Hi everybody! I'm a newbie seeking to design my own amplifier. I would be delighted if someone from this hobby could guide me along. Let me try to explain the features and what was intended.
1. I'm designing a stereo amplifier for Headphones/IEMs.
2. I use 1x NE5532 for each channel for a HPF (15Hz - 23kHZ) -> LPF + HPF = HBF
3. I then used 1x LME49720 for each channel for the unity buffer stage + "power" amplification stage.
4. Each op-amp ICs are connected to 100nF & 1uF capacitors for both power rails as decoupling capacitors.
5. I'm using 3.5mm TRS for now, but I'm considering a 4.4 TRRS Balanced connection as well. Maybe both?
6. This is a derivative from the "47-amp", but I'm thinking of making it into a portable amplifier...
7. I understand that L48xx/L49xx voltage regulator ICs is common, but it seem to be far too noisy as the power supply for audio applications and will need further circuit additions to be usable. Is there any recommendations for ICs replacement? With portable in mind, OOTB regulator ICs would be preferred?
8. I intend to do op-amp rolling [the LME49720s] so my first prototype would be DIP8 package for now on a breadboard. If it works, I will move on to PCB design using SOIC chips, the passive components shall be surface mount as well.
9. I will be separating the ground into signal ground (which is use AGND here) and true ground (GND)
Any feedback is welcomed 😀 2 random question out of this project:
1. Does it makes sense to replace the unity-buffer stage with JFET transistor amp design? I heard it makes the amp slightly colored to be more warm... in fact if I could implement both and choose between JFET (Warm) and Op-amp (Sterile) in one circuit would be fantastic!!
2. Am I the only one that thinks the commercial "portable amplifiers" are much larger than needed?
1. I'm designing a stereo amplifier for Headphones/IEMs.
2. I use 1x NE5532 for each channel for a HPF (15Hz - 23kHZ) -> LPF + HPF = HBF
3. I then used 1x LME49720 for each channel for the unity buffer stage + "power" amplification stage.
4. Each op-amp ICs are connected to 100nF & 1uF capacitors for both power rails as decoupling capacitors.
5. I'm using 3.5mm TRS for now, but I'm considering a 4.4 TRRS Balanced connection as well. Maybe both?
6. This is a derivative from the "47-amp", but I'm thinking of making it into a portable amplifier...
7. I understand that L48xx/L49xx voltage regulator ICs is common, but it seem to be far too noisy as the power supply for audio applications and will need further circuit additions to be usable. Is there any recommendations for ICs replacement? With portable in mind, OOTB regulator ICs would be preferred?
8. I intend to do op-amp rolling [the LME49720s] so my first prototype would be DIP8 package for now on a breadboard. If it works, I will move on to PCB design using SOIC chips, the passive components shall be surface mount as well.
9. I will be separating the ground into signal ground (which is use AGND here) and true ground (GND)
Any feedback is welcomed 😀 2 random question out of this project:
1. Does it makes sense to replace the unity-buffer stage with JFET transistor amp design? I heard it makes the amp slightly colored to be more warm... in fact if I could implement both and choose between JFET (Warm) and Op-amp (Sterile) in one circuit would be fantastic!!
2. Am I the only one that thinks the commercial "portable amplifiers" are much larger than needed?
After giving it a good look over, I think you need to uncouple yourself from this design (cart way before the horse), take a couple of steps back, and clarify the design intent and specifications you are trying to design for. Those do not appear to be present, and will drive the topology of the design.
1. What is your source? Phone, DAQ, DAP, etc? Is it portable, what connectivity, output impedance and does the source have volume/tone controls?
2. What headphones are you driving with this? What is their impedance, and possible lowest impedance if multiple/range of sets? What is their connectivity?
3. Do you want to ever drive speakers with this? This will require more power than for phones and needs to be planned for up front. What output power?
4. Do you want this portable (battery) or desktop (adapter) or both? Significant difference/difficulty in design depending.
5. What size are you shooting for? Pocket, or desktop box? =Significant= compromises in performance or increase in complexity if you want pocket size.
6. Why the filtering? This is not EQ, and is not normally needed at all- do you have frequency concerns such that filtering is required?
Please answer clearly the above questions, then I can help.
So far, obvious problems-
1. Why the filtering? You may want an input (passive) filter to block RF. The rest really is not needed unless you have a really bad source?
2. The buffer you have chosen only drives 600 Ohms. Op-amps are terrible for driving modern headphones, only used for low power, higher-Z phones.
3. Depending upon portability requirements, it all starts with the power supply- that is the most difficult part to get right.
4. Regulator choices come after answering basic power requirements and power topology.
5. Separating GND is a decision for later in the design, what you do with this depends upon point to point, through hole, or PCB.
1. What is your source? Phone, DAQ, DAP, etc? Is it portable, what connectivity, output impedance and does the source have volume/tone controls?
2. What headphones are you driving with this? What is their impedance, and possible lowest impedance if multiple/range of sets? What is their connectivity?
3. Do you want to ever drive speakers with this? This will require more power than for phones and needs to be planned for up front. What output power?
4. Do you want this portable (battery) or desktop (adapter) or both? Significant difference/difficulty in design depending.
5. What size are you shooting for? Pocket, or desktop box? =Significant= compromises in performance or increase in complexity if you want pocket size.
6. Why the filtering? This is not EQ, and is not normally needed at all- do you have frequency concerns such that filtering is required?
Please answer clearly the above questions, then I can help.
So far, obvious problems-
1. Why the filtering? You may want an input (passive) filter to block RF. The rest really is not needed unless you have a really bad source?
2. The buffer you have chosen only drives 600 Ohms. Op-amps are terrible for driving modern headphones, only used for low power, higher-Z phones.
3. Depending upon portability requirements, it all starts with the power supply- that is the most difficult part to get right.
4. Regulator choices come after answering basic power requirements and power topology.
5. Separating GND is a decision for later in the design, what you do with this depends upon point to point, through hole, or PCB.
@wparks many thanks for your advice. let me answer your query and start it right.
1. The source would be phone/DAP, purely wired connection (3.5mm, 4.4mm balanced or both). The output impedance of the source let's base on the DAP which I believe is <2ohm.
