doing compensation regimes for a power amp, and finding that the results of the simulations are at best a guide. The ear seems to be good at discerning tiny changes, too. Not very scientific, but the changes cause such sonic variations that telling the differences is easy.
The basic principles are elucidated by the simulator, but you still need to know precisely what is going. I'm using nested feedback, lag comp and phase lead, and the relationships are complex.
Cheers,
Hugh
Hugh, - we use simulation extensively in our products design and research phases, in fact CAD/CAM is an integral and heavily funded part of both our companies making extensive use of simulation, especially in safety critical developments such as an Anti-Lock Braking systems for road- and rail vehicles.
Professional simulators, provided with accurate data and run under all known conditions are exact, especially if your control loop includes actual measurements during prototyping it in fact becomes the system specification. It become accurate to a few fractional parts per million.
Your lag and lead compensation can be predicted very accurately after a few runs but you have to specify exactly or find from simulation which capacitors should be used to achieve the goal or would perform best in the required application.
The problem exists in mainly two areas, simple simulators assume perfect capacitors, and resistors and other basic components as well as perfectly matched semi-conductors with a few fixed parameters, this is not anywhere near reality. I see models of op amps based on a three terminal block and a macro describing it characteristic performance, which is absurd. Try simulating an op amp by building the whole circuit in your simulator, you would not know what to believe.
A simulator should be integrated with the PCB design program as well as the EMI program and this requires operators with an very good understanding of what they are simulating and what is displayed.
Very few companies can afford their own environmental test laboratories and third party testing is expensive, thus one has to be able to rely on a cost effective simulation during development in order to reduce the risk of performing any re-testing or the expense of lawsuits in the event of a critical failure.
Homologation testing both functionally and environmentally for an an ABS module at MIRA in the UK costs hundreds of thousands, if there were any failure, it requires rework and retest. How many amplifiers out there has actually been formally tested by anyone - legally it should. I know of only one that claims to have been tested at least for EMI by an accredited UK test facility, one of the Bryston models of three years or so ago, I cannot remember exactly, but this impressed me. There are at least one audio company out there that does more than listen to their product.
An expensive exercises and simulators provides a good alternative if you cannot afford to have such facilities available on your own premises.
I would say for simulating audio and the like were the result is really subjective, programs such as FREE SPICE could be a useful tools as you say, primarily because of its limited and very basic analysis. Unfortunately most seem to think it is the ultimate design tool and whatever it says is Gospel.
Anyway enough about simulators, I shall go back to sleep.
Professional simulators, provided with accurate data and run under all known conditions are exact, especially if your control loop includes actual measurements during prototyping it in fact becomes the system specification. It become accurate to a few fractional parts per million.
Your lag and lead compensation can be predicted very accurately after a few runs but you have to specify exactly or find from simulation which capacitors should be used to achieve the goal or would perform best in the required application.
The problem exists in mainly two areas, simple simulators assume perfect capacitors, and resistors and other basic components as well as perfectly matched semi-conductors with a few fixed parameters, this is not anywhere near reality. I see models of op amps based on a three terminal block and a macro describing it characteristic performance, which is absurd. Try simulating an op amp by building the whole circuit in your simulator, you would not know what to believe.
A simulator should be integrated with the PCB design program as well as the EMI program and this requires operators with an very good understanding of what they are simulating and what is displayed.
Very few companies can afford their own environmental test laboratories and third party testing is expensive, thus one has to be able to rely on a cost effective simulation during development in order to reduce the risk of performing any re-testing or the expense of lawsuits in the event of a critical failure.
Homologation testing both functionally and environmentally for an an ABS module at MIRA in the UK costs hundreds of thousands, if there were any failure, it requires rework and retest. How many amplifiers out there has actually been formally tested by anyone - legally it should. I know of only one that claims to have been tested at least for EMI by an accredited UK test facility, one of the Bryston models of three years or so ago, I cannot remember exactly, but this impressed me. There are at least one audio company out there that does more than listen to their product.
An expensive exercises and simulators provides a good alternative if you cannot afford to have such facilities available on your own premises.
I would say for simulating audio and the like were the result is really subjective, programs such as FREE SPICE could be a useful tools as you say, primarily because of its limited and very basic analysis. Unfortunately most seem to think it is the ultimate design tool and whatever it says is Gospel.
Anyway enough about simulators, I shall go back to sleep.
That was anything but a grand soliloquy, more a near-sighted, ill informed rant.
Anyone who has been using simulation for not much more than a year shouldn't hold himself up as an expert in it's use, even if he did buy the program.
As far as I know, all of the simulators (commercial or free) work on the same basic premise, that's why the models are pretty much interchangeable - I can use the same models in Multisim (NOT a "cheap" program) as one would use in LT or Microcap or Orcad.
