|
Post by marveltone on Aug 14, 2015 22:34:12 GMT
Interspersed amongst these threads, I keep running into references to filters. My poor befuddled mind is conjuring up images of purpose built sound processors, or miniature, preset headphone EQs, as it were. Am I on track? Do such beasts exist? If so, how does one acquire one?
I'm quite familiar with graphic EQs, as well as their parametric cousins, but this unknown filter talk is bugging me.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 15, 2015 5:52:57 GMT
You're not going mad. There is a filter in the pipeline, although it'll only be for certain headphones like the HD 650 which has a basic frequency response suitable for filtering in the first place. It'll be active & placed in-line between the source & amp. When it's actually going to happen I'm not sure but the powers that be are working on it.
Someone else who knows more about it will likely be along to explain further.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,882
|
Post by solderdude on Aug 15, 2015 7:53:21 GMT
There is lots 'o filters for various purposes.
filters can remove or increase or change a frequency response, phase response or other aspects.
Filters for musicians: Little boxes that can add effects to a single instrument so the sound changes. They can/should not be used for all instruments nor for hifi. Then there are tone controls that also can alter the sound.
Filters for power supplies:
Mains filters in power supplies can lower emitted or induced frequencies other than the ones that are supposed to be on the mains.
Emitted: All SMPS power supplies generate LOTS of high frequency switching noise signals which can be between 10kHz and 1GHz (10,000Hz to 10,000,000,000 Hz) and you don't want them to go onto the mains. Once on the mains they could cause some (sensitive) equipment to go bonkers in one way or another. There are rules to which manufacturers have to obide (but some simply don't, yet say they do) which determine how much 'polution' can be injected into mains. The rules differ for household and industrial, railroad, medical and automotive to name but a few. Manufacturers ONLY tend to filter the garbage just enough to pass tests as filter parts are expensive and making a minimum parts count filter is cheaper.
Induced: Then there are input filters that are designed to keep 'nasties' on the mains from entering (sensitive) equipment. Their purpose is to filter out incoming 'nasties' above say... 10kHz or at specific frequency bands to avoid influence in the circuits behind it. In this case too often a minimal amount of filtering is just for the same financial reasons to pass immunity tests. Just like in emitted screening can be a big part of the filtering by the way.
DC filters: You can also filter irregularities in power supplies after the voltage has been lowered and rectified. Regulators filter out ripples (up to certain frequencies). LC filters or LCR filters or simple RC filters (L stands for inductors, C for capacitors, R for Resistors) all have their positive and negative effects. 2 simple capacitors close to a single component in the power supply line can also provide enough filtering for a component or circuit to work 'properly' so tests can be passed or performance be guaranteed. Espicially SMPS (Switch-Mode-Power-Supplies) often have lots of LC filtering on their outputs (as well as their inputs)
In all cases above you can add extra mains or DC power line filtering which may or may not improve things measurably and or (tecnically inexplicable) subjectively.
Filters in speakers: Filters in speakers can exist to cut-off frequency bands a speaker(driver) cannot or may should not or the dsigner does not want to reproduce. Here too LC(R) filters are used. Filters in speakers can also be used to obtain a more even load impedance for amplifiers driving it. Filters can also be constructed to filter out narrow or wider bands per driver to make them flatter (similar to a fixed EQ)
Filters in digital: These exist in many forms and for many reasons but most are used for relaxing analog filters or needed for up-downsampling reasons. There is IIR and FIR, apodizing, pre- and or post-ringing and is always used for either up- or down-sampling (where one changes the bitrate of a signal). Too many aspects to cover here but in essence all of these filters are so called 'low pass' and the low part which is passed is usually 0Hz to 20khz miniumum. That what is filtered is above the (considered to be) audible limits, i.e. above 20kHz or even up to 100kHz for higher bitrates. In general all these filters do not filter in the audible range nor do they pre- or post-ring INSIDE the audible range. Some filters DO affect the audible band and even start to filter as low as 10kHz. This is really audible and usually are 'slow' or 'gentle' roll-off filters that are said to be more 'musical'. Both frequency and phase can be affected as well as time by these filters and they do so in different ways. Some filters allow high amounts of 'aliasing' to enter the audible band (though a process called mirroring against the sample frequency) which COULD but does not always have to become audible or could (but does not always) interfere with electronics and create unwanted signals. The audibility is heavily debated everywhere. More nonsense is spread (even by 'knowledgable' people than not and it is very hard for laymen to 'filter' out the wisdom form the nonsense and marketting talk.