2. The key output device I'm using would be IEMs (typically 10-50ohms), capable to drive headphones of 600ohms is a plus.
3. I have no intention to drive speakers at this point, so let's leave it as it is for now.
4. The ideal answer would be to have the capability of both battery (portability) and power adapter (desktop) power. But I would prefer portable over desktop
5. My ultimate goal is something portable, pocket in size.
6. In terms of complexity vs compromises, I think this part is something I really need help and advice to get a sweet spot?
7. I wasn't aware that filtering is not normally required, back in school my teacher have always insisted filtering. But probably due to breadboard circuit where EMI can run bad.
I'm using op-amp for now simply because it would be my first project and want to start small and slow. Although I was taught transistor in school, it was all theory and didn't have the chance to play with them practical. It would likely be another project (maybe 2-3 down the road) for a transistor amp approach.
Yes, I agree having a good power supply is the cornerstone of a good amplifier, thus I really need help building an stable, efficient & low noise one.
I would want it to be PCB, but to start off... everything will be on breadboard.
Once again, many thanks for helping me.
1. The source would be phone/DAP, purely wired connection (3.5mm, 4.4mm balanced or both). The output impedance of the source let's base on the DAP which I believe is <2ohm.
2. The key output device I'm using would be IEMs (typically 10-50ohms), capable to drive headphones of 600ohms is a plus.
3. I have no intention to drive speakers at this point, so let's leave it as it is for now.
4. The ideal answer would be to have the capability of both battery (portability) and power adapter (desktop) power. But I would prefer portable over desktop
5. My ultimate goal is something portable, pocket in size.
6. In terms of complexity vs compromises, I think this part is something I really need help and advice to get a sweet spot?
7. I wasn't aware that filtering is not normally required, back in school my teacher have always insisted filtering. But probably due to breadboard circuit where EMI can run bad.
I'm using op-amp for now simply because it would be my first project and want to start small and slow. Although I was taught transistor in school, it was all theory and didn't have the chance to play with them practical. It would likely be another project (maybe 2-3 down the road) for a transistor amp approach.
Yes, I agree having a good power supply is the cornerstone of a good amplifier, thus I really need help building an stable, efficient & low noise one.
I would want it to be PCB, but to start off... everything will be on breadboard.
Once again, many thanks for helping me.
Ok, let's break it down.
1. (DAP, very low source impedance.)
This is good, this will have volume control already, and tone controls, and output a very nice clean signal excellent for low noise, no filtering inside the phones amp is necessary.
Driving 10 - 50 Ohms definitely puts this outside of the realm of op-amps. People will play games using multiple parallel op-amps, etc, but it just really isn't practical for this difficult of a load. You're down into speaker impedance range (at lower power), so you are most likely in the small chip-amp territory. When you can do this, driving higher impedance headphones (300-600) is a cake, as you will probably have lots of output voltage from your amp.
3. (Speakers) You probably can, small desktop speakers at least, you will have enough power and voltage output.
Here is where it gets tricky. In the 'olden days' we could knock out something simple with a couple of 9V batteries and a transistor circuit because most of the common headphones were in the 300 Ohm range or higher. Driving such low impedance IEMs will require more current (lower output resistance) than 9V batteries can provide. You can still do this with a bank of AA's or AAA's, but now that starts getting heavier and cumbersome.
The solution modern manufacturers use is Lithium cell phone type batteries, running ~3.7V, which can output a LOT of current, then use tiny switching supplies to bring the voltage up to whatever the amplifier needs. But then you need a charger circuit to properly and safely charge the battery, etc. I think building the switching supply and charging architecture for a built-in lithium battery would lead me to suggest you just purchase a rechargeable portable headphone amplifier, as this problem has been solved and many many exist to choose from, and will be much smaller and safer than you can build.
Since this is likely a 'learning project' first and foremost, you may want to consider consigning this to desktop only, where power solutions abound, space is more available, and the path to a functional prototype is fairly short so you can learn what you need and evaluate what your next step is.
This is perhaps the most challenging phase of a design. Everything is a compromise, but you can't make trade-offs unless you have a clear definition of the desired requirements/specs for the final project. Figure out must-have's and non-starters first, then areas where you have flexibility. You will find that the vision you had for your project at the beginning will evolve as you understand the actual costs, difficulties and limitations of the differing options. My designs, whatever they are, spend a considerable time evolving and working out kinks both in my mind and on paper before I ever start drawing a schematic. Hopefully what you are able to achieve in the end resembles your beginning idea. It's OK if it does not, it's a learning process. Finding a solution that is truly elegant, simple but powerful, is rare, but so rewarding. They are out there, but I will warn you, there are a LOT of very smart people out there ahead of you, and they have probably already found the solution you are looking for, worked out the kinks, and published the schematic. Most of the wheels have already been invented. Don't be afraid to copy what has already been done. Copy exact, or modify and make it your own. It saves a LOT of time and hair pulling.
I've got a simple project idea I think you are going to like- Rod Elliott's (Elliott Sound Products) Hi-Fi Headphone Amp Project #113
This little amplifier runs on +/- 15V, and uses a 2-transistor output stage driven by an op-amp. The transistors are inside the op-amp feedback loop! It is relatively easy to build, very inexpensive, and has impressive performance. (lots of power and incredibly low distortion) It will have no problem driving whatever headphone or small speaker you are interested in. I built this, and it sounds great. As is, it has too much gain, but in the description he indicates how to change one resistor to tame it down, and I have more info on that. I can share with you the perf-board component layout I came up with, or even buy Rod's PCB if you want a faster build. Don't sweat the power supply, it's a pretty easy build, and there is lots of help on that one too.
1. (DAP, very low source impedance.)
This is good, this will have volume control already, and tone controls, and output a very nice clean signal excellent for low noise, no filtering inside the phones amp is necessary.
Driving 10 - 50 Ohms definitely puts this outside of the realm of op-amps. People will play games using multiple parallel op-amps, etc, but it just really isn't practical for this difficult of a load. You're down into speaker impedance range (at lower power), so you are most likely in the small chip-amp territory. When you can do this, driving higher impedance headphones (300-600) is a cake, as you will probably have lots of output voltage from your amp.
