After looking at existing designs, I have decided I want to build a custom in ear monitor headphone amp. I am designing this amp to be used with the Shure SE215 phones, which are 20 ohms and 107 dB/mW.
I have a few questions:
1) I have a single 9 VDC source. Is this enough? If not, how much is enough? Will using a virtual ground have acceptable results? Would using a charge pump be of any benefit?
2) I am using the NJM4556. Will using one op amp per channel give significant SPL for live in ear monitoring? The live ambient sound may average at about 90 dB. The IEMs create a seal, so I believe that SPL does not need to be as high as normal headphones.
3) How do I determine the proper compromise between safety and performance when it comes to choosing the output resistor?
4) If using a virtual ground, and using a blocking capacitor on the output, what value of resistance do I use to calculate the corner frequency of the high pass filter?
Thanks for your help.
I have a few questions:
1) I have a single 9 VDC source. Is this enough? If not, how much is enough? Will using a virtual ground have acceptable results? Would using a charge pump be of any benefit?
2) I am using the NJM4556. Will using one op amp per channel give significant SPL for live in ear monitoring? The live ambient sound may average at about 90 dB. The IEMs create a seal, so I believe that SPL does not need to be as high as normal headphones.
3) How do I determine the proper compromise between safety and performance when it comes to choosing the output resistor?
4) If using a virtual ground, and using a blocking capacitor on the output, what value of resistance do I use to calculate the corner frequency of the high pass filter?
Thanks for your help.
Re: (1) -
Essentially a single 9V supply ought to be adequate - sensitivity works out to about 124 dB / V after all, so 1 Vrms (2.83 Vpp) would be quite sufficient to blow your hearing out. '4556A datasheet says it can go up to 7.5 V and down to 2.1 V for 40 mA source/sink. Non-rechargeable 9V blocks tend to have a short-circuit current of only ~200 mA though, so you can calculate what that means for internal resistance - it works out to 45 ohms or so. Serious power into 20 ohms x2 would have the voltage sag quite a bit, so I wouldn't skimp on buffer capacitance.
It may be a smarter idea to go for a split supply using 2x 2 AA (or maybe AAA) cell holders. These tend to have far smaller internal resistance than 9V blocks with their teeny tiny AAAA cells (and generally, rechargeables are better than alkalines), so voltage under load may not actually be that much lower, and you would be able to get by without the massive output coupling caps required otherwise (470 µF per channel at the very least) and with moderately-sized power filter caps (I'd say 2x 220µ/10V ought to do).
If you go for an AC-coupled feedback network, virtual ground would only be needed for + input bias, so it wouldn't have to be anything particularly beefy - 10k / 10k || 22µ with 47k to + input would be just fine.
A charge pump to generate a negative supply is the next best thing to a real split supply, but their maximum current tends to be limited. Note that higher total supply voltage for amplifier electronics = higher battery drain.
Re: (2) -
I am fairly certain that this would suffice. Broadband isolation for SE215s should be 20+ dB depending on tips.
Re: (3) -
You don't. '4556As require only minimal short-circuit protection, especially at low supplies, so 10 ohm combining resistors ought to be ample. The O2 gets by with 1 ohm, though there are indications that this already increases current draw by a few mA, so for optimum battery life I'd suggest 2.2-4.7 ohms. These values should also still yield acceptably low output impedance in case you ever decide to upgrade to a multi-driver BA IEM (SE215s are a bit of an odd one out, being a dynamic driver affair with the corresponding very flat impedance response).
As far as hearing safety goes, you should know what sort of input levels to expect. You may be able to get along with unity gain, possibly with an extra input attenuator.
Re: (4) -
For the output, the sum of output impedance (usually dominated by the two combining resistors in parallel) and load impedance.
For the feedback network of a non-inverting amplifier, it's the lower resistor, Rg.
And before going through all that trouble, I'd suggest trying out a Topping NX1 (<40 bucks US 1-chip wonder) and seeing how well that works out.
