Hi Pete,
I'm not sure if you were aware of this, but one of peufeu's interests is memory distortion. I haven't really investigated memory distortion, so I don't have an opinion on it one way or another. But, given that interest, it seems to me that having a simulator that supports models like VBIC and Mextram would be just the thing to study it in an objective way. So in a transient analysis, the instantaneous junction temperature could vary from one sample to the next, giving variations in things like Is and beta. I'm just guessing, but maybe this would show in increase in distortion at low frequencies that's not predicted by the normal theory?
I'm not sure if you were aware of this, but one of peufeu's interests is memory distortion. I haven't really investigated memory distortion, so I don't have an opinion on it one way or another. But, given that interest, it seems to me that having a simulator that supports models like VBIC and Mextram would be just the thing to study it in an objective way. So in a transient analysis, the instantaneous junction temperature could vary from one sample to the next, giving variations in things like Is and beta. I'm just guessing, but maybe this would show in increase in distortion at low frequencies that's not predicted by the normal theory?
Interesting!
I'm looking at something similar, a protection circuit for a power MOSFET output stage that uses an analog multiplier and filter to estimate the instantaneous junction temperature of the output devices. Based on the datasheet SOA curves, the time constants for the power devices are in the range of 5-15 msec.
In the Fairchild app note AN-7532 for power MOSFETs, they have a thermal transfer function that's the series connection of a bunch of parallel RC networks (like a Foster's expansion of an RC impedance). Assuming positive component values, this implies poles and zeros that alternate on the negative real axis. So I'll have to do an empirical fit to pulse responses of several pulse widths to determine the poles and zeros of the thermal transfer function.
Different application but same sort of idea.
I'm looking at something similar, a protection circuit for a power MOSFET output stage that uses an analog multiplier and filter to estimate the instantaneous junction temperature of the output devices. Based on the datasheet SOA curves, the time constants for the power devices are in the range of 5-15 msec.
In the Fairchild app note AN-7532 for power MOSFETs, they have a thermal transfer function that's the series connection of a bunch of parallel RC networks (like a Foster's expansion of an RC impedance). Assuming positive component values, this implies poles and zeros that alternate on the negative real axis. So I'll have to do an empirical fit to pulse responses of several pulse widths to determine the poles and zeros of the thermal transfer function.
Different application but same sort of idea.
Well, I'm an old dog, and microcontrollers are a new trick that I want to delay learning for a while - maybe when I do a preamp 
.
My power amp design has some Halcro-like bootstrapping of the output stage, so there's voltages that float 15V above and below the output. Some op-amps and an AD633 could float on these, using the power amp's output as a virtual ground. Then the op-amps could easily interface to a standard limiting circuit.
.My power amp design has some Halcro-like bootstrapping of the output stage, so there's voltages that float 15V above and below the output. Some op-amps and an AD633 could float on these, using the power amp's output as a virtual ground. Then the op-amps could easily interface to a standard limiting circuit.
PB2 said:
This is one of the problems with splitting threads--things get shunted off into other places. I don't know whether my relevant posts ended up in this thread or the one that was spawned off of this and I don't have time to look.
However...
You get poor marks for reading comprehension.
Go back and reread my posts (you may have to locate the other thread--not my fault). I never said that simulation was useless for other fields. I said it was a waste of time for audio. It doesn't bother me in the least if someone uses simulations to design a motor control circuit. I care not a whit if someone uses simulations to design a computer. I don't loose sleep at night if someone uses simulations to design a car ignition system. Trying to impress me with what 'most professionals' do or what 'major segments of the engineering industry' do doesn't work. This is audio and audio has its own difficulties; difficulties not shared by other segments of the electronics industry. As a result, I care very much when people assume blindly that simulations have much to do with real world audio. Audio has more subtleties than brute force motor control or a zero/one computer.
Your final paragraph almost redeemed you, but then you failed the test. Congratulations, you were the first person to note the fly in the ointment (good), but you ignored the implications (bad). Yes, in terms of reliability a design should be able to take any reasonable ambient temperature in stride. This we can agree upon. But I had a more subtle point in mind. Temperature changes device behavior->device behavior (in this case think transconductance) changes gain->gain changes feedback->feedback changes performance. You see in simulations exactly what you expect to see...and only that.
To put it another way, most everyone will agree that a piece of equipment has to be turned on for thirty minutes to an hour before testing, yes? Why? Primarily because of temperature. Now, imagine for a moment that a listener is operating that piece of equipment in a room at 68 degrees F. He hears one thing. Imagine another listener with the same piece of equipment at 105 degrees F, installed above a piece of tube equipment. He reports something different. Each calls the other deaf because they can't agree on what the unit sounds like. So who's right?
Is it so difficult to imagine that they're both right? That the same design could sound different under different conditions?
Will a simulation show you that?