3. (Speakers) You probably can, small desktop speakers at least, you will have enough power and voltage output.
Here is where it gets tricky. In the 'olden days' we could knock out something simple with a couple of 9V batteries and a transistor circuit because most of the common headphones were in the 300 Ohm range or higher. Driving such low impedance IEMs will require more current (lower output resistance) than 9V batteries can provide. You can still do this with a bank of AA's or AAA's, but now that starts getting heavier and cumbersome.
The solution modern manufacturers use is Lithium cell phone type batteries, running ~3.7V, which can output a LOT of current, then use tiny switching supplies to bring the voltage up to whatever the amplifier needs. But then you need a charger circuit to properly and safely charge the battery, etc. I think building the switching supply and charging architecture for a built-in lithium battery would lead me to suggest you just purchase a rechargeable portable headphone amplifier, as this problem has been solved and many many exist to choose from, and will be much smaller and safer than you can build.
Since this is likely a 'learning project' first and foremost, you may want to consider consigning this to desktop only, where power solutions abound, space is more available, and the path to a functional prototype is fairly short so you can learn what you need and evaluate what your next step is.
This is perhaps the most challenging phase of a design. Everything is a compromise, but you can't make trade-offs unless you have a clear definition of the desired requirements/specs for the final project. Figure out must-have's and non-starters first, then areas where you have flexibility. You will find that the vision you had for your project at the beginning will evolve as you understand the actual costs, difficulties and limitations of the differing options. My designs, whatever they are, spend a considerable time evolving and working out kinks both in my mind and on paper before I ever start drawing a schematic. Hopefully what you are able to achieve in the end resembles your beginning idea. It's OK if it does not, it's a learning process. Finding a solution that is truly elegant, simple but powerful, is rare, but so rewarding. They are out there, but I will warn you, there are a LOT of very smart people out there ahead of you, and they have probably already found the solution you are looking for, worked out the kinks, and published the schematic. Most of the wheels have already been invented. Don't be afraid to copy what has already been done. Copy exact, or modify and make it your own. It saves a LOT of time and hair pulling.
I've got a simple project idea I think you are going to like- Rod Elliott's (Elliott Sound Products) Hi-Fi Headphone Amp Project #113
This little amplifier runs on +/- 15V, and uses a 2-transistor output stage driven by an op-amp. The transistors are inside the op-amp feedback loop! It is relatively easy to build, very inexpensive, and has impressive performance. (lots of power and incredibly low distortion) It will have no problem driving whatever headphone or small speaker you are interested in. I built this, and it sounds great. As is, it has too much gain, but in the description he indicates how to change one resistor to tame it down, and I have more info on that. I can share with you the perf-board component layout I came up with, or even buy Rod's PCB if you want a faster build. Don't sweat the power supply, it's a pretty easy build, and there is lots of help on that one too.
Heyss!
I really appreciate your help~ Thank you so much! Little have I expected a stranger would put in so much effort to guide me along. The project you showed me certainly seem interesting and convinced me to try it out. But I still would like to design something myself no matter how simple it would be.
And you are right that I shouldn't be so ambitious to go for a portable build at this stage... what I need is to boost my knowledge and confidence. So... let's scrap the idea of portable for now and design a desktop headphone amp. For now, I'm gonna start off stage by stage and gel them together piece by piece.
Part I (Power supply):
(a) I guess this part I should just buy? I doubt I can design any thing better than commercially available and this part is basically decides say 70% of the outcome? Is there something you could recommend?
Part II (Buffer + Amplification):
Let me start off clearing my doubts first. From what I gather from your previous reply is lower impedance devices (referring to IEMs and lower impedance headphones, 10-60ohms) requires a lot more current to drive right? Doesn't a buffer follower stage resolve the issue?
You also mentioned that op-amp application does not seem viable for my needs. So I'm only left with the option of transistor amp? What are some potential applications for op-amp amps?
I'm leaving the expectations and compromises in my next reply, after I hear your thoughts to the above... For now I'm gonna go read the project u shared and try to understand the ciruit.
I really appreciate your help~ Thank you so much! Little have I expected a stranger would put in so much effort to guide me along. The project you showed me certainly seem interesting and convinced me to try it out. But I still would like to design something myself no matter how simple it would be.
And you are right that I shouldn't be so ambitious to go for a portable build at this stage... what I need is to boost my knowledge and confidence. So... let's scrap the idea of portable for now and design a desktop headphone amp. For now, I'm gonna start off stage by stage and gel them together piece by piece.
Part I (Power supply):
(a) I guess this part I should just buy? I doubt I can design any thing better than commercially available and this part is basically decides say 70% of the outcome? Is there something you could recommend?
Part II (Buffer + Amplification):
Let me start off clearing my doubts first. From what I gather from your previous reply is lower impedance devices (referring to IEMs and lower impedance headphones, 10-60ohms) requires a lot more current to drive right? Doesn't a buffer follower stage resolve the issue?
You also mentioned that op-amp application does not seem viable for my needs. So I'm only left with the option of transistor amp? What are some potential applications for op-amp amps?
I'm leaving the expectations and compromises in my next reply, after I hear your thoughts to the above... For now I'm gonna go read the project u shared and try to understand the ciruit.
In my experience, the NJM4556 opamp will drive low impedance headphones at reasonable levels (i.e. not hearing-damaging) without problems. I have my opinions on opamp rolling but if you really want to do it, you could use the 4556 as a unity gain output buffer and have an opamp gain stage before it (volume control in between) where you can try pretty much any opamp you like. High and low pass filtering can be easily included without any additional opamp stages.
@cabirio
mind sharing your opinions on opamp rolling? what is your concern?
actually... from what you have commented, isnt that what my original design was? the only difference was that I implemented a active BPF for both channels using the NE5532 chip. I then use LME49720 for an unity-gain buffer stage followed by an "power" gain stage. Like what you have mentioned, I also implemented a volume control using a potentiometer between the unity-gain and "power" gain stage.