Filters in analog: After each DAC chip some low-pass filtering is required to remove the 'steps' and smooth out the signal so the measured numbers are impressive and to prevent 'mirrors' or problems in gear behind it. Filters are usually 'active' meaning amplifiers (opamps usually or discrete) are used together with RC filters (seldom L's are used) to create low pass filters. These too can have different 'slopes' or phase/amplitude characteristics based on several aspects such as minimum sample rates or maximum samplerates but are usually based on the 'working frequency' of the DAC chip when internal upsampling is used (think most DS (Delta-Sigma) and dereivative DAC chips. Audibile effects of sharp filters (needed for Non Over Sampling ladder (R2R or otherwise) DACs are heavily debated and some even leave them completely out claiming they do more harm then good. In that case, the propreties of the gear following that DAC will proved the filtering and this may or may not work out for the better.
On inputs of most amplifiers there is an input filter. Some amplifiers have high-pass (capacitor coupled) filters that filter DC or frequencies below 1 to 20Hz for technical or 'religious' reasons. Most amplifiers (certainly not all) have low pass filters on their inputs preventing signals above the slew-rate (ability of the amplifier to follow the input signal speed changes) to enter the amp and thereby avoiding or lowering TIM (transient InterModulation) or 'A.M. detection' (think the buzzing sounds you hear sometimes just before your handy tells you someone is calling you) and/or amplification of unwanted frequencies from (ill designed) DAC's.
Filters in radio: In radios there are lots of filters that tune to frequencies you want to receive and convert them all to a single frequency (for FM 10.7MHz, for AM 455kHz) so one only needs to amplify and band filter one frequency. Then there are low-pass filters depending on the bandwidth of the signal. For AM these can be very low and filter well within the audible band. For FM these filters (when present) are usually till 15kHz as above that there is no 'information' present anyway. Then there is the MPX filter which 'notches' (very sharp filter that removes only a very narrow band) at 19kHz. The reason for that is that some young people can actually hear that + it may interfere with recording equipment. That 19kHz continuous 'pilot-tone' that is sent along with the music is there to tell a 'demultiplexer' when the left and when the right signal is transmitted. For mono radios L and R are thus automatically added. This 'demultiplexing' is done at double the frequency of 19khz , i.e. 38khz and stereo decoders (de-multiplexers) thus often also have a 38kHz notch filter as well to remove switching 'steps'.
Filters in Vinyl: As it is impossible to record an audio signal directly in a groove as is a tilting (and gradual sloping) filter is used to lower the amplitude of low frequencies and to increase the higher frequencies in amplitude. In a music signal the lower frequencies usually have the highest amplitude and highest frequencies have to lowest amplitude. This means that when untreated the needle would swing violently on bass frequencies and almost not on higher frequencies. Also the 'grain' of the vinyl would drown out the higher frequencies with their small amplitudes.
