I'm not sure if that was the product, but even if the photocell's signal has minimal or no processing, surely the focus/position servos will affect the intercepted beam and hence the audio signal. And I use this as an example of the contradictions inherent in the feedback antipathy of my friend, not as a criticism (however the details go) of the design.
Actually, after a while he found he couldn't bear listening to the system, despite insisting that it was still the best, because he heard a deterioration as things got further into the given LP that he found difficult to describe, but that became more and more noticeable and intolerable.
This reminds me, peripherally, of a chatroom I was in, where a participant alleged that the grooves in LPs were simply encoding individual bits. His profession, upon questioning: Law. You can't make this stuff up.
Not really good for that. They don't translate. Really good for rapid prototyping of ACTIVE crossovers. Also good for all the testing of raw drivers in the preliminary design.
There are lots of "improvements" for the DCX, many of dubious value that seem to be from people who don't believe in digital or op amps. I am waiting for a tube buffer for one.
I will keep using mine for what I bought it for. Quick evals and prototyping active crossovers. You can dial in basic sub crossover LT and notches pretty quick for a given room. Next time I put in in line with my subs, I may just leave it there.
Hi,
It is the commercial one I am aware of.
First, as said, I have not taken one apart, but supposedly the signal is plain directly from the photocell.
As for the rest, yes, they may have some impact at very low frequencies, but it is worth looking at the patent and the mechanical design of the actual product to get a feel how fast things can be adjusted.
Essentially, your claim of "feedback city", at least within the audio range to the best of my knowledge inaccurate.
And inadvertently you have also illustrated something else.
I found the "Laser Turntable incredibly interesting technologically but was very happy to go back to the "three-armed bandit", a pure "traditionalist" unsuspended, brute force "heavy metal" turntable, fitted with three different arm/cartridge combo's.
Ciao T
It is the commercial one I am aware of.
First, as said, I have not taken one apart, but supposedly the signal is plain directly from the photocell.
As for the rest, yes, they may have some impact at very low frequencies, but it is worth looking at the patent and the mechanical design of the actual product to get a feel how fast things can be adjusted.
Essentially, your claim of "feedback city", at least within the audio range to the best of my knowledge inaccurate.
And inadvertently you have also illustrated something else.
I found the "Laser Turntable incredibly interesting technologically but was very happy to go back to the "three-armed bandit", a pure "traditionalist" unsuspended, brute force "heavy metal" turntable, fitted with three different arm/cartridge combo's.
Ciao T
Hi,
First, the analogue design of the 2496 series is extremely hamfisted, especially when compared to the elegance of circuits in the earlier 8024 range.
It seems someone was paid to find a way to cram the maximum number of the cheapest possible Op-Amp's possible between a balanced DAC and an XLR Socket.
The most "dubious"options (direct output via R/C or Transformer) actually make the most sense.
If you want "tube buffers", I have designed a "universal" one that can be applied to the DCX/DEQ (and pretty much anything else with a DAC Chip), available from one of the sponsors/advertisers of this site (diyhifisupply).
Ciao T
There are lots of "improvements" for the DCX, many of dubious value that seem to be from people who don't believe in digital or op amps. I am waiting for a tube buffer for one.
First, the analogue design of the 2496 series is extremely hamfisted, especially when compared to the elegance of circuits in the earlier 8024 range.
It seems someone was paid to find a way to cram the maximum number of the cheapest possible Op-Amp's possible between a balanced DAC and an XLR Socket.
The most "dubious"options (direct output via R/C or Transformer) actually make the most sense.
If you want "tube buffers", I have designed a "universal" one that can be applied to the DCX/DEQ (and pretty much anything else with a DAC Chip), available from one of the sponsors/advertisers of this site (diyhifisupply).
Ciao T
I don't see why not ? A transfer function is a transfer function regardless of whether you got there passively or actively. What matters is the signal seen at the driver terminals.
Sure, passive and active designs respond slightly differently to dynamic driver parameter variations such as Re with temperature and Le with excursion etc, but a good passive design will minimize these effects as much as possible anyway. (For example using high sensitivity drivers whose voice coils don't get that hot because they're not dissipating much power...)
