Spice simulation

John, I think I can more or less agree with all that you said. It is important to know that one cannot rely on a simulation and that it has limitations. But that is true of most tools. Measurements are not always correct or even remotely correct. Just as one has to realize the limitations of measurements and the potential error sources in measurements, one has to realize the limitations and error sources in a simulation. Any tool can give the wrong result if handled wrong. (That is especially important to know if using tools like axes or knives.) A simulation program is basically not very different from other tools. It is just that it is so complex that it is much harder to see if one is using it correctly or not, and if it gives the intended result. Computers do make it easier to mess things up. Spice is a tool that can be very useful, if handled right. But it is not a necessary tool, as both you and many others prove (with the exception of IC design which I think would not be possible today withouth advanced simulation programs).

I think nobody has tried to claim that Spice results are always reliable. There are some people, though (not thinking of anyone in this thread) who seem to believe they are the only ones who are bright enough to realize that Spice results may differ from a real circuit and thus, everybody else must be stupid enough to think that Spice gives a perfect prediction of a real circuit.

Further, don't believe that everybody who likes Spice must be young. I have also used slide rules and mechanical calculators, as I am sure many others here have. That is fun to have done, but I see no place for such tools today, since pocket calculators and computers are much better replacements. There is however one tool that I can never dispense with, and which is the oldest of them all, pen and paper. Pen and paper is one of the most ingenious, useful and reliable tools mankind has ever invented.
 
I would like to thank those who responded to my last couple of posts for participating in an impromptu experiment in reading comprehension and human psychology. (One of my degrees is in psych--old habits die hard.)
I trust the parallels between my post on Science as Religion and some of the responses are obvious enough.
As for my mainframe/network analogy, it was deliberately oblique, though not inappropriate. Again, the responses were predictable. In passing I'll note that it's almost always the ones on the butt-end of an analogy who complain that it isn't relevant.
John, at least, appears to have gotten my point, but that's not surprising, in that he seems to hold some of the same viewpoints that I do. I'm not saying that he agrees with everything I say--he may, he may not; he's a free man and has his own experiences and opinions.
You can't simulate for imaging. You can't simulate for dynamics. You can't simulate for clarity. These things exist...
*****
Brief aside--I'm always amazed by people who claim that imaging doesn't exist. Close your eyes. Point to the musician. Did your finger end up somewhere between the speakers? That's lateral imaging. Close your eyes again. Throw a wad of paper at a well recorded sax player taking a solo. Did the wad of paper end up behind the speakers? That's depth of image. Dynamics is harder to "prove" but I trust that most people who actually listen have heard amps that you can feel in your gut, whereas others don't produce the same visceral response. Etc.
*****
...yet it's not entirely clear how to manipulate a circuit to produce them in a predictable way. It's trivial to say that more capacitance in the power supply might help with dynamics, but there are examples of amps with huge power supplies that don't impress. Likewise it seems probable that lower feedback is beneficial for imaging, but it's not an open and shut case; note that I said beneficial, not necessary (just to head off a few protests--another one of those reading comprehension things...).
Yes, it's true that some of the subjective stuff is poorly understood, and as such hard to design for. That's part of what I'm saying. But the flipside of the coin is that--like the Aleph-X "not working"--you can't depend on simulations to tell you whether the circuit will actually "work." In the case of the Aleph-X, the simulations indicated that the topology wouldn't even idle, much less make music. Those were gross failures, easy enough to spot. But how about more subtle things? I happen to like imaging. There's no simulation software on Earth that will tell me whether one topology will image better than another. Dynamics? Likewise. And so forth.
You can simulate THD until the cows come home (English idiom meaning "for a long time"), but those numbers have only the loosest of correlations with reality because you're dealing with cloned parts; perfect in every way. Monte Carlo analysis can shake up the numbers a bit, but it's still a badly flawed approximation of reality.
How long will it take for the models to get better? I don't know. But it isn't there yet and isn't likely to get close enough in the next couple of years to accomplish the things that we need to do. If you want to know real answers, you've still got to build the thing. And it's likely to remain that way for a while yet.

