Sorry for the huge wall of text, but I would like to be as detailed and specific with my questions as I can be.
After playing with microcontrollers and PWM drivers and the like for a while, I want to take my first stab at building an analog circuit. I chose to design a battery-powered portable light organ using an electret microphone for the input signal, and I think I'm pretty close on the design but am running into some challenges.
I started with this design (http://www.jameco.com/Jameco/workshop/diy/basicledcolororgan-fig11.jpg) from a Jameco video but made some (hopefully!) improvements. My intent is to power it off some CR123A batteries, because they have good energy density, high enough discharge rate for what I need, and I already have some laying around. I'm currently planning on using this cheap mic (http://www.cui.com/product/resource/digikeypdf/cma-4544pf-w.pdf), 20-20000 Hz range, -44 dbV/Pa gain, 3-10V rated voltage.
Here's the circuit (http://i.imgur.com/9J3YEb2.png) I currently have drawn up, I've been doing some LTSpice simulation (though I'm a total novice at it). I hooked up the mic according to the schematic in the data sheet and changed the filters to low pass and high pass filters (aka a three-stage second-order crossover network) instead of the narrow bandpass filters in the original design, and they give the frequency response I want with the component values I calculated and put in the model. I'm not sure if just treating the mic as an AC voltage source is the right way to model it, but I'm not confident enough in my LTSpice skills to model the mechanics of the mic itself. I'm hoping to be able to run off +/- 3V (two batteries, low voltage to dissipate with current-limiting resistors on the LEDs for better battery life), though I'm not sure yet if that's feasible. I could add 2 more for +/-6V. More would probably hurt the portability of the device.
I'm currently using the UniversalOpAmp2 op amp model because the op amp I've been looking at using (more on this because the one I picked out is probably the wrong one) isn't in the default LTSpice library and I don't know how to make my own model yet, and for rectifying diode and transistor I just used some common "classic" components for now.
The major issue I'm encountering is the signal strength of the mic. For -44 dbV/Pa @ 1000mV/1Pa reference, I calculate a signal strength of 6.3 mV/Pa. A pretty usable range of 40 dB to 100 dB SPL gives sound pressure values of 2e-3 to 2 Pa, so signal strength of 12.6 uV (!) to 12.6 mV. I think that range guarantees I will need an adjustable gain stage to avoid clipping the signal at the rails - I verified in simulation that clipping the signal does add significant high frequency components, which I expected based on my knowledge of digital filters but wasn't sure about for analog filters.
Anyway, finally onto the questions. The first two should be easy ones. For the input gain stage, should I use an inverting configuration for low (matched?) input impedance, or a non-inverting configuration for high input impedance? I'm not sure which is better for input from a microphone. I'm fairly sure I want the high impedance configuration for the filter stages so they don't affect the rest of the circuit. Second, in the shown circuit, I set up the first stage gain to the highest without clipping for the maximum input signal, with some margin (3v / 0.0126V = 238, so 200 to give some margin, or 400 with +/-6V rails, as shown). Should I move some of that gain to the filter stages to avoid possible GBW limitations? I'm not super concerned with the fidelity of the signals as long as they switch the LEDs on and off, if that is relevant to the question.
On the gain topic, I would (obviously) need adjustable gain from 1-1000 to get the same signal strength between 40 dB and 100 dB SPL. Does it matter if the adjustable gain stage goes before or after the fixed gain stage? 1000x gain would require 20MHz GBW, which isn't too tough to find, but I'm not sure if I would run into any gain/phase margin issues at that level of gain and frequency - would it be better to instead spread the gain over two op amps (connected to the same pot)?
The biggest question for me is op amp selection. My needs are pretty demanding - rail-to-rail, high GBW, low input voltage requirements, either very low input offset or offset null pins (maybe?), and DIP package. My current simulation effectively uses an ideal op amp model for lack of a better component to use (or the knowledge to make my own component model). The one I had been looking at based on an initial Digikey search was TL974 (http://www.ti.com/lit/ds/symlink/tl971.pdf), but it has 4mV input offset and no offset null pins - I think that would kill my gain overhead without clipping for a uV-mV order signal. I actually don't really know what to search for to even find op amps with offset null pins. It might not even be the problem I think it is - I'm not sure if the offset voltage gets multiplied by the gain, and I read somewhere that it really only concerns DC signals and not AC signals. Regardless, I'm open to op amp component suggestions.
Moving on to the output side, I'm not sure how the component values were selected for the signal conditioning in the Jameco schematic, but I verified that it worked with the provided values and simulated it with the different components removed to prove to myself that they were all necessary. I plan to use a Schottky diode to rectify instead of a standard one for better signal strength. One thing that confused me in the original schematic was the different values for filter caps - is there a reason for the bigger caps on the lower frequency outputs? I found that the voltage ripple was about the same on each channel with the same capacitor value, and the huge filter cap on the bass appeared to noticeably hurt transient response. Can I use tantalum caps instead of electrolytic to save space? In the same vein, what is the purpose of the coupling cap after the input amp? I didn't see any difference with or without it.
Last question (!) is about the drivers. With the circuit as-shown (+/-6V rails), I wasn't outputting high enough Vbe to put the transistors in saturation, so I was only getting 10mA collector current. I could obviously use a Darlington pair, but would it be better to drive MOSFETs with the BJTs? Without the relatively high Vce, I could maybe put an extra LED in series and lower power dissipation.
To summarize my questions so they don't get lost in the huge post:
- Setting up input gain (fixed and variable)
- Op amp selection - offset voltage important?
- Filter cap sizing
- Output driver selection - Darlington pair vs MOSFET
- Any general component selection suggestions (also any way to reduce the board footprint like useful ICs) or any other ways to improve the design
Thanks a lot for reading this long post and for any suggestions or comments!
