|
Post by PinkFloyd on Dec 27, 2004 12:15:50 GMT
I've really been testing the LM6171 without output caps extensively over the past couple of days whilst the rest of you have probably been stuffing yourself with turkey I tried 6 x LM6171's and no two gave measurements that suggested simply whacking a resistor between 3 & 4 is the ideal solution. With 10M all of the results were "-mV" and here are the results: Chip 1: -48.1mV Chip 2: -41.3mV Chip 3: -23.5mV Chip 4: -13.7mV Chip 5: -37.2mV Chip 6: -25.8mV With 15M all of the results were "+mV" and here are the results: Chip 1: +15.9mV Chip 2: +24.7mV Chip 3: +34.9mV Chip 4: +49.0mV Chip 5: +27.8mV Chip 6: +31.2mV I made up quite a few 15M resistors using 3 x 10M resistors suitably configured (series / parallel) to make 15M: Pretty small chips to work with but easy when you get the hang of it. I then tested each one and selected 6 which where pretty spot on to 15M and then encased these 6 "modules" in hot glue to ensure there was no way of them coming apart: I attached these "modules" to the 6 x LM6171's and identified them by heatshrinking different colours onto their wires. I used the left channel of the WNA to perform the tests and then the right channel...... both channels measured pretty identical using the same LM6171 in each channel. My conclusions are that neither 10M or 15M can be considered "ideal" as all of the chips have to be individually "tuned" to give very low offset.. simply sticking a 15M between 3&4 is far from satisfactory. If you are going to remove the output caps I strongly suggest you have a few different resistors to hand as you'll have to "make" the correct ohmage resistor to suit each individual LM6171..... it could be anywhere from 10M to 16M dependant on the individual LM6171 you have... from the above results 12M - 13M may be a good starting point but be prepared to spend a very long time finding the value that will give you zero offset. I listened to a WNA with the two lowest offset measured LM6171's fitted (-13.7mV & 25.8mV) against a WNA with output caps fitted: I won't waste time typing here as I could hear "no" difference between the two.... NO difference at all. So, is it worth removing the output caps? IMO no it isn't. With the output caps in place there is ZERO offset and, as such, the full dynamic range will not be affected (any dc offset will pull or push the diaphragm away from its resting point which will reduce it's travel in one direction whilst uselessly increasing it in the other direction) You also risk blowing the crap out of your headphones if one of those resistors falls off Nope, I can fully understand why Dr. White has chosen to use output caps.... zero dc offset, guaranteed protection for your headphones and all this with minimum (if any) insertion loss..... Just my two cents but output caps it is for me considering the variations between LM6171's.... if a set value resistor gave an exact offset reading with "every" LM6171 then fine but due to the wide fluctuations output caps are the bullet proof solution.
|
|
gsphukd
<100
The Sultan Of Swing.
Veda-Audio
Posts: 46
|
Post by gsphukd on Feb 5, 2005 3:35:07 GMT
If the output caps are in the signal path - doesnt this filter out some of the low frequencies?
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Feb 9, 2005 0:58:21 GMT
"If the output caps are in the signal path - doesnt this filter out some of the low frequencies? "
Not if the input of the next stage has a high enough input impedance and/or the caps are large enough in value.
i happen to be on the side of not fearing cpacitors in the signal path and if done corrctly work fine audibly.I also am in the camp that says to do "DC coupled or bust" but add so much complexity to just eliminate a simple capacitor is like walking backwards.
|
|
gsphukd
<100
The Sultan Of Swing.
Veda-Audio
Posts: 46
|
Post by gsphukd on Feb 9, 2005 1:13:19 GMT
"If the output caps are in the signal path - doesnt this filter out some of the low frequencies? " Not if the input of the next stage has a high enough input impedance and/or the caps are large enough in value. Hiya Rick!!! Can you explain how the impedance of the next stage nullifies the filtering effect of the cap? Thanks bud!
|
|
|
Post by PinkFloyd on Feb 16, 2005 19:14:48 GMT
The next stage after the output caps are the headphones GS, quite literally, the caps are in series with the output and go between L / R output and the headphone socket. Here's the deal: With 300 ohm cans you should be using about 47uF Headphones with an impedance above 400 ohms use 22uF Headphones with an impedance below 400 ohms use 100uF Headphones with an impedance of 8 ohms use 1000uF I personally use 470uF output caps as they cover most bases (32R - 600R).... if you want the amp to be 8ohm to 600 ohm compatible then sling in a 1000uF non polar. I now use the Nitai non polar caps, being non polar they introduce no distortion and they are pretty cheap (sound great too!) The 16V 470uF are 22 pence each so they won't break the bank www.rapidelectronics.co.uk/rkmain.asp?PAGEID=80010&CTL_CAT_CODE=31198&STK_PROD_CODE=M62536&XPAGENO=1
|
|
FritzS
Been here a while!