Grey
 
I'd like to say that it appears that Nelson and I hit the Submit key at the same time. There's no possible way that he could have read my post and agreed that quickly.
He is free to agree with what I said or not, but his post should not be taken as approval without further input from him.

Grey

EDIT: Ah, I see he has edited his post. Good.
 
Greg, aren't you missing the point? Assuming there are properties like imaging that do not follow automatically from the usual figures like low THD and low IM etc. they are properties we don't even know how to measure and probably nodody knows how to design for with predictable results. Of course you could not then expect a simulator to tell you in advance how the imaging of an amplifier will be. If it exists as clearly identifiable property, then obviosuly there is something lacking in our theory (or understanding thereof) and in out measurements. Spice is a simulation using very elaborate models of our current understanding of the electronics and semiconductor physics. No more, no less. If our theory is wrong or incomplete, how could Spice know better? Neither Spice nor any other simulator is designed to cope with X-factors, to use Bobs terminology, and neither is the theory in electronics textbooks. Are those textbooks also completely useless for amplifier design? I think not.

Spice also has very interesting uses apart from just simulating a particular design. It seems that it is far beyond some peoples comprehension to even imagine that, so I am happy to see that at least Rodolfo has realized that too.


EDIT: You must have read an earlier version of Nelson post than I did, but I think I can imagine what Nelson edited. :)
 
Right on, Nelson. My greatest leaps in design are often after I make a mistake.
One time, however, a business rival accosted me at the CES where we were both in 'business mode' including wearing business suits, and declared to me that I was a washout and a has-been at the old age of about 40. Two weeks later, I came up with the phono input circuit that launched the Vendetta Research preamp.
 
AX tech editor
Joined 2002
Paid Member
john curl said:
[snip]You, newbe engineers, are taught in school how to use Spice, and you can't see a world without it. [snip]

John,

I am not a newbee engineer. I am probably as old or older than you. I have been using Spice for at least 15 years, using its powers for proof-of-concept work, and researching issues that are hard or vey cumbersome to do in actual hardware, like measuring the sum of several currentsall at once. I am well aware of Spice's limitation.

It's not just the newbees that use Spice. Some of us absorb new tools of the trade as they come available and use them to advantage. Some others stopped moving 30 years ago. That's OK, but then they shouldn't tell us how we should do it.

Jan Didden
 
Christer said:


Greg, aren't you missing the point?


My name is spelled Grey, not Greg...like I said, it's a matter of reading comprehension.
Go back and reread my posts and you will find that I've already addressed your point.


Nelson Pass said:


Interestingly, some of the best insights into these issues come
from negatives. You start with a circuit that has the qualities
you want, and you make a small change that degrades them.

:cool:


Nelson's post addresses my experience exactly. Case in point: When I was in school, I thought negative feedback was the ultimate answer. It cured all ills, right? The professor said so and I believed him. All you needed was enough open loop gain and you could conquer the world. Of course, a lot of other people felt the same way, so we ended up with junk like Nikko amps (amongst many others) with THD specs featuring a zillion zeros after the decimal point. Only they sounded like crap, didn't they?
(John, this is your cue to cover alternate distortion measurements.)
That was, as Nelson put it, a negative. A big one. What should have been better, wasn't.
A more subtle question that never seems to get asked, much less answered, is how those circuits made it from the test bench to the shelf at the retailer. Could it be because they skipped the listening room? No wonder the tube manufacturers of the day gained such a large market share--their solid state competitors were their own worst enemies. Tubes really did sound better than most solid state, and by no small margin.
I often wonder where tube gear would be today had it not been for the low distortion craze back in the 70's and early 80's.
I do not claim to have the circuit-related expertise that John and Nelson have, but I've done a little here and there and have found that theory needs checking a lot more often than some would have you believe. If it makes it more palatable, think of it as negative feedback; use your ears to tell you when you've gone too far. It's for sure that simulations won't tell you...they're positive feedback; they will confirm all your preconceived notions. Of course it's better! Look how many extra zeros we've tacked on behind the decimal point!
Audio history says otherwise.