So... RIAA was born which is a 2 step RC filter. 1 section lowers the frequencies below 1kHz and does so by decreasing the amplitude of the lowest frequencies the most. The 2nd section increases the frequencies above 1kHz and in a way that the hisghest frequnecies are amplified the most. This frequency response altered signal is then written into the disc and has to be 'undone' on the exact same way again to let the original signal (amplitude/frequency wise) re-emerge. For this we need either a mechanical electrical converter that is 'speed dependant' such as the old 'Chrystal or Piezo' head which outputs the original signal without filtering as the 'filter' is embedded in the piezo principal (and is the background on which the RIAA is founded by the way) This also provides almost line levels without any amplification when connected to an input R of 1MOhm or higher) OR with a magneto dynamic cartridge (be them if the cantilever has a coil or magnet attached) followed by an amplifier, which is needed as the signal level is very small. That amplifier must also undo the frequency alterations and so the phono preamp also needs to have the RIAA characteristic. As a bonus of this RIAA system you also obtain a great reduction in high frequency noise (which is more audbile than lF noise) because higher frequencies (including the vinyl noise) are lowered considerably by the RIAA filter.
Then there is the input IMPEDANCE and CAPACITANCE of that RIAA amplifier that also alters the frequency response and differs per 'cartridge' A wrong impedance or capacitance can cause peaking at high frequencies, roll-off or even detecting of a radiostation or other unwanted signals. Rarely the cartridge and input characteristics match perfectly so different SQ is lurking because of this 'RC input filter'.
Then there is the accuracy of the RIAA filter as well which are not all created equal.
On top of that there is the rumble filter which is a high pass filter that lowers the amplitude of frequencies below 40Hz (filters can range from 10Hz to 60Hz in general) They prevent the woofer from making large excursions because of warped discs which could damage the woofers. As there (generally) is little relevant info below 10Hz the 'rumble' can be lowered in amplitude.
Filters in tape recorders:
In Tape recorders filters are also needed for various reaons. Just like with digital audio the bandwith is limited because of the bias frequency used for recording. Bias is needed to 'move' teh magnetic particles on the tape. These particles must remain in there 'position' after the recording in order to 'hold' the sound. For this reason you need quite some magnetic 'force' to get them to be positioned by an audio signal. This bias can be applied by using a DC current (cheap cassette drecorders) but these magenetise the heads which is unwanted. So a high frequency tone (the bias) is applied and the audio signal rides on top. Once the magnetic particle is 'free' the audio signal can align it. That bias amplitude must not be too high nor too low. Too low results in 'distortion' and too high means the smallest amplitudes in the audio (the treble) cannot align the particles as the bias force is overpowering it. For this reason some tapes worked 'better' on your cassette player than others, because it needed a slightly lower or higher bias. Bias levels could be adjusted internally by the factory or service guy to a tape of your perefrence. The better decks had these bias adjustments on the front panel. FeCr, CrO2 and metal all needed substantially different amounts of magnetic 'force'compared to Ferro tape. When Fe = 100% FeCr = 110%, CrO2 = 150% and metal 200% (ballpark numbers).
The bias frequency is usually between 80kHz and 100kHz (for commercial equipment) and the input signal must not have audio information of those frequencies. The input signal is thus filtered away above 30-40kHz. Then there is also the MPX filter which is needed so Dolby systems aren't 'fooled' by the stereo MPX signal of some tuners (that do not have a filter inside) and so the 19kHz or 38kHz signals can not interact with the bias frequency and cause aliassing. The 'bias signal' also is recorded alongside the audio. For this reason on playback the bias frequency is also filtered out again using a notch filter. This is to create nice 'numbers' and not too fool Dolby or others decompressors on playback.
Then there is also some equalizing needed somewhat like RIAA but NOT the exact same.
Equalizers.
These exist in various forms of which the graphic one (with sliders for different bands) and bandwidths depending on the numbert of sliders (bands) are most common. These are all analog and often introduce (some) noise. They are fine for 'correcting' the frequency response of speakers/headphones in a general way or to 'correct' for (some) room issues. They can also be used to change the frequency response more to your liking (to 'colour' the sound) Most common in commercial equipment (with upto 10 sliders mostly). Some of the more expensive ones had 20 to 30 sliders. Some have separate sliders for left and right (for correcting speakers in rooms) and others have 1 slider for left + right which are more useful as a 'tone'control'
Then there is the parametric one where each 'section' can be changed in frequency (usually with a band only), in steepness (Q) and amplitude. These are ideal to filter out nasty (room) resonances exactly. More common in studios or professional equipment and requires almost expert knowledge and measuring equipment to set it up.