I understand that NJM4556 is better in terms of output current... would it make more sense for me to use 1x NJM4556 (for both channels buffer) and 1x LME49720 for power gain?
Back on the issue of BPF, do you think it is even necessary? Is a passive filter enough? I have been taught in school to implement active filter whenever possible due to the steeper cut-off...
anyways, what is your take? do you agree with wparks that my original schematics is quite problematic?
mind sharing your opinions on opamp rolling? what is your concern?
actually... from what you have commented, isnt that what my original design was? the only difference was that I implemented a active BPF for both channels using the NE5532 chip. I then use LME49720 for an unity-gain buffer stage followed by an "power" gain stage. Like what you have mentioned, I also implemented a volume control using a potentiometer between the unity-gain and "power" gain stage.
I understand that NJM4556 is better in terms of output current... would it make more sense for me to use 1x NJM4556 (for both channels buffer) and 1x LME49720 for power gain?
Back on the issue of BPF, do you think it is even necessary? Is a passive filter enough? I have been taught in school to implement active filter whenever possible due to the steeper cut-off...
anyways, what is your take? do you agree with wparks that my original schematics is quite problematic?
My concern with opamp rolling is that if you consider the specific requirements of your circuit in terms of topology, supply voltage and current, input impedance, noise, load driving capability, DC offset, bandwidth, stability, etc., any difference in sound you may perceive between different opamps that meet those requirements adequately is going to be due expectation bias. On the other hand I know it's fun (I went through an opamp rolling phase myself decades ago) and there isn't much harm you can do, so if you want to do it, go for it.
The circuit I propose would be something like this:
From left to right:
- R1/C1 is a low pass, in this case with F3 = 720 kHz. Yes, it is a good idea to have some RF filtering at the input. Note that if the source has significant output impedance, the F3 will go down accordingly and you may want to decrease C1.
- C2 blocks the source from any DC at the + input of U1, which could be significant if it's a bipolar opamp. R2 is a bleed resistor for C2 and R3 biases U1. The overall input impedance is ~20k, easy for any source.
- R4/R5 set the gain of U1 to ~4, which I find adequate most of the time. You can adjust to taste. C4 is for stability.
- I find channel imbalances most annoying on headphones. Since Log pots are notorious for bad balance between tracks, here I use a linear 20k pot loaded by the 4k7 biasing resistor of U2, which results in a roughly Log response with better channel balance. C5 is required to prevent any DC current through the pot wiper, which will make it scratchy.
- Also, R7 being small ensures the DC offset at the output of U2 is kept negligible (~280 uV in this case).
- The total load seen by U1 here is ~2k, which many decent opamps handle without problems.
- C2/R3, C3/R4 and C45/R7 will determine your high pass cutoff frequency, which you can adjust to your requirements.
The only "problem" with your circuit is that it's unnecessarily complicated (and lacks some things like DC blocking between stages). Yes, you can get steeper slopes with active filters (not the ones you have though), but this isn't required in this application. Also, if you're considering making it portable, any extra opamp stages you add will result in shorter battery life.
The circuit I propose would be something like this:
From left to right:
- R1/C1 is a low pass, in this case with F3 = 720 kHz. Yes, it is a good idea to have some RF filtering at the input. Note that if the source has significant output impedance, the F3 will go down accordingly and you may want to decrease C1.
- C2 blocks the source from any DC at the + input of U1, which could be significant if it's a bipolar opamp. R2 is a bleed resistor for C2 and R3 biases U1. The overall input impedance is ~20k, easy for any source.
- R4/R5 set the gain of U1 to ~4, which I find adequate most of the time. You can adjust to taste. C4 is for stability.
- I find channel imbalances most annoying on headphones. Since Log pots are notorious for bad balance between tracks, here I use a linear 20k pot loaded by the 4k7 biasing resistor of U2, which results in a roughly Log response with better channel balance. C5 is required to prevent any DC current through the pot wiper, which will make it scratchy.
- Also, R7 being small ensures the DC offset at the output of U2 is kept negligible (~280 uV in this case).
- The total load seen by U1 here is ~2k, which many decent opamps handle without problems.
- C2/R3, C3/R4 and C45/R7 will determine your high pass cutoff frequency, which you can adjust to your requirements.
The only "problem" with your circuit is that it's unnecessarily complicated (and lacks some things like DC blocking between stages). Yes, you can get steeper slopes with active filters (not the ones you have though), but this isn't required in this application. Also, if you're considering making it portable, any extra opamp stages you add will result in shorter battery life.
Last edited:
@cabirio
Allow me to start off with some questions on your proposed circuit.
1. Why you set the -3db mark at 720khz? Doesn't that defeat the purpose of a LPF to begin with (more than a decade above 22khz)?
2. Doesn't C2 and R3 form a passive HPF?
3. The log pot you mentioned is referring to r6 right?
4. U2 is working as a unity gain buffer right? Won't it be more stable to have a negative feedback resistor?
And now back to my design...:
I think everyone here agrees that having a LPF is beneficial (and a passive one would suffice). However there is no need for a HPF to form BPF.
So I'm proposing to reduce 1 NE5532 op-amp and only use 1 NE5532 to make an active LPF with cut-off frequency at roughly 23khz.
Followed by the issue of DC offset (I believe that is what you refer as DC blocking), how do I decide what cap value to use? I should add 1 at every op-amp input right?
In fact I'm puzzled by your statement, "Yes, you can get steeper slopes with active filters (not the ones you have though)". Is there some thing with my filter?
Allow me to start off with some questions on your proposed circuit.
1. Why you set the -3db mark at 720khz? Doesn't that defeat the purpose of a LPF to begin with (more than a decade above 22khz)?
2. Doesn't C2 and R3 form a passive HPF?
3. The log pot you mentioned is referring to r6 right?
4. U2 is working as a unity gain buffer right? Won't it be more stable to have a negative feedback resistor?
And now back to my design...:
I think everyone here agrees that having a LPF is beneficial (and a passive one would suffice). However there is no need for a HPF to form BPF.
So I'm proposing to reduce 1 NE5532 op-amp and only use 1 NE5532 to make an active LPF with cut-off frequency at roughly 23khz.
