Hopefully this is the right place for this question. The problem I'm having is that when I use my IEMs with my portable device, I have to set the volume almost as low as it goes in software to get a comfortable listening level. I can hear really bad quantization noise in quiet parts of the music. It sounds like I'm listening to just a couple of bits in the quietest parts.
My headphones are Sennheiser IE6, and the impedance looks like a flat 17 ohms across the frequency response. http://www.innerfidelity.com/images/SennheiserIE6.pdf
I'll be getting a Hifiman RE-400, and it looks like those have 36 ohm impedance and just slightly less sensitivity. I will probably use one at home and the other at work.
My device is a Samsung Galaxy S4, and I assume the output impedance is low, but I have not attempted to measure it yet. It's not the best quality, as there is some noise and of course the quantization noise, but it's good enough and convenient to use at work. It can drive my IEMs to painful levels, and it can even drive something like the Sennheiser HD700 I tried in a store to as loud as I would ever need.
I have a similar problem when I listen on my old Macbook Pro. I have to set the volume super low, around 1 bar or even a fraction of a bar using option+shift+volume up/down key. If it's in 16 bit mode, which it defaults to, I can hear quantization noise at these low levels and background noise. If I force it to 24 bit mode, it's better, but it won't stay there.
My desktop PC has a better sound card (integrated Realtek ALC888S, Intel HD Audio). The noise floor is lower, it sounds a bit better, and I have more control over the output levels. But it's not very portable!
I think I need an attenuator to use with the first two sources, but I can't figure out how to design it properly. I'm an electrical engineer, but I don't have a very good understanding of audio yet.
The problem I'm running into is that the impedance of the source and headphones are kind of close together. As I mentioned, I haven't measured the source output impedance, but if I had to guess it's probably 0.5-2 ohms. The headphones are 17 ohms. I want the load on the device output to be high impedance so it doesn't load the output down too much. But I want to connect the headphones to a low impedance for good damping and driver control. I'm blanking on what configuration of resistors could do this. Would it work to use a higher impedance (like 20 ohm) resistor in series and a lower impedance (like 2 ohm) resistor as a shunt to form a voltage divider? The 20 ohm resistor would keep the load on the source low (22 ohms), and the 2 ohm would lower the output impedance the headphones see (2 ohms in parallel with [20 ohms plus the source output impedance]).
A fixed attenuator (pad?) is fine (preferable, actually) if I can design it right. But how do I know how much attenuation I need? Maybe I just have to measure the voltage going to the headphones at the highest comfortable listening volume and design a circuit that scales the full output to a comfortable level.
Will this setup work without compromising sound quality, or do I need an amp? It seems silly to me to get an amp to reduce the signal level. But if it's the best way to get the needed impedance at each end, I don't mind building one. The O2 looks easy enough, but it's a lot more expensive than a few resistors!
My headphones are Sennheiser IE6, and the impedance looks like a flat 17 ohms across the frequency response. http://www.innerfidelity.com/images/SennheiserIE6.pdf
I'll be getting a Hifiman RE-400, and it looks like those have 36 ohm impedance and just slightly less sensitivity. I will probably use one at home and the other at work.
My device is a Samsung Galaxy S4, and I assume the output impedance is low, but I have not attempted to measure it yet. It's not the best quality, as there is some noise and of course the quantization noise, but it's good enough and convenient to use at work. It can drive my IEMs to painful levels, and it can even drive something like the Sennheiser HD700 I tried in a store to as loud as I would ever need.
I have a similar problem when I listen on my old Macbook Pro. I have to set the volume super low, around 1 bar or even a fraction of a bar using option+shift+volume up/down key. If it's in 16 bit mode, which it defaults to, I can hear quantization noise at these low levels and background noise. If I force it to 24 bit mode, it's better, but it won't stay there.
My desktop PC has a better sound card (integrated Realtek ALC888S, Intel HD Audio). The noise floor is lower, it sounds a bit better, and I have more control over the output levels. But it's not very portable!
I think I need an attenuator to use with the first two sources, but I can't figure out how to design it properly. I'm an electrical engineer, but I don't have a very good understanding of audio yet.
The problem I'm running into is that the impedance of the source and headphones are kind of close together. As I mentioned, I haven't measured the source output impedance, but if I had to guess it's probably 0.5-2 ohms. The headphones are 17 ohms. I want the load on the device output to be high impedance so it doesn't load the output down too much. But I want to connect the headphones to a low impedance for good damping and driver control. I'm blanking on what configuration of resistors could do this. Would it work to use a higher impedance (like 20 ohm) resistor in series and a lower impedance (like 2 ohm) resistor as a shunt to form a voltage divider? The 20 ohm resistor would keep the load on the source low (22 ohms), and the 2 ohm would lower the output impedance the headphones see (2 ohms in parallel with [20 ohms plus the source output impedance]).
A fixed attenuator (pad?) is fine (preferable, actually) if I can design it right. But how do I know how much attenuation I need? Maybe I just have to measure the voltage going to the headphones at the highest comfortable listening volume and design a circuit that scales the full output to a comfortable level.