I have seen one of the models Kean modified (I actually looked in the thread he linked to) and it looks like he is making some real progress with this, I would encourage him to continue. Andy_c has done some outstanding work, providing improved models for a few well known and often used devices - as far as I know he doesn't have the equipment that the semiconductor manufacturers have.
I have used the free utilities available, like RMAA to test my amps using the soundcard in my computer. It is not as convenient as a purpose-built unit, like an HP distortion meter but it can be very accurate. I have tested the same amp using both methods (brought it to Chris (Anatech) and he tested it's distortion on an HP meter) and the results concur.
Not all of us can afford to go out and buy a rack of equipment and the latest software. That is why some of these free utilities exist and you should try them for myself first before condemning them as useless, you might be surprised.
Anyone who has been using simulation for not much more than a year shouldn't hold himself up as an expert in it's use, even if he did buy the program.
As far as I know, all of the simulators (commercial or free) work on the same basic premise, that's why the models are pretty much interchangeable - I can use the same models in Multisim (NOT a "cheap" program) as one would use in LT or Microcap or Orcad.
I have seen one of the models Kean modified (I actually looked in the thread he linked to) and it looks like he is making some real progress with this, I would encourage him to continue. Andy_c has done some outstanding work, providing improved models for a few well known and often used devices - as far as I know he doesn't have the equipment that the semiconductor manufacturers have.
I have used the free utilities available, like RMAA to test my amps using the soundcard in my computer. It is not as convenient as a purpose-built unit, like an HP distortion meter but it can be very accurate. I have tested the same amp using both methods (brought it to Chris (Anatech) and he tested it's distortion on an HP meter) and the results concur.
Not all of us can afford to go out and buy a rack of equipment and the latest software. That is why some of these free utilities exist and you should try them for myself first before condemning them as useless, you might be surprised.
Hi John, Yes I would agree with you totally. These tools are helpful especially for the enthusiast. But you would not expect to find them were it matters. We were the first company in South Africa to purchase MicroCAP, I recall in 1993 and we have upgraded since then. We also purchase models of components that we use.
Spice is a universal simulation application but some uses more and other less model parameters. You can make a spice program that uses only one parameter. The accuracy of the models is quite important in order to obtain valuable information.
I have played a little with other spice simulators since I do not always want to bring a dongle home as I forget it in my machine and the next day have to pay an engineers salary for doing nothing. So yes these free programs do have their place and for that reason Hugh is absolutely right they offer great guidelines, like the calculator on your cellular phone, great for emergencies but not really an accountant's tool.
John I am not an audio know-all. I have been a qualified R&D engineer for about 40 years, keeping abreast with changes and advancements in electronics and also in software.
But in the audio forum there is much to speculate about since there is almost nothing anyone agrees with besides, everybody is an expert in this nebulous field of controversy and subjectivity. There is little science and very few proven rules.
In engineering the first thing that you learn is the theory of an amplifier, every single person with an electronic degree should be able to design a good performing amplifier in an afternoon. This was in the days of slide rules. Nowadays I guess it would take a good engineer about 10 -20 minutes to design a good amplifier from nothing. With the advent of simulators another 10 or so minutes to optimize it.
I used spice simulators and in particularly MicroCAP since 1993. It has been a great tool for almost any electronic simulation, whether it contains logic, mixed signal, etc. If the models are good and you know how to modify parameters then you will succeed with your design and can go from simulation to PCB.
Spice is a universal simulation application but some uses more and other less model parameters. You can make a spice program that uses only one parameter. The accuracy of the models is quite important in order to obtain valuable information.
I have played a little with other spice simulators since I do not always want to bring a dongle home as I forget it in my machine and the next day have to pay an engineers salary for doing nothing. So yes these free programs do have their place and for that reason Hugh is absolutely right they offer great guidelines, like the calculator on your cellular phone, great for emergencies but not really an accountant's tool.
John I am not an audio know-all. I have been a qualified R&D engineer for about 40 years, keeping abreast with changes and advancements in electronics and also in software.
But in the audio forum there is much to speculate about since there is almost nothing anyone agrees with besides, everybody is an expert in this nebulous field of controversy and subjectivity. There is little science and very few proven rules.
In engineering the first thing that you learn is the theory of an amplifier, every single person with an electronic degree should be able to design a good performing amplifier in an afternoon. This was in the days of slide rules. Nowadays I guess it would take a good engineer about 10 -20 minutes to design a good amplifier from nothing. With the advent of simulators another 10 or so minutes to optimize it.