Essentially a single 9V supply ought to be adequate - sensitivity works out to about 124 dB / V after all, so 1 Vrms (2.83 Vpp) would be quite sufficient to blow your hearing out. '4556A datasheet says it can go up to 7.5 V and down to 2.1 V for 40 mA source/sink. Non-rechargeable 9V blocks tend to have a short-circuit current of only ~200 mA though, so you can calculate what that means for internal resistance - it works out to 45 ohms or so. Serious power into 20 ohms x2 would have the voltage sag quite a bit, so I wouldn't skimp on buffer capacitance.
It may be a smarter idea to go for a split supply using 2x 2 AA (or maybe AAA) cell holders. These tend to have far smaller internal resistance than 9V blocks with their teeny tiny AAAA cells (and generally, rechargeables are better than alkalines), so voltage under load may not actually be that much lower, and you would be able to get by without the massive output coupling caps required otherwise (470 µF per channel at the very least) and with moderately-sized power filter caps (I'd say 2x 220µ/10V ought to do).
If you go for an AC-coupled feedback network, virtual ground would only be needed for + input bias, so it wouldn't have to be anything particularly beefy - 10k / 10k || 22µ with 47k to + input would be just fine.
A charge pump to generate a negative supply is the next best thing to a real split supply, but their maximum current tends to be limited. Note that higher total supply voltage for amplifier electronics = higher battery drain.
Re: (2) -
I am fairly certain that this would suffice. Broadband isolation for SE215s should be 20+ dB depending on tips.
Re: (3) -
You don't. '4556As require only minimal short-circuit protection, especially at low supplies, so 10 ohm combining resistors ought to be ample. The O2 gets by with 1 ohm, though there are indications that this already increases current draw by a few mA, so for optimum battery life I'd suggest 2.2-4.7 ohms. These values should also still yield acceptably low output impedance in case you ever decide to upgrade to a multi-driver BA IEM (SE215s are a bit of an odd one out, being a dynamic driver affair with the corresponding very flat impedance response).
As far as hearing safety goes, you should know what sort of input levels to expect. You may be able to get along with unity gain, possibly with an extra input attenuator.
Re: (4) -
For the output, the sum of output impedance (usually dominated by the two combining resistors in parallel) and load impedance.
For the feedback network of a non-inverting amplifier, it's the lower resistor, Rg.
And before going through all that trouble, I'd suggest trying out a Topping NX1 (<40 bucks US 1-chip wonder) and seeing how well that works out.
Thank you very much for your input, it was very helpful.
I was thinking along the same lines as your explanation, so this is reassuring for me. The Topping NX1 looks like a great solution. I need to continue designing my own amp because my stereo input will actually come from two balanced mono signals (mixing board). I might as well put everything in one box to keep things simple and at about the same cost.
I have a preliminary design that follows your suggestions. I followed everything except for understanding the corner frequency calculation. You mentioned it is the combining resistors in parallel...I'm not sure what you mean by combining resistors, and why there would be two. In my schematic, are you referring to the 10R output resistor?
If I understand correctly,
Freq = 1 / sqrt( 2 * pi * R * C)
R = R_opamp + R_out + R_driver = 1 + 10 + 20 = 31 ohms
C = 470E-6 Farads
Freq = 11 Hz
I was thinking along the same lines as your explanation, so this is reassuring for me. The Topping NX1 looks like a great solution. I need to continue designing my own amp because my stereo input will actually come from two balanced mono signals (mixing board). I might as well put everything in one box to keep things simple and at about the same cost.
I have a preliminary design that follows your suggestions. I followed everything except for understanding the corner frequency calculation. You mentioned it is the combining resistors in parallel...I'm not sure what you mean by combining resistors, and why there would be two. In my schematic, are you referring to the 10R output resistor?
If I understand correctly,
Freq = 1 / sqrt( 2 * pi * R * C)
R = R_opamp + R_out + R_driver = 1 + 10 + 20 = 31 ohms
C = 470E-6 Farads
Freq = 11 Hz
Attachments
- Status
- Not open for further replies.