Will this setup work without compromising sound quality, or do I need an amp? It seems silly to me to get an amp to reduce the signal level. But if it's the best way to get the needed impedance at each end, I don't mind building one. The O2 looks easy enough, but it's a lot more expensive than a few resistors!
Welcome to diyAudio 🙂
I'm sure you'll get lots of good advice on this but FWIW here are my thoughts...
I tend to find that when looking at getting levels subjectively correct (that is to say that it all feels right and correct) that you can spend as long as you want sweating the details but ultimately you need to just try it and get a feel for how it will be.
I would suggest rigging the phones with a series resistor of say 22 ohms initially and then add shunt values across it to get the level you want. Also, some (not all) Class D output stages tend to need the presence of the low impedance load (the phones) as these form part of the output filter.
So no hard and fast rules. Grab some resistors and give it a try. You need that starting point 🙂
I'm sure you'll get lots of good advice on this but FWIW here are my thoughts...
I tend to find that when looking at getting levels subjectively correct (that is to say that it all feels right and correct) that you can spend as long as you want sweating the details but ultimately you need to just try it and get a feel for how it will be.
I would suggest rigging the phones with a series resistor of say 22 ohms initially and then add shunt values across it to get the level you want. Also, some (not all) Class D output stages tend to need the presence of the low impedance load (the phones) as these form part of the output filter.
So no hard and fast rules. Grab some resistors and give it a try. You need that starting point 🙂
Thanks. I know getting the levels right will probably take some experimentation. But I want to make sure there aren't any unintended effects on the signal that are subtle or hard to hear. For example, if I load the device output too much (too low impedance) it might start to distort or clip. Clipping would be easily audible, but a little distortion might be hard to detect except in a careful listening test. Or if I drive the headphones with too much impedance the frequency response might change, or the damping might not be as good, which could alter the sound a little.
Using resistors will always be a compromise, but it can be a pretty good compromise and work really well.
Don't be afraid of using empirical methods where there are unknowns. If your worried over distortion or its ability to drive a low impedance then try it for real. Rig your 'phones up with a series resistor (or attenuator using series shunt) and then also load the output before the attenuator. See what it will drive realistically before obvious distortion sets in and then raise the minimum value by say a factor of two or three.
I think your imagining problems where there probably won't be any in practice.
Don't be afraid of using empirical methods where there are unknowns. If your worried over distortion or its ability to drive a low impedance then try it for real. Rig your 'phones up with a series resistor (or attenuator using series shunt) and then also load the output before the attenuator. See what it will drive realistically before obvious distortion sets in and then raise the minimum value by say a factor of two or three.
I think your imagining problems where there probably won't be any in practice.
Generally speaking, an amplifier does make the best attenuator - it's very hard to obtain <<1 ohm of Zout otherwise. That being said, all of the in-ears mentioned seem to have virtually flat impedance responses and as such ought to be relatively immune to source impedance related coloration (the RE-600 mostly is around 36 and peaks at around 38 ohms, which computes to a mere 0.34 dB deviation at 100 ohms).
So I'd agree that a resistive attenuator ought to work fine here. Maybe 100 : 33 ohms for a start - I try to avoid loading outputs with <100 ohms if I can help it, headphone drivers tend to be a fair bit happier that way (crossover distortion in particular). Don't use tiny thick-film resistors, low power handling + high voltage and temperature coefficients = bad. TH metal-film or decent-sized SMD thin-film are much preferred.
So I'd agree that a resistive attenuator ought to work fine here. Maybe 100 : 33 ohms for a start - I try to avoid loading outputs with <100 ohms if I can help it, headphone drivers tend to be a fair bit happier that way (crossover distortion in particular). Don't use tiny thick-film resistors, low power handling + high voltage and temperature coefficients = bad. TH metal-film or decent-sized SMD thin-film are much preferred.
Your system suffers from poor gain structure.
I would be looking to hack its circuit and improve it; but honestly, modern SMD subminiature circuitry is hard to hack. (I personally hate it.)
Series resistors can affect the frequency response of some headphones (for better or for worse). The overall trend in headphones is towards high efficiency, low impedance phones that are designed to be driven by low impedance sources.
Your best bet for sound quality would be to rig a unity gain buffer with an attenuator in front of it.
First step is a series resistor or ladder, but you will waste headroom and I suspect that you don't have a lot to waste.
I would be looking to hack its circuit and improve it; but honestly, modern SMD subminiature circuitry is hard to hack. (I personally hate it.)
Series resistors can affect the frequency response of some headphones (for better or for worse). The overall trend in headphones is towards high efficiency, low impedance phones that are designed to be driven by low impedance sources.
Your best bet for sound quality would be to rig a unity gain buffer with an attenuator in front of it.
First step is a series resistor or ladder, but you will waste headroom and I suspect that you don't have a lot to waste.
This The HISSBUSTER - for sensitive headphones might be an option for you. If you don't want to build them yourself and don't mind waiting for a month or so, I plan on building a few, probably in March.
- Status
- Not open for further replies.