Followed by the issue of DC offset (I believe that is what you refer as DC blocking), how do I decide what cap value to use? I should add 1 at every op-amp input right?
In fact I'm puzzled by your statement, "Yes, you can get steeper slopes with active filters (not the ones you have though)". Is there some thing with my filter?
1. There's no reason / benefit to low-passing your signal strictly to the audible range. You can have substantially flat frequency and phase response up to 20 kHz if you low-pass higher. As I said it is a good idea to have some filtering to reject RF interference. The F3 isn't critical, you can go lower but, as I said, the output impedance of the source will lower it, so here I've left a healthy margin just in case.
2. Yes, as I said, C2/R3, C3/R4 and C5/R7 are all high pass filters. Well, to be precise, C3/R4 is a low shelf that raises the overall cut-off frequency. With the values shown, the frequency response is down ~0.1 dB @ 20 Hz and the final slope is 12 dB/oct.
3. R6 is a linear pot, which loaded by R7, behaves like a log one, see here.
4. The NJM4556 is stable at unity gain, no feedback resistor required.
You need DC blocking at the input (you don't want to put a DC voltage on the output of your source), after the pot wiper (see above) and, if there was significant DC offset at the output, there as well, but it's not the case here. Between other stages, it may or may not be required depending on the specific circuit.
Your filters, though active, are first-order, i.e. 6 dB/oct. You don't need steeper slopes but, if you did, you'd have to use different topologies, see e.g. here.
2. Yes, as I said, C2/R3, C3/R4 and C5/R7 are all high pass filters. Well, to be precise, C3/R4 is a low shelf that raises the overall cut-off frequency. With the values shown, the frequency response is down ~0.1 dB @ 20 Hz and the final slope is 12 dB/oct.
3. R6 is a linear pot, which loaded by R7, behaves like a log one, see here.
4. The NJM4556 is stable at unity gain, no feedback resistor required.
You need DC blocking at the input (you don't want to put a DC voltage on the output of your source), after the pot wiper (see above) and, if there was significant DC offset at the output, there as well, but it's not the case here. Between other stages, it may or may not be required depending on the specific circuit.
Your filters, though active, are first-order, i.e. 6 dB/oct. You don't need steeper slopes but, if you did, you'd have to use different topologies, see e.g. here.
I don't see a need for higher order filter as well. So these are things I will make amendment to my initial design:
1. Remove the HPF section
2. Redesign the active 1st order LPF to cutoff at 40khz
3. Replace the unity gain buffer stage with a NJM4556
4. Op-amp rolling for the potentiometer controlled gain stage. Will study the linear pot with a biasing resistor topology shared.
5. Adding suitable caps to deter DC offset at the inputs to the op-amp.
6a. Assuming a +/- 15V power supply. I should have more than enough power.
6b. Future proofing comments for portable implementation : lithium ion battery regulated to +/-9v?
1. Remove the HPF section
2. Redesign the active 1st order LPF to cutoff at 40khz
3. Replace the unity gain buffer stage with a NJM4556
4. Op-amp rolling for the potentiometer controlled gain stage. Will study the linear pot with a biasing resistor topology shared.
5. Adding suitable caps to deter DC offset at the inputs to the op-amp.
6a. Assuming a +/- 15V power supply. I should have more than enough power.
6b. Future proofing comments for portable implementation : lithium ion battery regulated to +/-9v?
So much noise and fluff going on here since I was able to chime in, I'm really not sure where to start.
The reason you don't use op-amps to drive loads below several hundred ohms is because they are not able to source enough current. They do not have a low enough output resistance. They are incredibly versatile, easy to use, but they are not output drivers. Using them in that role is like putting a go-cart engine in a car. Sure, it "works" but you won't be happy with the performance. Specialized versions of the go-cart engine are still go-cart engines. There are far better circuits for the output stage.
BTW I pointed you to a most excellent one, that is an incredible learning opportunity. That circuit combines an op-amp to provide all of those benefits, including incredible open-loop gain and high speed, with an incredibly simple but powerful diode biased class AB complimentary output that is able to deliver a LOT of current (so much so that it is tamed back by the 10 ohm emitter degeneration resistors), and because the output stage is enclosed within the op-amp's feedback loop the performance is incredible, distortion is essentially below measurable, you can roll the op-amp without changing anything other than the sound, and it will even drive speakers. Notice no output capacitor to cause distortion- The amp is biased to ground. This circuit will teach you a tremendous amount about power amplifiers that use this same basic ubiquitous topology- You will learn about class AB operation, complimentary transistor pairs, outout stage biasing and quiescent current, emitter degeneration, global feedback, and even zobel networks. That's a LOT of prime audio learning in one simple, inexpensive project.
If you would rather go shake-and-bake and use modular components, I can start looking into the wide wide world of chip-amps, any one of which will be very capable, make this a very quick and easy project with minimal learning.
Some fluff I have to go after-
Please don't do any active filtering. I know your professor pounded it into you as an exercise, but it is not needed here at all. It only adds complexity, parts count, and of course noise and distortion. The only reason I would include even a passive low pass filter at the input is that your digital source might have some stray RF riding on the audio cable. Your amplifier will likely have wide bandwidth, and no reason to spend power amplifying frequencies you can't hear, but no reason to use active filtering. Put f3db >> 100Khz (it's RF after all ) and don't worry further about it.
Inter-stage coupling caps are used far too often. In the above design, C2 and C5 are not needed. U1 is biased to ground, U2 is biased to ground. There is =no= voltage differential of any consequence between them. It is good practice to put a DC blocking coupling cap at the input of your design, because you can never really guarantee that all of the sources you will use do not have some DC offset that will show up downstream. But keep in mind, coupling caps are another source of distortion and noise. If you are going to use one, make sure you need it (actual DC to block), and choose a film capacitor, any of the poly, polypropylene, MKT, type caps will work. 'Wima' is an excellent name, but almost anything film will work. Avoid ceramic, electrolytic (bipolar or polarized) or tantalum for audio signal coupling. Electrolytic coupling caps are only used at amplifier outputs when you have to block significant DC voltage, and they are driving a speaker type load.