Grey
 
GRollins said:


My name is spelled Grey, not Greg...like I said, it's a matter of reading comprehension.

Of course it is Grey. I was tired when posting. I am terribly sorry. Please accept my apologies.



Go back and reread my posts and you will find that I've already addressed your point.

I really cannot see that, so one of us must be seriously misunderstanding the other. It still seems to me that you are trying to say that Spice is useless because it doesn't do what you want it to, totally disregarding that it is not and never was intended to do that. I could say that cars are useless since we cannot fly to the moon with them, but it would be unfair and pointless to do so.

Please note, however, that I am not trying to convince you, or anybody else, to start using Spice. As I said, Spice is not necessary, but it can be very useful for those who realize what it can be used for.
 

GK

Disabled Account
Joined 2006
GRollins said:
I would like to thank those who responded to my last couple of posts for participating in an impromptu experiment in reading comprehension and human psychology. (One of my degrees is in psych--old habits die hard.)
I trust the parallels between my post on Science as Religion and some of the responses are obvious enough.


The content of your own post was a greater parallel and you come off as more of a patient than a practitioner.

Someone who makes a dogmatic assertion that spice simulation has no application audio engineering clearly isn't in touch with reality and doesn't have much authority with which to denounce others as religious dogmatists.

Also, what remarkable powers of perception do you posess in order to berate certain people here as being at the "but-end" of your incoherent analogy - people whose knowledge, careers and achievments you know absolutely nothing about?
 
GRollins said:
.......................
You can simulate THD until the cows come home (English idiom meaning "for a long time"), but those numbers have only the loosest of correlations with reality because you're dealing with cloned parts; perfect in every way. Monte Carlo analysis can shake up the numbers a bit, but it's still a badly flawed approximation of reality.
................
Grey

Hi Grey,

I've simulated a lot of amplifiers, just to see how reliable the THD predictions are. According to my experiences, those number do have a close correlations with reality, say +/- 3dB, just what one would expect when taking into account the lot to lot spread of the various semiconductor parameters.
As for cloned parts, do you really think spice adepts are so naive using only clones?

Regards, Edmond.
 
Nelson Pass said:
washed up at 40

Dude,

that was '91, you were too busy washing cash in your "little laundrette" then. :clown:

The car shop owner in the village i reside in needed a week and a thousand bucks to figure out the thing refused to start because the $2 chip inside the car key had taken off to the twilight zone.
However, the guy still knows how to flatten my engine head by sanding it on a mirror with diamond paste.

Not talking Spice, without simulation tools i'd be a goner.
Just like a range of other applications, Spice has progressed a long way compared to a decade ago.
But i'd gladly trade software for 30 years of experience, a wet finger, and a set of ears.

A quote from someone we haven't seen here in quite some time, regretably :
jcarr said:
The key is to understand the topology and operating conditions of the circuit, and know what characteristics are required from individual components at each location in the circuit.
To summarize: "Do you understand what you are doing?"
 
john curl said:
Actually, we just take SPICE a little less seriously than some others. (dog barking)


If you don't use it at all, that would truly be an understatement.

SPICE is just another tool, but a very good one. Comparing it to ECAD of 40 years ago is just plain silly.

I didn't avoid using PC's because of the limitations of the Commodore 64.

As I said, you can design a fine amplifier without SPICE. You can often do the same without a spectrum analyzer in your toobox. I also do not SPICE everything in sight.

And yes, it is important to have the basic design understandings and insights BEFORE you SPICE the circuit, and YES, you are a fool to believe all that SPICE tells you without subsequently building the design and measuring it thoroughly.

You can also drive a car without a seatbelt, and yes, there is the infrequent occasion when the seatbelt doesn't work perfectly, or makes it more difficult to exit the car. That is not a reason not to wear one. Our Governor just found that out the hard way.

Cheers,
Bob