You might be able to infer that an increase in current here would lead to a change there, but it never occurred to you to consider it. Ah, but a good pair of ears could raise the question easily. That's real world audio. That's one of many reasons to distrust simulations in audio.
Grey
GRollins said:
You certainly make numerous assumptions, and you would be wrong about most of them. You don't know anything about what I have designed (you've not asked) or of the thousands of other designs done in this industry. Further, you comment about my comprehension? You tested me? How arrogant. This is where I stop reading you, not worth my time.
Pete B.
john curl said:
John, unlike you, I don't design things as elementary and basic as audio power amplifiers for a living - that's just something that I do on the side and for a hobby.
Secondly, anyone fitting your "newbie engineer" profile, employed in their capacity as a electrical or electronics engineer, wouldn't last a week in the real world. It's amazing that the world of electronic and electrical engineering has progressed so much since your school days, isn't it?.
If characterising and contrasting yourself with “newbie engineers” in such a manner makes you feel better about yourself then that’s fine, but I’d just point out that your characterisations don’t have much relation to reality.
Your 'test it by building it philosophy' is definately admirable, but don’t forget that many designers work on products of such complexity that optimising them by building physical prototypes may literally take many years, while doing it in computer simulation could take a week.
And yes, this does apply to audio engineering –particularly the areas in which computers, digital mastering, DSP, microcontrollers, computers, etc play an increasingly dominant role.
If good hi fi amplifiers are so easy to make, why does anyone discuss anything on this website, in the solid state area? I think you are just inserting a hidden insult to successful audio designers everywhere.
I know how to make microphone electronics, master analog tape recorders, studio boards, ultra low noise pre-preamps, preamps, power amps, and electronic xovers. In fact, I am famous for most of my contributions in designing these devices. What have you done lately?
I know how to make microphone electronics, master analog tape recorders, studio boards, ultra low noise pre-preamps, preamps, power amps, and electronic xovers. In fact, I am famous for most of my contributions in designing these devices. What have you done lately?
john curl said:
John, don't over do it. Audio engineering is only one engineering domain, with many vastly varied facets. Being a “guru” in a few doesn’t make your generalisations on SPICE, audio engineering and “newbie engineers” as a whole right.
And yes, to make a good Hifi amplifier does require skill and knowledge, but so do other things, and some more so.
andy_c said:Well, I'm an old dog, and microcontrollers are a new trick that I want to delay learning for a while - maybe when I do a preamp
My power amp design has some Halcro-like bootstrapping of the output stage, so there's voltages that float 15V above and below the output. Some op-amps and an AD633 could float on these, using the power amp's output as a virtual ground. Then the op-amps could easily interface to a standard limiting circuit.
Andy, have you had a look at the very simple PIC type controllers?
http://www.bobblick.com/techref/projects/picprog/picprog.html
I've worked with processors ranging from 8080s to large microcoded
custom VLIW hardware, but not PICs, yet. I expect to take a look at
them if I ever do a remote control pre-amp, and that they'll
be easy to use.
Pete B.
PB2 said:
I've used a PIC16F876 for the exact same purpose as described here. It has an internal 10 bit ADC and multiple multiplexed inputs.
The great thing about this method is that you can program SOA secondary breakdown protection into the uP. Implementing such with an analogue multiplier would be a bit of a pain. I currently do all my PIC programming with the windows based C compiler by CCS (but started with the DOS version):
http://www.ccsinfo.com/
The complier comes with a very good manual and heaps of example files to study and run. With a software package such as this, getting into uP programming quickly is real easy, even if you only have rudimentary programming experience in BASIC - which was my prediciment 8 or so years ago.
If fact, I'm currently using the CCS complier to program a bank of PIC's for a gamma/neutron nuclear bore hole logging tool I'm designing, right now.
Cheers,
Glen
Yes, it looks like Microchip has about the best support community around. I was thinking about the ones that have a USB interface as described here. Apparently Microchip makes a C compiler for these that's free, and only lacks some of the optimizations of the pay version. I have lots of C and C++ programming experience, but not with any embedded solutions. I think that with some study of the architecture I could figure out what's going on.
My goal for a preamp would be to use the Microchip part and have a USB interface that I could use to update the firmware by hooking the computer straight up to it. They apparently have software on the PC and microcontroller side that's free and can help make this happen.
My goal for a preamp would be to use the Microchip part and have a USB interface that I could use to update the firmware by hooking the computer straight up to it. They apparently have software on the PC and microcontroller side that's free and can help make this happen.
andy_c said:
Andy,
Definitely take the plunge, you are in for a lot of fun. I started hand coding microprocessors almost 30 years back, used cross assemblers later and C compilers later still. For engineers with gusto for hardware it is the best of both worlds!!! You actually scope the program goings!!
Definitely a good developement environment is a must, with the ability to full debugging.
Rodolfo