More later on the power supply, and whatever else comes up.
The reason you don't use op-amps to drive loads below several hundred ohms is because they are not able to source enough current. They do not have a low enough output resistance. They are incredibly versatile, easy to use, but they are not output drivers. Using them in that role is like putting a go-cart engine in a car. Sure, it "works" but you won't be happy with the performance. Specialized versions of the go-cart engine are still go-cart engines. There are far better circuits for the output stage.
BTW I pointed you to a most excellent one, that is an incredible learning opportunity. That circuit combines an op-amp to provide all of those benefits, including incredible open-loop gain and high speed, with an incredibly simple but powerful diode biased class AB complimentary output that is able to deliver a LOT of current (so much so that it is tamed back by the 10 ohm emitter degeneration resistors), and because the output stage is enclosed within the op-amp's feedback loop the performance is incredible, distortion is essentially below measurable, you can roll the op-amp without changing anything other than the sound, and it will even drive speakers. Notice no output capacitor to cause distortion- The amp is biased to ground. This circuit will teach you a tremendous amount about power amplifiers that use this same basic ubiquitous topology- You will learn about class AB operation, complimentary transistor pairs, outout stage biasing and quiescent current, emitter degeneration, global feedback, and even zobel networks. That's a LOT of prime audio learning in one simple, inexpensive project.
If you would rather go shake-and-bake and use modular components, I can start looking into the wide wide world of chip-amps, any one of which will be very capable, make this a very quick and easy project with minimal learning.
Some fluff I have to go after-
Please don't do any active filtering. I know your professor pounded it into you as an exercise, but it is not needed here at all. It only adds complexity, parts count, and of course noise and distortion. The only reason I would include even a passive low pass filter at the input is that your digital source might have some stray RF riding on the audio cable. Your amplifier will likely have wide bandwidth, and no reason to spend power amplifying frequencies you can't hear, but no reason to use active filtering. Put f3db >> 100Khz (it's RF after all ) and don't worry further about it.
Inter-stage coupling caps are used far too often. In the above design, C2 and C5 are not needed. U1 is biased to ground, U2 is biased to ground. There is =no= voltage differential of any consequence between them. It is good practice to put a DC blocking coupling cap at the input of your design, because you can never really guarantee that all of the sources you will use do not have some DC offset that will show up downstream. But keep in mind, coupling caps are another source of distortion and noise. If you are going to use one, make sure you need it (actual DC to block), and choose a film capacitor, any of the poly, polypropylene, MKT, type caps will work. 'Wima' is an excellent name, but almost anything film will work. Avoid ceramic, electrolytic (bipolar or polarized) or tantalum for audio signal coupling. Electrolytic coupling caps are only used at amplifier outputs when you have to block significant DC voltage, and they are driving a speaker type load.
More later on the power supply, and whatever else comes up.
Revised. I'm still out and jotting down notes for amendment in the evening.
1. Design a 1st order passive LPF to cutoff at 100khz
2. Replace the unity gain buffer stage with a NJM4556
3. Op-amp rolling for the potentiometer controlled gain stage. Will study the linear pot with a biasing resistor topology shared.
4a. Assuming a +/- 15V power supply. I should have more than enough power.
4b. Future proofing comments for portable implementation : lithium ion battery regulated to +/-9v?
1. Design a 1st order passive LPF to cutoff at 100khz
2. Replace the unity gain buffer stage with a NJM4556
3. Op-amp rolling for the potentiometer controlled gain stage. Will study the linear pot with a biasing resistor topology shared.
4a. Assuming a +/- 15V power supply. I should have more than enough power.
4b. Future proofing comments for portable implementation : lithium ion battery regulated to +/-9v?
You have the opamps the wrong way round, the 4556 is the one with high current drive capability, so that's the one you want at the output. You need to bias the + input of the first opamp with a resistor to ground. Depending on which opamp you put there and what input impedance you're aiming for (which will determine the value of that resistor), the bias current may put a significant DC voltage on the input, hence the requirement for a DC blocking cap. The feedback arrangement of the second opamp doesn't make sense, you're feeding the signal to both the + and - inputs. And again, you do need a DC blocking cap after the pot wiper and then a resistor to ground to bias the + input of the second opamp, to a) make sure no current goes through the pot wiper and b) the biasing of the opamp doesn't depend on the pot position, which will result in a variable DC offset at the output.
Please take another look at the schematic I posted above, read carefully my comments and ask for clarification on anything I left out or you didn't understand in my explanation.
Please take another look at the schematic I posted above, read carefully my comments and ask for clarification on anything I left out or you didn't understand in my explanation.
Ok, ignore a direct warning- The NJM4556 is a dog, and won't drive your 32 ohm IEM's much less any lower. If that is the goal, as is, you are wasting your time. The 4556's datasheet is proud that it can drive 150 Ohm loads. If you HAVE to use an op-amp, at least parallel two as is done in this example contributed to Rod Elliott's site. (Disregard the "crossfeed" crap, just look at the U2/U3 output stage. There is probably a lot in this project that will appeal to you. I don't think it will drive these low loads very well, especially with 9V batteries, but he seems pretty happy. (Crossfeed isn't magic, it just shrinks your soundstage.) BTW Rod has a LOT of designs at his site that use single and dual rail op-amps in a myriad of different applications, and you can learn a lot from his designs and explanations.
Since you bypassed the opportunity to break your transistor cherry and build something that =will= actually cleanly drive 10 Ohm IEM's with authority, there are some other less intimidating shake and bake options. Check out the BUF634. It's a bandaid for op-amp current limits, but works very well. Since you want through-hole your best bet to find two cheap would be Ebay. Otherwise, either surface mount, pricey, or both.
If you want even more shake and bake, try the TPA6120. It's surface mount only, but you can get PCB modules for cheap on Ebay. If you are going that way, there are a TON of PCB modules on ebay, from Class D, to BTL styles, etc. Sky is the limit when it's being built for you.
Since you bypassed the opportunity to break your transistor cherry and build something that =will= actually cleanly drive 10 Ohm IEM's with authority, there are some other less intimidating shake and bake options. Check out the BUF634. It's a bandaid for op-amp current limits, but works very well. Since you want through-hole your best bet to find two cheap would be Ebay. Otherwise, either surface mount, pricey, or both.