After playing with microcontrollers and PWM drivers and the like for a while, I want to take my first stab at building an analog circuit. I chose to design a battery-powered portable light organ using an electret microphone for the input signal, and I think I'm pretty close on the design but am running into some challenges.
I started with this design (http://www.jameco.com/Jameco/workshop/diy/basicledcolororgan-fig11.jpg) from a Jameco video but made some (hopefully!) improvements. My intent is to power it off some CR123A batteries, because they have good energy density, high enough discharge rate for what I need, and I already have some laying around. I'm currently planning on using this cheap mic (http://www.cui.com/product/resource/digikeypdf/cma-4544pf-w.pdf), 20-20000 Hz range, -44 dbV/Pa gain, 3-10V rated voltage.
Here's the circuit (http://i.imgur.com/9J3YEb2.png) I currently have drawn up, I've been doing some LTSpice simulation (though I'm a total novice at it). I hooked up the mic according to the schematic in the data sheet and changed the filters to low pass and high pass filters (aka a three-stage second-order crossover network) instead of the narrow bandpass filters in the original design, and they give the frequency response I want with the component values I calculated and put in the model. I'm not sure if just treating the mic as an AC voltage source is the right way to model it, but I'm not confident enough in my LTSpice skills to model the mechanics of the mic itself. I'm hoping to be able to run off +/- 3V (two batteries, low voltage to dissipate with current-limiting resistors on the LEDs for better battery life), though I'm not sure yet if that's feasible. I could add 2 more for +/-6V. More would probably hurt the portability of the device.
I'm currently using the UniversalOpAmp2 op amp model because the op amp I've been looking at using (more on this because the one I picked out is probably the wrong one) isn't in the default LTSpice library and I don't know how to make my own model yet, and for rectifying diode and transistor I just used some common "classic" components for now.
The major issue I'm encountering is the signal strength of the mic. For -44 dbV/Pa @ 1000mV/1Pa reference, I calculate a signal strength of 6.3 mV/Pa. A pretty usable range of 40 dB to 100 dB SPL gives sound pressure values of 2e-3 to 2 Pa, so signal strength of 12.6 uV (!) to 12.6 mV. I think that range guarantees I will need an adjustable gain stage to avoid clipping the signal at the rails - I verified in simulation that clipping the signal does add significant high frequency components, which I expected based on my knowledge of digital filters but wasn't sure about for analog filters.
Anyway, finally onto the questions. The first two should be easy ones. For the input gain stage, should I use an inverting configuration for low (matched?) input impedance, or a non-inverting configuration for high input impedance? I'm not sure which is better for input from a microphone. I'm fairly sure I want the high impedance configuration for the filter stages so they don't affect the rest of the circuit. Second, in the shown circuit, I set up the first stage gain to the highest without clipping for the maximum input signal, with some margin (3v / 0.0126V = 238, so 200 to give some margin, or 400 with +/-6V rails, as shown). Should I move some of that gain to the filter stages to avoid possible GBW limitations? I'm not super concerned with the fidelity of the signals as long as they switch the LEDs on and off, if that is relevant to the question.
On the gain topic, I would (obviously) need adjustable gain from 1-1000 to get the same signal strength between 40 dB and 100 dB SPL. Does it matter if the adjustable gain stage goes before or after the fixed gain stage? 1000x gain would require 20MHz GBW, which isn't too tough to find, but I'm not sure if I would run into any gain/phase margin issues at that level of gain and frequency - would it be better to instead spread the gain over two op amps (connected to the same pot)?
The biggest question for me is op amp selection. My needs are pretty demanding - rail-to-rail, high GBW, low input voltage requirements, either very low input offset or offset null pins (maybe?), and DIP package. My current simulation effectively uses an ideal op amp model for lack of a better component to use (or the knowledge to make my own component model). The one I had been looking at based on an initial Digikey search was TL974 (http://www.ti.com/lit/ds/symlink/tl971.pdf), but it has 4mV input offset and no offset null pins - I think that would kill my gain overhead without clipping for a uV-mV order signal. I actually don't really know what to search for to even find op amps with offset null pins. It might not even be the problem I think it is - I'm not sure if the offset voltage gets multiplied by the gain, and I read somewhere that it really only concerns DC signals and not AC signals. Regardless, I'm open to op amp component suggestions.
Moving on to the output side, I'm not sure how the component values were selected for the signal conditioning in the Jameco schematic, but I verified that it worked with the provided values and simulated it with the different components removed to prove to myself that they were all necessary. I plan to use a Schottky diode to rectify instead of a standard one for better signal strength. One thing that confused me in the original schematic was the different values for filter caps - is there a reason for the bigger caps on the lower frequency outputs? I found that the voltage ripple was about the same on each channel with the same capacitor value, and the huge filter cap on the bass appeared to noticeably hurt transient response. Can I use tantalum caps instead of electrolytic to save space? In the same vein, what is the purpose of the coupling cap after the input amp? I didn't see any difference with or without it.
Last question (!) is about the drivers. With the circuit as-shown (+/-6V rails), I wasn't outputting high enough Vbe to put the transistors in saturation, so I was only getting 10mA collector current. I could obviously use a Darlington pair, but would it be better to drive MOSFETs with the BJTs? Without the relatively high Vce, I could maybe put an extra LED in series and lower power dissipation.
To summarize my questions so they don't get lost in the huge post:
- Setting up input gain (fixed and variable)
- Op amp selection - offset voltage important?
- Filter cap sizing
- Output driver selection - Darlington pair vs MOSFET
- Any general component selection suggestions (also any way to reduce the board footprint like useful ICs) or any other ways to improve the design
Thanks a lot for reading this long post and for any suggestions or comments!
