I had some basic questions about the design and development process of any audio design. I plan to use ltspice to simulate, and then maybe breadboard it or use eagle and do a toner transfer pcb. So far I have just made other people designs and never did much of designing my self. I have worked on non audio designs though, led drivers, microcontrollers, non audio multichannel dacs, some digital circuits etc.
How do you test a design in ltspice. I have only used pulse to generate square waves so far for non audio work. I know it can do sine waves and 'raw waves' also. how do you capture the noise numbers etc.
Same thing with a real circuit also, I have an old tek scope, without the usb; and a square wave signal generator. What waveforms do you test with, what do you check on the scope to know the noise numbers are good enough etc.
How do you test a design in ltspice. I have only used pulse to generate square waves so far for non audio work. I know it can do sine waves and 'raw waves' also. how do you capture the noise numbers etc.
Same thing with a real circuit also, I have an old tek scope, without the usb; and a square wave signal generator. What waveforms do you test with, what do you check on the scope to know the noise numbers are good enough etc.
Last edited:
I think you're on the right path basically, but you just have to figure out what you're trying to do in order to know what you need to do. 🙂
Yeah, that sounds silly, but I'm replying because I'm doing the "theory, LTSpice, Eagle, toner-transfer prototype, put it on the test gear" loop right now, and it is working out for me.
In my case, I'm breadboarding specific circuits to make sure that they behave the way that I think they should, and also to see if there are any unforeseen problems that could be addressed before I send a pile of CAD and components off to a PCB/assembly house to generate a pile of pricey landfill. In your case, maybe you just want to make some nice gear that does something useful for you, like a power amp or a headphone amp.
In that case, use the tools to whip up a prototype and maybe listen to it? I have a nice Audio Precision analyzer, so I can see in great detail what the thing is doing, but still, a series of listening tests might be able to discover much of the same information as you tweak the design. Of course, I too will listen to some of these circuits, but only after I know that the circuits work well enough to pass the simple measurements and tests.
What are those measurements? I like to measure distortion, but not as a percentage, but as a spectrum of the distortion analyzer residual. In this way, I can see what makes up that (for example) "0.05%" THD+N number… is it hum? broadband noise? and which harmonics are present and in what quantity? You need a distortion analyzer to do that, but you can probably take the analyzer residual of a simple analyzer and feed it into a PC/soundcard to take the FFT of it to see what it's made of. Or, get an AP2 test set.. ;-)
Your question about LTSpice: I usually use a .tran analysis to test a random circuit, since it can handle startup behavior as well as all sorts of funky problems that might not be visible with a simple 'frequency response' type test. Still, a simulation is only as good as the component models you feed it, and sometimes, it's tough to accurately model the entire design, including the PCB.
I end up using LTSpice to gain insight on how a circuit works, and to be able to mess with the component values and choices quickly and to see how they affect the circuit. I then try to whip up a simple prototype using real components and perhaps a "non-real" layout (i.e. a single sided toner-transfer PCB layout) just to see if reality follows the model.
I would not expect to judge performance on a 'scope, other than to see if the circuit is oscillating or otherwise misbehaving. Oscilloscopes are not meant to give you high precision measurements (unless you pay $$ for a modern digital scope) but they're very useful to tell you some basic information very quickly: is each part of the circuit behaving like I think it should? Is each stage running at the right DC operating point? Is the circuit oscillating because of a bad layout or poor component choices, or a design defect? Stuff like that…
Anyway, I'm rambling, but you're not far off, and all you need to do is dive in a little further and see what works for you.
Best regards!
Yeah, that sounds silly, but I'm replying because I'm doing the "theory, LTSpice, Eagle, toner-transfer prototype, put it on the test gear" loop right now, and it is working out for me.
In my case, I'm breadboarding specific circuits to make sure that they behave the way that I think they should, and also to see if there are any unforeseen problems that could be addressed before I send a pile of CAD and components off to a PCB/assembly house to generate a pile of pricey landfill. In your case, maybe you just want to make some nice gear that does something useful for you, like a power amp or a headphone amp.
In that case, use the tools to whip up a prototype and maybe listen to it? I have a nice Audio Precision analyzer, so I can see in great detail what the thing is doing, but still, a series of listening tests might be able to discover much of the same information as you tweak the design. Of course, I too will listen to some of these circuits, but only after I know that the circuits work well enough to pass the simple measurements and tests.
What are those measurements? I like to measure distortion, but not as a percentage, but as a spectrum of the distortion analyzer residual. In this way, I can see what makes up that (for example) "0.05%" THD+N number… is it hum? broadband noise? and which harmonics are present and in what quantity? You need a distortion analyzer to do that, but you can probably take the analyzer residual of a simple analyzer and feed it into a PC/soundcard to take the FFT of it to see what it's made of. Or, get an AP2 test set.. ;-)
Your question about LTSpice: I usually use a .tran analysis to test a random circuit, since it can handle startup behavior as well as all sorts of funky problems that might not be visible with a simple 'frequency response' type test. Still, a simulation is only as good as the component models you feed it, and sometimes, it's tough to accurately model the entire design, including the PCB.
I end up using LTSpice to gain insight on how a circuit works, and to be able to mess with the component values and choices quickly and to see how they affect the circuit. I then try to whip up a simple prototype using real components and perhaps a "non-real" layout (i.e. a single sided toner-transfer PCB layout) just to see if reality follows the model.
I would not expect to judge performance on a 'scope, other than to see if the circuit is oscillating or otherwise misbehaving. Oscilloscopes are not meant to give you high precision measurements (unless you pay $$ for a modern digital scope) but they're very useful to tell you some basic information very quickly: is each part of the circuit behaving like I think it should? Is each stage running at the right DC operating point? Is the circuit oscillating because of a bad layout or poor component choices, or a design defect? Stuff like that…
Anyway, I'm rambling, but you're not far off, and all you need to do is dive in a little further and see what works for you.
Best regards!
halo mister
i see you using eagle
is the variable .step (like in ltspice) is limited to only 3 variable ?
This will probably be helpful:
Distortion measurements with LTSPICE - Audio PerfectionAudio Perfection
LTspice doesn't simulate noise very well, it's only really useful to get a ballpark THD of discrete circuits.
Distortion measurements with LTSPICE - Audio PerfectionAudio Perfection
LTspice doesn't simulate noise very well, it's only really useful to get a ballpark THD of discrete circuits.
LTSPICE simulates noise quite adequately. It is the quality of your model, that usually limits noise analysis precision
How do I generate a sine wave. Is there a circuit that can convert a square wave to a sinewave? Or should I burn some wave files to a cd and use a cd player. I have a waste discman lying around somewhere.
How about taking measurements on a pc. Do I need a good soundcard for taking measurements. The only external sound card I got is odac and I guess it doesnt have a mic input. Once I have the read wave files from mic input, whats a good freeware software to do the analysis.
To test an amp, I guess I would need some kind of circuit between the amp and the comp, any idea what it should be? I guess at the least I would need to reduce the voltage to below what the sound card can handle at its input, and that should be adding its own distortion, so probably a resistor based voltage divider is not really a good idea.
How about taking measurements on a pc. Do I need a good soundcard for taking measurements. The only external sound card I got is odac and I guess it doesnt have a mic input. Once I have the read wave files from mic input, whats a good freeware software to do the analysis.
To test an amp, I guess I would need some kind of circuit between the amp and the comp, any idea what it should be? I guess at the least I would need to reduce the voltage to below what the sound card can handle at its input, and that should be adding its own distortion, so probably a resistor based voltage divider is not really a good idea.
- Status
- Not open for further replies.