If you want even more shake and bake, try the TPA6120. It's surface mount only, but you can get PCB modules for cheap on Ebay. If you are going that way, there are a TON of PCB modules on ebay, from Class D, to BTL styles, etc. Sky is the limit when it's being built for you.
First of all, many thanks to both of you for the inputs.
However it seems like there is a fork here on the opinions. And again, like I have mentioned countless times that I'm a newbie... I have to admit I'm struggling to keep up with the depths of knowledge and wisdom both of you have on circuit designing.
Anyways, what I want to say is this: Please cut me some slack. As you can see, I'm dedicated to accomplishing my first project thus the revisions and questions. But please be reminded that I'm doing this out of a hobby not a career, I don't spent 16hrs a day on this. I read the comments, I try to digest the information and I then make challenges/questions to the unknown.
Now, treat me like a pure newbie with no electronics background and I'm humbled to learn... Let's focus on a decent enough design for now.
1st stage: Passive LPF
Is a first order passive lpf using 1.6k ohm and 1nf enough?
Is there a need to go to a 2nd order?
Does the components value here make sense?
2nd stage: Unity gain buffer
The NJM4556 seem to a problem now?
1. This should be used as the "power" amp component instead of the buffer component?
2. Did I make a mistake to have an unity gain buffer before power amp? I was guided somewhere else that the unity gain buffer increases the current output, allowing more current to flow into the next component (power amp stage).
3rd stage: Power gain
Is there an issue with op-amp rolling? I know every op-amp works differently and knock off the balance of the design equation. But this is part of DIY fun no? Or am I missing out on something I should know?
Other than the connections to the pot and the need to add a bias resistor in parallel, are there other issues?
Once again thanks for the inputs, do let me know if the comments by both of you should be considered separately as different design/project or I could use unilaterally into 1 design. It is getting more and more confusing...
However it seems like there is a fork here on the opinions. And again, like I have mentioned countless times that I'm a newbie... I have to admit I'm struggling to keep up with the depths of knowledge and wisdom both of you have on circuit designing.
Anyways, what I want to say is this: Please cut me some slack. As you can see, I'm dedicated to accomplishing my first project thus the revisions and questions. But please be reminded that I'm doing this out of a hobby not a career, I don't spent 16hrs a day on this. I read the comments, I try to digest the information and I then make challenges/questions to the unknown.
Now, treat me like a pure newbie with no electronics background and I'm humbled to learn... Let's focus on a decent enough design for now.
1st stage: Passive LPF
Is a first order passive lpf using 1.6k ohm and 1nf enough?
Is there a need to go to a 2nd order?
Does the components value here make sense?
2nd stage: Unity gain buffer
The NJM4556 seem to a problem now?
1. This should be used as the "power" amp component instead of the buffer component?
2. Did I make a mistake to have an unity gain buffer before power amp? I was guided somewhere else that the unity gain buffer increases the current output, allowing more current to flow into the next component (power amp stage).
3rd stage: Power gain
Is there an issue with op-amp rolling? I know every op-amp works differently and knock off the balance of the design equation. But this is part of DIY fun no? Or am I missing out on something I should know?
Other than the connections to the pot and the need to add a bias resistor in parallel, are there other issues?
Once again thanks for the inputs, do let me know if the comments by both of you should be considered separately as different design/project or I could use unilaterally into 1 design. It is getting more and more confusing...
Ok, I'm sorry. This is something that will happen when you ask for help on a forum. Many people with different depths and areas of experience, different styles and preferences, and lots of conflicting ideas, good advice and bad. The good, it provides you with a lot of different perspectives and fresh ideas. The bad, it can conflict, you can be lead astray, and posters can even go off topic arguing amongst themselves, leaving you scratching your head. Take it all with a grain of salt, pick out what is interesting and you can use, discard the rest.
You don't know the bios of the people helping you, so you have to take a lot on faith. So you know where I am coming from- I learned to solder when I was 8, and immediately began building Heathkit kits. I built many custom electronic hobby projects including guitar amplifiers for my musician friends. I spent several years working in a consumer electronics repair shop while getting my electrical engineering degree. I graduated with my BSEE in 1996, and my senior design graduation project was a fully discrete built, hand-wired on proto board stereo Class-D integrated amplifier. I worked for Hewlett Packard and Intel for 18 years as a senior design engineer on switching power supplies and VLSI processor circuitry design. I taught myself vacuum tube radio theory, at one time owned over 200 tube radios of every different sort, and beautifully restored dozens. I have fully restored several antique cars and do all of my own mechanical work. I burned out, retired early in 2016, and moved to the mountains to homestead and play hookie. Since then I have been rekindling my interest in discreet analog electronics and high quality audio. I am working on the build of my third tube amp, in addition to having restored multiple germanium and other vintage amplifiers, turn tables, and radios. I have built several of Rod Elliott's audio designs and power supplies, and I'm just getting started. I recognize a lot of myself in you, and want you to learn, but will not always justify in depth the advice I have given. I will try better to.
I was steering you pretty hard to the transistor output design for the learning opportunity. It will meet your original specs (IEMs down to 10 ohms) and is an excellent performer. Because I have built it, I can offer tons of direction and advice as to the building. But, it does not sound like a good fit for where you are right now, and what you want to use. Given that, let's proceed with cabirio's schematic in post #8 as a rough starting point.
If you are using the schematic that cabirio posted in post #8, which is basically what you should be following if you want to stick to op-amps, then yes, you got it backwards. The NE5532 stage is not unity gain, it has a low gain (4) and drives the volume control. The final stage, the 4556 is used in unity gain as a power output.
This first active stage with the NE5532 is your voltage amplifer- not unity gain. This is the stage that provides a small amount of variable gain (amplification), and is only lightly loaded, because it's job is to present the amplified signal to the volume control and input of the last stage, which is the buffer (power) stage. Because this NE5532 stage is lightly loaded, it can be almost any op-amp you desire. Here is where I would use an 8-pin dip socket (machine pin is my favorite), and an NE5532 or any other non-special cheap op-amp for initial testing, and you can roll this op-amp if you desire in the future for possibly audible different sound.