Both types of equalizers (there are other types too) are also in completely digital form these days. Even the cheapest MP3 players often have these.
The SQ stands or falls with the 'software' implementations of the filters, just like the analog's SQ stands or falls with the components used .
The 'headphone' filters I make are all analog and a crossing between gentle RC filtering and fixed 'parametric' filters that do closely the 'opposite' of the frequency response of the headphone. Thereby 'correcting' that what the headphone itself does not do properly. It cannot correct al'problems' nor can it make a headphone 'do' what is physically impossible for it. One can merely alter that waht can be altered and bring a headpone closer to 'flat' performance or give it a different sound 'coulour' if so desired. The headphone itself MUST be capable of reproducing everything and MUST be of high quality to begin with to achieve excellence. Lesser headphones can be improved 'somewhat'.
If you want to know more about other or specific type of filters... please ask in this thread!
Will make the info available on the website as well once more questions have been asked and answerred.
|
|
|
Post by marveltone on Aug 15, 2015 12:53:22 GMT
Thanks for the detailed answer. My reason for asking is simple enough. My preferred (and only, at the moment) set of cans are the well known, yet, certainly flawed Grado SR80i. I was looking into 31 band equalizers, which I have experience and good luck with, to help settle down the rather bumpy frequency response these cans are known for. Then I stumbled across this: www.minidsp.com/products/minidspkits/2-x-in-4-x-out. With my rather feeble way of reasoning, it seems that I could hook this up to my laptop, set my eq, then leave It in my soundpath to do its thing, with only a tiny footprint on the table, rather than a full rack sized box. The small size is what caught my attention. Assuming I have some experiance setting equalizers, will this do the trick, or am I better off with a standard graphic eq?
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 15, 2015 14:45:48 GMT
Thanks for the detailed answer. My reason for asking is simple enough. My preferred (and only, at the moment) set of cans are the well known, yet, certainly flawed Grado SR80i. I was looking into 31 band equalizers, which I have experience and good luck with, to help settle down the rather bumpy frequency response these cans are known for. Then I stumbled across this: www.minidsp.com/products/minidspkits/2-x-in-4-x-out. With my rather feeble way of reasoning, it seems that I could hook this up to my laptop, set my eq, then leave It in my soundpath to do its thing, with only a tiny footprint on the table, rather than a full rack sized box. The small size is what caught my attention. Assuming I have some experiance setting equalizers, will this do the trick, or am I better off with a standard graphic eq? What playback software are you using?
|
|
solderdude
Administrator
measureutternutter
Posts: 4,882
|
Post by solderdude on Aug 15, 2015 15:28:16 GMT
There are plenty of good plugins for players like foobar which can also do the trick. This is more of a 'stand alone' device when want to replace an equalizer or want to do more things than just EQ. However, if you are a hires junkie (have sample rates >48kHz) using this device is rather pointless.
It basically is a sound card with some signla processing inside working on 48kHz. Should you want to use it as a DAC then you should realise all redbook will be upsampled to 48kHz.
If it just for 1 or more headphones and want to EQ and already have a desktop amp and DAC the (still upcoming) filter board may be of use. In that case you don't have to figure out and set the EQ but rather use a module.