Sound of Blue Danube
Sound of Blue Danube
Posts: 1,364
|
Post by FritzS on Mar 6, 2005 11:56:35 GMT
Have anyone heared a difference with/without caps? Is the quality high from Nitai non polar caps? I confused - the price are low - the price from high quality ELNA, Rubicon, etc. caps are expensive! Have you a polypropylene cap in parallel? I think a cap should not have more distortion as a transformer - looks like a Lundal www.lundahl.se/It's a good option to put a resistor (suggestion 100 ohm) in parallel to the lamp?
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Apr 24, 2006 15:53:23 GMT
Don't know the Nitai caps but have a 1000uf Elna Cerfine sitting on the output of my single ended class-a mosfet follower (based on the Headwize Szekeres Amp) and it sounds fine.In fact better than fine.I am also in the "single series capacitor" camp. I have no problem with multiple bypasses for SHUNT connected capacitors such as those used in a power supply bypass scheme but SERIES connected (that is where the signal must go through the capacitor) it is my opinion you screw up the timing integrity of the audio signal when as it passes through the different values of capacitor in the "multicap" byapss the final result at the other end is the signal being reassembled as a disjointed whole due to the speed differences of the caps. Think speaker crossover where each of the speaker elements (woofer,midrange,tweeter) have an individual frequency selective crossover which passes the signal on to its own driver. Being designed for a different section of the musical spectrum the one that needs to push large quantities of air to get the bass is big heavy and slow. Next will be the midrange which is smaller than the woofer because it need no reproduce extreme low bass or move large quantities of air but still large and slow in comparison to the tweeter.It has no shot at really high frequencies and even if it CAN reproduce them usually it will be at a very narrow window so dispersion will suck. Next up is the tweeter.Because it must reproduce the highest notes it must be small fast and light.try to make it play notes too low and it will detonate but used in the range it was designed for all is well. There is also the full range driver which can sound scary real with the right set up and the right music but being in the 4-8 inch diameter size range is really not a "full range" transducer but in reality a "Super Midrange" transducer.Used with a simple amp (single ended triodes or single ended Class-A transistor amp) and because there is no crossover in the frequence range where 90 % of music acxtuallyu lives the end sound is very coherent,very natural,and with simple vocals or aciustic instruments is as close to "real' as you are likely to get in the home. Reality of modern systems means this SUPER MID is usually mated to a subwoofer to regain the bass and a supertweeter to get the top end and that is when the crossover again becomes part of the system though being very low/high and in separate "boxes" the phasing integrity can be maintained. Headphones are like the full range driver.One single transducer to cover everything of musical significance so why would anyone want to much up the time.phase/space relationship of this by tossing in a "crossover" which is what the multiple capacitor really is ? Signal in/single cap/signal out or Signal in/path one capacitor out/path two capacitor out/mixed together as one again.Worse is the overlap area where before a cap hands off to the next they are both reproducing the same exacxt signal but at two different speeds and with to different "tones".Not good from my perspective K.I.S.S (keep it simple stupid) has even more meaning at the headphone level than it does at the speaker level becaus ethe interface is transducer/ears instead of tranducer(s)/air/room boundries/ ears.the room itself changes the timing and phasing unless you listen "nearfield" but with headphones there is no place for any inconsistancies to hide. simple and clean path to a simple interface.Signal in/signal out just personal opinion and YMMV (your mileage may vary)
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Apr 24, 2006 16:03:08 GMT
oh yeah,the output resistor.Serves one or two purposes 1-output current limiting to protect the output stage from damage in case of a short (like plugging in headphones while the amp is on) 2-adds series resistance which in combination with the capacitance of the cable and the next stage forms a low pass filter which protects feedback loop amplifiers from RFI "reflections" entering through the output jack,being fed back to the input via the feedback loop where this RFI will be treated as a signal and amplified (unlike RFI entering the input side which should be cancelled out by the differential input nature of a feedback amp). For "1" the resistor can be inside the feedback loop where it will have practical zero impact on the output of the amp. For "2" the resistor must be outside the loop but will have an impact on the sonics due to A-directly in the signal path B-current limits the output and raises overall output impedance making amps with marginal output current iffy with low Z cans. The cure of course is an overbuilt output stage combined with the standard (at least it was once ) 120 ohm series resistance which will provide both current limit/short circuit protection and protection from cable reflections entering the feedback path rock solid and glitch free in a real world system over theoretical perfection that would make for a potentially unstable or self destructive amp
|
|
|
Post by PinkFloyd on Apr 24, 2006 16:54:51 GMT