Again, in keeping with cabirio's schematic, the third stage, after the volume control, is the unity gain output buffer (driver). The NJM4556 op-amp was suggested by cabirio because it is a "higher output current" op-amp that would be somewhat more able to drive the heavy load of your headphones, so if you use this, it needs to be in the output stage.
I strongly suggest that if you are going to use this, that the op-amp is hopefully doubled up with two driving in parallel for twice the output current as I showed you in the Rod Elliott project for the portable headphone amp. Look at the last part of that schematic how U2 and U3 are wired together like a team of two horses. This is a safe way to couple these two amps together. The two 51 Ohm resistors allow the amplifiers to have minor differences in output voltage without fighting each other too much. You should be able to tack this at the end of cabirio's schematic in place of the single unity gain buffer.
Digest this, come up with another rough schematic, and we can hash it out. Or, I can hand sketch what I am talking about if you need a better start than the discussion.
You don't know the bios of the people helping you, so you have to take a lot on faith. So you know where I am coming from- I learned to solder when I was 8, and immediately began building Heathkit kits. I built many custom electronic hobby projects including guitar amplifiers for my musician friends. I spent several years working in a consumer electronics repair shop while getting my electrical engineering degree. I graduated with my BSEE in 1996, and my senior design graduation project was a fully discrete built, hand-wired on proto board stereo Class-D integrated amplifier. I worked for Hewlett Packard and Intel for 18 years as a senior design engineer on switching power supplies and VLSI processor circuitry design. I taught myself vacuum tube radio theory, at one time owned over 200 tube radios of every different sort, and beautifully restored dozens. I have fully restored several antique cars and do all of my own mechanical work. I burned out, retired early in 2016, and moved to the mountains to homestead and play hookie. Since then I have been rekindling my interest in discreet analog electronics and high quality audio. I am working on the build of my third tube amp, in addition to having restored multiple germanium and other vintage amplifiers, turn tables, and radios. I have built several of Rod Elliott's audio designs and power supplies, and I'm just getting started. I recognize a lot of myself in you, and want you to learn, but will not always justify in depth the advice I have given. I will try better to.
I was steering you pretty hard to the transistor output design for the learning opportunity. It will meet your original specs (IEMs down to 10 ohms) and is an excellent performer. Because I have built it, I can offer tons of direction and advice as to the building. But, it does not sound like a good fit for where you are right now, and what you want to use. Given that, let's proceed with cabirio's schematic in post #8 as a rough starting point.
Those values sound fine. When in doubt, I seek inspiration from professional designs. (hundreds of free schematics online) The input stage of my favorite vintage transistor amp, the Dynaco ST-80 uses 4.7K and 220pF, with an F3db of 154Khz. I indicated >> 100Khz, so 100Khz is a minimum, I'd go higher because you want this well above audible frequencies, and you are trying to block RF, which is in the 100's of Khz to megahertz. Don't go second order. First order is plenty. At the very beginning at the input, throw a 100K or so to ground, followed by a series 1uF film capacitor for a DC blocking coupling cap. (The 100K holds the input side of the coupling cap lightly to ground, preventing noise when the input cable is not plugged in, and elminating pop when you plug a cable in.) then the RC filter above.
If you are using the schematic that cabirio posted in post #8, which is basically what you should be following if you want to stick to op-amps, then yes, you got it backwards. The NE5532 stage is not unity gain, it has a low gain (4) and drives the volume control. The final stage, the 4556 is used in unity gain as a power output.
This first active stage with the NE5532 is your voltage amplifer- not unity gain. This is the stage that provides a small amount of variable gain (amplification), and is only lightly loaded, because it's job is to present the amplified signal to the volume control and input of the last stage, which is the buffer (power) stage. Because this NE5532 stage is lightly loaded, it can be almost any op-amp you desire. Here is where I would use an 8-pin dip socket (machine pin is my favorite), and an NE5532 or any other non-special cheap op-amp for initial testing, and you can roll this op-amp if you desire in the future for possibly audible different sound.
Again, in keeping with cabirio's schematic, the third stage, after the volume control, is the unity gain output buffer (driver). The NJM4556 op-amp was suggested by cabirio because it is a "higher output current" op-amp that would be somewhat more able to drive the heavy load of your headphones, so if you use this, it needs to be in the output stage.
I strongly suggest that if you are going to use this, that the op-amp is hopefully doubled up with two driving in parallel for twice the output current as I showed you in the Rod Elliott project for the portable headphone amp. Look at the last part of that schematic how U2 and U3 are wired together like a team of two horses. This is a safe way to couple these two amps together. The two 51 Ohm resistors allow the amplifiers to have minor differences in output voltage without fighting each other too much. You should be able to tack this at the end of cabirio's schematic in place of the single unity gain buffer.
Digest this, come up with another rough schematic, and we can hash it out. Or, I can hand sketch what I am talking about if you need a better start than the discussion.
Don't sweat it Deci1223, we're just approaching it differently. I agree that an amp with a discrete output like Rod Elliot's or mine (here) will have much better output drive capability. I disagree that the NJM4556 is a dog, especially for a first project like yours. I base this opinion on personal experience and actual measurements of the opamp that go well beyond its datasheet specs (here). I use my design above to drive a variety of headphones with very different impedances and sensitivities, but given that IEMs often have sensitivities in the 100-110 dB SPL/mW range and long term exposure to SPL's above 85 dB can cause permanent hearing damage, you really don't need a lot of power for this.
I also think that, for a first project, you're better off keeping it as simple as possible, making it more likely that you'll succeed both in building something that works and understanding how and why it works, which to me is as important. In the end it's up to you to decide how complex a project you're willing to tackle.
I also think that, for a first project, you're better off keeping it as simple as possible, making it more likely that you'll succeed both in building something that works and understanding how and why it works, which to me is as important. In the end it's up to you to decide how complex a project you're willing to tackle.
Sorry, too late to edit the last post: @Moderation Team , could you please move this thread to the Headphone Systems forum?