I used spice simulators and in particularly MicroCAP since 1993. It has been a great tool for almost any electronic simulation, whether it contains logic, mixed signal, etc. If the models are good and you know how to modify parameters then you will succeed with your design and can go from simulation to PCB.
Hi Nico,
I got a different impression from you here.
On this forum there are those that pretend they are experts and there are those that are genuine experts.
Of course, I am neither. I just give my opinion (see post #994 in this thread) and go about my business of learning and building.
🙂
I got a different impression from you here.
On this forum there are those that pretend they are experts and there are those that are genuine experts.
Of course, I am neither. I just give my opinion (see post #994 in this thread) and go about my business of learning and building.
🙂
Last edited:
John,
Please stop taking cheap shots at Nico and other forum (= thread) members.
The experts who are pretending might get upset (including me!), and besides, are you really that clever?
Cheers,
Hugh
Please stop taking cheap shots at Nico and other forum (= thread) members.
The experts who are pretending might get upset (including me!), and besides, are you really that clever?
Cheers,
Hugh
The way I see it, Nico is entitled to his self-confidence since he's been in the field so long. If he says something that's wrong, we'll figure it out between all of us, no problem. His input is valued; knowledge is knowledge, regardless of the giver.
I don't see the logic here, it just seems like a witch hunt to me. Nico does not need to be "exposed" by anyone, we can all think for ourselves you know...
For the record I don't think Nico has done anything wrong.
- keantoken
I don't see the logic here, it just seems like a witch hunt to me. Nico does not need to be "exposed" by anyone, we can all think for ourselves you know...
For the record I don't think Nico has done anything wrong.
- keantoken
Thanks Anthony,
I have PMed John, asked him to moderate his comments. He indicates he won't be staying. I make it very clear I will not delete posts or ask individuals to leave. BUT, those who remain must be civil otherwise there will be conflict with me. I would say that while I'm not the best technocrat in the world I'm pretty good with words, and that is what these issues come down to.
Nico, I thought your response very measured, restrained even, and I take my hat off to you.
Cheers,
Hugh
I have PMed John, asked him to moderate his comments. He indicates he won't be staying. I make it very clear I will not delete posts or ask individuals to leave. BUT, those who remain must be civil otherwise there will be conflict with me. I would say that while I'm not the best technocrat in the world I'm pretty good with words, and that is what these issues come down to.
Nico, I thought your response very measured, restrained even, and I take my hat off to you.
Cheers,
Hugh
Aye...
So, I suppose there is a little more experimentation to do with prototypes and then a final schematic will be drawn? Where are we really in the development?
- keantoken
So, I suppose there is a little more experimentation to do with prototypes and then a final schematic will be drawn? Where are we really in the development?
- keantoken
Kean,
Crossfeed, possible revamp of tone control, level/balance, amp all ready, then we should be OK for a large schemat. Then it's the pcb layout, after a few important component choices.
Hugh
Crossfeed, possible revamp of tone control, level/balance, amp all ready, then we should be OK for a large schemat. Then it's the pcb layout, after a few important component choices.
Hugh
I've suggested a 2SC3423 VAS, but no one has commented on that yet. I wonder how much it matters, really, the circuit already performs so well. Any second opinions?
- keantoken
- keantoken
Kean,
Can't really see the point of the 3423 - it's hard to get and rated to 120V, which is wasted here. The device chosen has very low Cob, is running at just 4.25mA, and easily obtainable. Can you think of a good reason?
Hugh
Can't really see the point of the 3423 - it's hard to get and rated to 120V, which is wasted here. The device chosen has very low Cob, is running at just 4.25mA, and easily obtainable. Can you think of a good reason?
Hugh
Makes sense. The perfectionist in me though has suggested to use the 2N4124 here however; compare the Cob curves.
http://www.jaycar.com.au/images_uploaded/BC558.PDF
http://www.fairchildsemi.com/ds/MM/MMBT4124.pdf
Pulling teeth?
- keantoken
http://www.jaycar.com.au/images_uploaded/BC558.PDF
http://www.fairchildsemi.com/ds/MM/MMBT4124.pdf
Pulling teeth?
- keantoken
Looks good, I agree, nice fast device at 300MHz and quiet......
BUT, more difficult to find, Cob 4pF versus 6pF for the BC557, beta no higher.
However, I did note that the BC560 had 2db of noise compared to the 10dB of the BC557, so maybe we should use that? Easy to find, very good noise performance, all other parameters comparable or better than the 557!
Thanks, Kean, good suggestion....
Hugh
BUT, more difficult to find, Cob 4pF versus 6pF for the BC557, beta no higher.
However, I did note that the BC560 had 2db of noise compared to the 10dB of the BC557, so maybe we should use that? Easy to find, very good noise performance, all other parameters comparable or better than the 557!
