Hi,
I'm on the lower end of the learning curve with chip amp design but I'm at the stage where I want to develop some designs of my own. Given that layout is so critical is there any point in prototyping with breadboard or veroboard or would my time be better spent learning how to use SPICE? The other option I guess is to go straight to PCB and make the layout flexible enough to experiment with component values.
Any tips as to what you guys do?
Thanks
Simon.
I'm on the lower end of the learning curve with chip amp design but I'm at the stage where I want to develop some designs of my own. Given that layout is so critical is there any point in prototyping with breadboard or veroboard or would my time be better spent learning how to use SPICE? The other option I guess is to go straight to PCB and make the layout flexible enough to experiment with component values.
Any tips as to what you guys do?
Thanks
Simon.
Hi Simon,
I would start with the SPICE (I still have to learn it by myself
). Once you stabilized and compensated your design in SPICE then make a PCB for it. Be prepared to make several PCB design iterations before you get what you aim for. Breadboarding leaves you mostly through-hole components option while many "interesting" parts these days are SMD. It is also hard to optimize the layout on a breadboard which can be critical for achieving best performance especially if you decide to try a composite amp topology.
Regards,
Oleg
I would start with the SPICE (I still have to learn it by myself
). Once you stabilized and compensated your design in SPICE then make a PCB for it. Be prepared to make several PCB design iterations before you get what you aim for. Breadboarding leaves you mostly through-hole components option while many "interesting" parts these days are SMD. It is also hard to optimize the layout on a breadboard which can be critical for achieving best performance especially if you decide to try a composite amp topology.Regards,
Oleg
My first effort at a chip amp wasn't good.
It oscillated.
Turned out the layout was poor.
Decoupling needs to be close to IC as possible
Feedback resistor needs as short a path has possible.
I also made the mistake of lowering gain and didn't realise there is a minimum gain required for stability.
I saved the situation by add 1000pf between inverting and non-inverting inputs.
The oscillation was getting in everywhere.
The TDA7294 oscillated at 1MHz and this got back into my valve pre amp on the same pcb. Then it just fed around the loop.
I tamed the valve with 47k grid stopper 100pf from grid to ground.
It oscillated.
Turned out the layout was poor.
Decoupling needs to be close to IC as possible
Feedback resistor needs as short a path has possible.
I also made the mistake of lowering gain and didn't realise there is a minimum gain required for stability.
I saved the situation by add 1000pf between inverting and non-inverting inputs.
The oscillation was getting in everywhere.
The TDA7294 oscillated at 1MHz and this got back into my valve pre amp on the same pcb. Then it just fed around the loop.
I tamed the valve with 47k grid stopper 100pf from grid to ground.
You can get some ideas for which simulations to run from my Taming the LM3886 series.
The first thing that'll throw you off is the terminology, so here's a quick primer:
Note that all simulations start with a DC operating point calculation. If your DC op point is wrong, the subsequent simulation (AC, transient, noise, etc.) will be invalid. Verify that you get the expected operating point by running a DC sim before proceeding to the other simulations.
In general: You simulation is only as good as the models. Crap models -> crap results. Thus, part of any simulation task is to familiarize yourself with the models to see how well they jive with reality.
Most models get the DC op point right. Many get the AC performance and noise performance right. Some (few?) get the transient response right to the point where you can get accurate THD simulation results, though you can usually tell if something is fundamentally broken.
Commonly used simulators include: LTspice, TINA-TI, and Pspice. I mostly use TINA-TI as I use National/Burr-Brown/TI parts. They each have their quirks and strengths. LTspice and TINA-TI are free from LT (now ADI) and TI, respectively. Pspice is from OrCAD. A student version is available for free. It's crippleware but it is somewhat useful as I recall.
I definitely encourage you to explore circuit simulation. Once you get over the initial hump of learning how to use the simulator you can learn circuit so much faster using the simulator. A resistor substitution that would take a few minutes in the lab (and cost $0.10) can be accomplished for free in a few seconds in the simulator.
Tom
The first thing that'll throw you off is the terminology, so here's a quick primer:
- DC Simulation: Calculates the DC operating point. Useful for the "is it on" types of diagnostics. Also useful to see if you have a wrong connection somewhere as the DC values will not be as expected.
- AC Simulation: Frequency sweep. Used to determine the frequency response of your circuit. Can also be used to determine stability, though there are may ways to do that wrong, so be careful.
- Transient simulation: Simulation vs. time. Basically, if you would use an oscilloscope to measure something in reality, you'll want to run a transient simulation. Also useful if you want to simulate distortion (run a transient sim with a sine wave input and perform an FFT on the output waveform). Some simulators (TINA-TI) have a harmonic distortion simulation that's separate from the transient sim.
- Noise simulation: Like AC sim but looks at the noise (either input-referred or output referred or both) of the circuit.
- Temperature sweep: This usually is an option under DC or AC sweep, but it can also be a separate simulation. It simulates the circuit at various temperatures. This is very useful for looking at bias stability of discrete output stages, for example.
Note that all simulations start with a DC operating point calculation. If your DC op point is wrong, the subsequent simulation (AC, transient, noise, etc.) will be invalid. Verify that you get the expected operating point by running a DC sim before proceeding to the other simulations.
In general: You simulation is only as good as the models. Crap models -> crap results. Thus, part of any simulation task is to familiarize yourself with the models to see how well they jive with reality.
Most models get the DC op point right. Many get the AC performance and noise performance right. Some (few?) get the transient response right to the point where you can get accurate THD simulation results, though you can usually tell if something is fundamentally broken.
Commonly used simulators include: LTspice, TINA-TI, and Pspice. I mostly use TINA-TI as I use National/Burr-Brown/TI parts. They each have their quirks and strengths. LTspice and TINA-TI are free from LT (now ADI) and TI, respectively. Pspice is from OrCAD. A student version is available for free. It's crippleware but it is somewhat useful as I recall.
I definitely encourage you to explore circuit simulation. Once you get over the initial hump of learning how to use the simulator you can learn circuit so much faster using the simulator. A resistor substitution that would take a few minutes in the lab (and cost $0.10) can be accomplished for free in a few seconds in the simulator.
Tom
Last edited:
Thanks for your responses, sounds like simulation is the way to go. I'll take a look at the different software out there, I'm on a mac so my choices may be limited, I think Eagle does some simulation so I may start there.
Maybe I'll switch from brushing up on my maths (which I seem to have forgotten in recent years) to learning SPICE.
Maybe I'll switch from brushing up on my maths (which I seem to have forgotten in recent years) to learning SPICE.
VirtualBox is another option. I didn't find it as good as VMware when I tried it, but it is free so the price is definitely right.
Oracle VM VirtualBox
Running Windoze in a VM is nice. It's very convenient to be able to share the clipboard between the two OSes.
Tom
Oracle VM VirtualBox
Running Windoze in a VM is nice. It's very convenient to be able to share the clipboard between the two OSes.
Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.