On this one I've got to agree "and" disagree with you Rick I've been experimenting with bypass caps for quite some time and have noticed in "some" amps they sound good in the signal path and in others you're better just leaving the 'lytic well alone to fend for itself. No measurement or theory will dictate how any one component will "sound" and the final tuning is always best done by trial and error using the el cheapo and most effective "ear" Output caps add a certain signature to the sound and this signature can be tweaked to suit you own taste and one good way of tweaking is to parallel some film caps onto the 'lytic until the sound suits your own ears. Sure, it's not perfect, ideally we wouldn't have output caps at all but sometimes they're required to block any DC that may be present on the output. Nothing at all wrong in experimenting with bypass caps in the signal path IMO. They either sound shite or they sound great and you never get the same results with different amps. This is a "highly" pleasurable pastime and one that takes great patience (and lots of experimentation with different values / types of caps) before you find the combo that suits "your" own personal set of ears. So, I agree with you that in some amps they're not a good idea but I certainly don't agree that they shouldn't be implemented in the signal path / period. All depends what the sound is like and, at this stage ot tuning, the ears are the only tools that I trust
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Apr 24, 2006 17:31:41 GMT
I am a "use the aboslute best single part you can" type of guy over a "lets put a bandaid on it" type.The bypass cap "fix" was invented to hide shortcommings on inferior parts where (usually) it was a big slow electrolytic "log jam" losing upper octave detail so it came about that using a smaller bypass to "speed up" the treble which is EXACTLY the prolem with the theory and in fact to my ears the use. Speed up one part of the signal then mix it back to a "slower version' of iteself is a good thing ? Why not just use a "treble boost" which is what in effect this is accomplishing ? Oh yeah.tome controls not only suck but muck up the time/phacse relatioinship so should be avoided while OOOPS ! That is what the cap is doing !!!!!!!!! Think funnels Mikester.You have a bucket of water and you want to get the water from bucket A to bucket B and being in a hurry you are using multiple funnels but they are not the same size You pur water into all at the same time yet the one with the big opening is already empty and needs to be re-filled while the one with the small opening is still half full. This is EXACTLY what capacitors of large size differences are doing but in a frequency selective way that if there was a quick cutoof and zero overlap may be valid even though the "faster" funnel will empty out quicker.This is what causes the mage in that area to wander and to become indistinct so is an attemot to get the "tone" right while butchering the time relationship maing the last part of the audio events arrive at different times even though the signal is identical. Listeng tests are usually the best test but there is a psychologicval event that always tells us different is better even when it is not and that is because thereIS a discernable difference rather than a refinement. Is overall tonality important ? Personally I am not convinced it is and the reason is every single system i have ever heard in my life sounds different yet the perfornmer is easily recognised for who he or she is.Take a Cd and play it on system #1 then pop the CD out and play it on the car system.Does it sound the same as the home rig ? It would be extreremely rare that it would yet you will have no problem at all,no question in your mind who the performer was even though the overall "tone" has changed. Take that same CD to a buddies house and what happens ? Tonally different again and so on in as many various systems you play that single CD in so which one is the CORRECT tone ? Obviously short of duplicating the mastering system NONE are correct yet all can not only be fully enjoyable but may actually be considered "dead nuetral" which when I read that is laughable because such a thing does not exist. ALL electronics,ALL resistors,ALL capacitors and ALL wires change the sound and no way around it so what we do when we pick out audio devices for personal use is to find the ones that sound pleasing to use.that do the least amount of harmonic damge but never ever are we dead nuts nuetral even though we fool ourselves and toss the phrase around (usually to describe either boring electronics or amps that tend to bright which the "bandaid cap" makes a reality) The test for me of capacitor audibility is not so much tone but transients in real time.I listent o the leading edge of a placked string or the decay of an acoustic instrument and if it strikes a reality nerve or does it in fact sound like a recorded instrument. Terms like "speed" are also bullshit since even 2V/uS is plenty fast for music yet you see RF amps (1000v/uS) touted as a good thing.I think listening fatique is an artifact of this race for speed and race for upper octave detail that while it may sound exiting at first becomes a very un natural,very non organic experience over time and make the search for a new amp imperitive even though most have no idea why and make the exact same mistake in amp selection in round ll or round lll. Music and equipment is all about choices and taste but to be honest I try to slow my amps down rather than speed them up and over time I hear MORE detail but it is presented in such a "natural for humans" way it is not as IN YOUR FACE exiting but three or four hours in you realise you were listening to music and not electronics and that to me is the test of an amp-forgetting it is there while dancing around the house naked playing air guitars,the only instrument the rickmonster is good at