Then there are programs like accudio which can 'load' certain headphones and bring you closer to perfect sound.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 15, 2015 15:58:51 GMT
The reason I asked about Joe's playback software was so that we could better recommend an equalizer plug-in. Parametric EQ's are the best type although they're not all that easy to set up if the user is not used to them. I scoured the net for a frequency response graph for the SR80i but I could only find the SR80 & I'm not sure how close they are. By looking at it though & guessing that the bass response is somewhat similar I hope that's not the area Joe's looking to enhance because that won't be easy.
|
|
|
Post by marveltone on Aug 15, 2015 16:35:23 GMT
The reason I asked about Joe's playback software was so that we could better recommend an equalizer plug-in. Parametric EQ's are the best type although they're not all that easy to set up if the user is not used to them. I scoured the net for a frequency response graph for the SR80i but I could only find the SR80 & I'm not sure how close they are. By looking at it though & guessing that the bass response is somewhat similar I hope that's not the area Joe's looking to enhance because that won't be easy. Playback software at this point is negotiable. For the time being, I've just defaulted to Windows Media Player for ease of use (It was there!) but I am not married to it, and am certainly welcome to better suggestions. As far as frequency correction goes the bass response is not a problem. It may not be as strong as in other cans, but it's tight, and definitely there. I'm mostly hoping to smooth out some of the peaks and irregularities in the upper end, where things tend to get ragged, such as the resonance around the 4k mark, for example. Up to this point, I've been pretty old school in my equipment, and have just recently been moving from CDs to using the laptop for my listening. So, while I'm no stranger to the older tech home sound equipment, the newer digital stuff has flown by me rather quickly, and I'm playing catch-up. I joined this site to learn, catch up, and am literally open to all suggestions. Thus far, I have no DAC, no amp, and no corrective hardware or software. I'me basically building from the ground up, given my current laptop and current phones. I'm fairly certain (but not written in stone) that I want to buy a Schiit Modi 2 for my DAC, and a Project Starlight for my amp. This is where I'm at a stand still. I really enjoy my Grados, but I know of their problems and limitations, so I'm looking to give them a little help, if possible without breaking the bank.
|
|
|
Post by marveltone on Aug 15, 2015 16:46:55 GMT
There are plenty of good plugins for players like foobar which can also do the trick. This is more of a 'stand alone' device when want to replace an equalizer or want to do more things than just EQ. However, if you are a hires junkie (have sample rates >48kHz) using this device is rather pointless. It basically is a sound card with some signla processing inside working on 48kHz. Should you want to use it as a DAC then you should realise all redbook will be upsampled to 48kHz. If it just for 1 or more headphones and want to EQ and already have a desktop amp and DAC the (still upcoming) filter board may be of use. In that case you don't have to figure out and set the EQ but rather use a module. Then there are programs like accudio which can 'load' certain headphones and bring you closer to perfect sound. So your recommendation would be a software solution, rather than hardware? I'm so used to my old days of running directly out of the headphone jack of my old CD player that I totally spaced out the software solution. I've only recently started moving my music library to the laptop and still have a hard time seeing it as a viable music source. Alas, the digital music player is still a new world to me,
|
|
|
Post by marveltone on Aug 15, 2015 17:43:40 GMT
BTW, just downloaded Foobar200 today and am starting to familiarize myself with it. Told you I was behind!
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 15, 2015 18:02:15 GMT
Foobar2000 is definitely the way to go on a Windows machine. There are other free options & many that cost money but Foobar2000 served me well for 13 years before I switched to a Mac. I switched for reasons other than music playback, there's no real difference (that's audible) between the platforms in that respect.
I'm glad you're happy with the bass as is because it's much easier to remove peaks than it is to enhance troughs. If you're going to be buying a valve amp there may be no need to alter the signal at all because most valves tend to smooth out the treble anyway. Still, if you want to experiment I'd recommend installing a parametric EQ plug-in rather than relying on the stock Foobar2000 graphic EQ. Not that it's a bad one of its type, it's actually pretty good.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,882
|
Post by solderdude on Aug 15, 2015 19:42:38 GMT
This could be worth a read when playing with foobar.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 15, 2015 19:56:11 GMT
This could be worth a read when playing with foobar. +1 I hadn't seen that before Frans. Nice work.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,882
|
Post by solderdude on Aug 15, 2015 19:57:46 GMT
It's written by Javier
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Aug 15, 2015 20:03:38 GMT
Well done Javier then. It's pretty much perfect.
|
|