Thanks, Kean, good suggestion....
Hugh
John, unlike yourself, I try not to make people look stupid. I go with someone who is new at something I will be new to it as well, I find it makes it a lot easier coaching, without telling, this is fortunately is my way of getting something across.
Also unlike yourself I did not progress from being a handyman to obtaining an electronic engineering degree through DIY audio.
In the past I had nothing but good things to say about John, but unlike him I do not need this forum to have a gripe at life.
I still have a very comfortable life and a few successful electronic engineering companies. I have enjoyed my hobby for the best part of my life and will continue to do so without DIY audio. I need not share anything, but I did because I wanted to, not because I had to.
Most of my life I attempt to be as humble as possible I never wanted to be perceived as superior. If I did so then I apologies to all.
Cheers all, and nice knowing you John.
Also unlike yourself I did not progress from being a handyman to obtaining an electronic engineering degree through DIY audio.
In the past I had nothing but good things to say about John, but unlike him I do not need this forum to have a gripe at life.
I still have a very comfortable life and a few successful electronic engineering companies. I have enjoyed my hobby for the best part of my life and will continue to do so without DIY audio. I need not share anything, but I did because I wanted to, not because I had to.
Most of my life I attempt to be as humble as possible I never wanted to be perceived as superior. If I did so then I apologies to all.
Cheers all, and nice knowing you John.
Last edited by a moderator:
We could use the MPSA18. 3pF Ccb, with 1.5db noise. And high Hfe.
http://www.fairchildsemi.com/ds/MP/MPSA18.pdf
Nico, please ignore John. We know enough not to take his personal comments seriously. If he attacks you and doesn't get a response, it makes him look bad, not you.
- keantoken
http://www.fairchildsemi.com/ds/MP/MPSA18.pdf
Nico, please ignore John. We know enough not to take his personal comments seriously. If he attacks you and doesn't get a response, it makes him look bad, not you.
- keantoken
Last edited:
Nico,
Please, stay with us. I speak for all here when I say that we hold you in the highest regard, not just for your knowledge, which is encyclopaedic, but for your willingness to explain carefully to others some of the finer points of electronics, a difficult endeavour - in short, to give willingly of your time and help others. John made mistakes, sure, but he tells me he would like to leave, and if that is the case, this unpleasantness is now behind us.
Folks, I cannot believe the time Nico has spent with me, explaining, encouraging, suggesting, and he has even rung me on several occasions to chat about electronics. I consider him a noble and kind friend, and I'm grateful I met him here. I believe we all are. To lose him would be a tragedy for the thread and a severe loss to DIYaudio.
To give John his credit, he is knowledgeable, generous with his circuits, and very energetic. He is welcome to stay with us but only if he immediately stops attacking the experience and beliefs of others. I resile from nothing I have said to you, John, either here or privately, you have been disruptive, and you have caused a lot of angst for others.
So, back to normal programming, MPSA18 looks good, Nico, what are your thoughts on this device? Perhaps a simulation would reveal any improvements? We have a short signal path, so a bit more feedback shouldn't hurt a bit......
Cheers,
Hugh
Please, stay with us. I speak for all here when I say that we hold you in the highest regard, not just for your knowledge, which is encyclopaedic, but for your willingness to explain carefully to others some of the finer points of electronics, a difficult endeavour - in short, to give willingly of your time and help others. John made mistakes, sure, but he tells me he would like to leave, and if that is the case, this unpleasantness is now behind us.
Folks, I cannot believe the time Nico has spent with me, explaining, encouraging, suggesting, and he has even rung me on several occasions to chat about electronics. I consider him a noble and kind friend, and I'm grateful I met him here. I believe we all are. To lose him would be a tragedy for the thread and a severe loss to DIYaudio.
To give John his credit, he is knowledgeable, generous with his circuits, and very energetic. He is welcome to stay with us but only if he immediately stops attacking the experience and beliefs of others. I resile from nothing I have said to you, John, either here or privately, you have been disruptive, and you have caused a lot of angst for others.
So, back to normal programming, MPSA18 looks good, Nico, what are your thoughts on this device? Perhaps a simulation would reveal any improvements? We have a short signal path, so a bit more feedback shouldn't hurt a bit......
Cheers,
Hugh
Member
Joined 2009
Paid Member
Moderator requested
Hugh - I will email you offline.
Moderators - I respectfully suggest that some of the posts above contain off topic comments that can be safely removed without hurting the flow of this thread with the benefit of leaving a better record for readers who follow us.
Hugh - I will email you offline.
Moderators - I respectfully suggest that some of the posts above contain off topic comments that can be safely removed without hurting the flow of this thread with the benefit of leaving a better record for readers who follow us.