|
|
|
Post by PinkFloyd on Apr 24, 2006 19:32:14 GMT
As I said Rick, some amps benefit from speeding up (some valve amps) and others don't
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Apr 24, 2006 20:23:08 GMT
BZZZZZZZT ! wrong answer BTW-what is "faster" than a vacuum? all capacitors are filters and the only difference between signal path and power supply use is in the PS connection they store voltage over/under the nads shunted to ground (filtered) and in the signal path pass everything above the hinge point so a dual cap is one where everything from the lowest bass of the cutoff is passed to the output which is then combined with another signal having a much higher hinge point but being single pole high-pass filters (-6dB per octave sloping) there is a seriously broad overlap between the two where you actually have two versions of the very same signal. The larger the "bypass" the lower down into the music this will take place Speed up triodes ? NOT ! More like use a proper transformer load so the coupling cap can be kept to a small value where the choices in good sounding caps are many. Way easier to load a cap into 5 or 10K primary than it is to load it into 40-600 ohm headphone load but for this you pay a heavy penalty in added cost.Why comprimises are made in commercial gear-too costly means it will not sell in quantity so many times it is get the sound as good as posible within a target price point rather than try and build the best possible device
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Apr 24, 2006 20:29:14 GMT
BTW-I will take great sonics in the 30-15khz range and a round square wave on the scope over what passes for SOTA in headphone amps.The rush to out spec and have pretty curve traces is precisely why the sonics are wrong but to be honest and build honest would just have some knucklhead reviewer come along and slam the product even while admitting the amp sounds good. how many times is it written " Amp a sounded great but I have a feeling amp B is more accurate so I chose amp B as the head to head winnder " WTF ! I always thought it was the sound that counted not pleasing the lab rats and it seems a bit backwards when a reviewer feels so guilty about liking an amp that he chooses the inferior sounding one to be a lot like walking backwards. you may get there but you risk walking right off a damn cliff you never seen first
|
|
|
Post by PinkFloyd on Apr 24, 2006 22:01:13 GMT
BTW-what is "faster" than a vacuum? A speeding bullet? You fell right into that one man I can wind you up faster than a speeding vacuum pump ;D
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Apr 24, 2006 22:06:40 GMT
heh,not exactly a secret is it Mikey BTW-don't make me come over there man,won't be pretty ;D
|
|
FritzS
Been here a while!
Sound of Blue Danube
Sound of Blue Danube
Posts: 1,364
|
Post by FritzS on Apr 26, 2006 8:23:57 GMT
|
|
rickcr42
Fully Modded
Rest in peace my good friend.
Posts: 4,514
|
Post by rickcr42 on Apr 26, 2006 13:44:28 GMT
The thing the "theorists" never adress is the simple "how do I eliminate DC from the output of an audio amp ?" with the answer being either a capacitor,a DC balance control that will at some point drift,a DC servo which brings in effects every bit as audible as a capaitor plus complicates the simple or a fully balanced amp from input to output with every single part matched to a tight tolerance which again means drift at some point so back to a DC offset. Trying to reinvent the wheel is where many would be audio circuit designers fall down and by elminating one thing either allow worse "glitches" to enter,way overcomplicate a simple circuit AND forget that an audio amp is happiest and most stable in real world use when that amp has both a high and low cutoff point. flat to DC (0-hz) is DC and flat to the mhz range means susceptable to RFI,oscillation through loop instability and more often than not veru tiring "etched" highs because of the tendency of these upper registers to have componants that are mathmatical copies of the original which being low level in nature are not often recognised as the source of the sonic problem. Most would be 100% happy with an amp dead flat from 30hz to 15khz until they KNEW the amp rolled off at the extreme top and bottom ends then no matter how good the amp sounded would be on a mission to find a replacement. Shitty opamp and shitty transistor amps measure great and have spoiled the so called "audiophile" into wanting similiar "lab" response and pretty square waves even though it is at the cost of SQ many times. the big cap/little cap bypass to me is on the same level.An attempt to both satisfy the NEED to reproduce a square wave not having rounded edges which you must have if there is a filter involved where it should be AND to replace part of the sonic signature most have grown up with in the didgital age that to me is a pale imitation of the harmonically pleasing/performance accurate sound of analog which BTW we humans are-analog creatures. what some call "dull" I call accurate and why I both do not fear the capacitor but would rather keep the timing between instruments dead on and as recorded rather than add a segmented multi-signal path "stew" of dissimiliar sonics by using the single best single capacitor i can find for the specific point in the signal path
|
|