My current system is a passive/active hybrid, the high pass and low pass filters for each driver are passive, level matching is done passively, (well, the drivers themselves are fortuitously sensitivity matched at the overlap region, so there are no L-Pad's needed) but the baffle step correction as well as some significant response shaping including 3 notch filters is done actively.
Doing a lot of it actively allows very rapid iteration and adjustment of the design. If I measure the response of the drivers and see a resonance that needs correcting, its extremely quick to prototype that with a parametric equalizer and see what the audible result of fixing it is.
With any passive design you don't have an unlimited number of coefficients available, so its very helpful to find out which response aberrations need correcting (and how accurately) and which don't...just because there is a wiggle in the response doesn't necessarily mean that you should do something about it, on the other hand sometimes an innocuous looking response error can turn out to be quite audible and in need of correcting.
When I'm happy with the measured and audible response if I measure the response at the driver terminals due to the sum of the passive and active networks, then modify the passive network to give the same response at the driver terminals without the correction applied actively then the active correction has served its rapid prototyping function...
It doesn't help design the network for me but it helps to test and confirm (with listening tests and measurements) the desired transfer functions before going to the trouble of actually building them passively.
Last edited:
BIG, BIG difference, but this you know. Using an active crossover, the reactive elements of the driver are isolated from the filter. What I was getting to were those who think a third order 1K active filter is the same as one calculated on the WEB for an "8 ohm driver" are the same. Not those who understand they want to realize the same result transfer function. Sorry if I miss interpreted. A DCX is a lot cheaper than the pile of caps and coils needed for their old spring board.
Went looking for FRED ringing in the DH-120. Put a lot more effort into it. Could not find any, but I sure confirmed the discussions of the garbage coming in the power line. Yuk. from Khz to 100Mkz. All were considerably diminished by putting my Trip-Lite on the bench. Two thirds maybe on the Khz end, pretty much on the Mhz end. I look forward to my CORCOM order.
I just knew somebody would have a tube buffer! No, I don't want one, as for me this is just test equipment.
Ham-fisted or cost driven? Same result I guess. The thing was dirt cheap and intended for low end PA work. I think they did a pretty good job. That we even think about using it for other things is some credit to them.
If I wanted to green one, I think I would start with another chassis to build decent power supplies, then maybe THAT for IO and then look at the rest of the analog. You probably have external clocking anyway if you are doing studio type work. I am not convinced one way or the other, Jensen or THAT. Chips sure are cheaper and smaller.
I bought some Chine's "kits" that were surface mound headphone amps to practice smt removal and refit. They were cheaper than a prototype board and came with enough parts I can barely see to practice with. I did OK with an iron. I am going to make a hot air gun and next time Goodwill has a $5 toaster oven, try that route. $15 for a dram of solder paste is pretty steep though. I think I would go after the op amps with just a little more practice. Vision is the hardest part.
Ham-fisted or cost driven? Same result I guess. The thing was dirt cheap and intended for low end PA work. I think they did a pretty good job. That we even think about using it for other things is some credit to them.
If I wanted to green one, I think I would start with another chassis to build decent power supplies, then maybe THAT for IO and then look at the rest of the analog. You probably have external clocking anyway if you are doing studio type work. I am not convinced one way or the other, Jensen or THAT. Chips sure are cheaper and smaller.
I bought some Chine's "kits" that were surface mound headphone amps to practice smt removal and refit. They were cheaper than a prototype board and came with enough parts I can barely see to practice with. I did OK with an iron. I am going to make a hot air gun and next time Goodwill has a $5 toaster oven, try that route. $15 for a dram of solder paste is pretty steep though. I think I would go after the op amps with just a little more practice. Vision is the hardest part.
Hi,
It's not a buffer. It's basically a 2-channel uncommitted Tube gain stage with selectable local feedback for 6, 12 and 30dB gain (no feedback with 30dB gain). Resistor I/V conversion for +/-1mA and +/- 2mA full scale current DAC's is also on board.
It's basically a building block for a range of audio designs. Diyhifisupply use the same module for various DAC's, there are line Stages and even Phono Stages and Mike Pre's.
No. Hamfisted. The original DEQ8024 (which I reviewed ages ago for TNT-Audio and which may have contributed to bringing Behringer gear to the attention of Audiophiles and DIY Folks) used a by far more elegant circuit with only one dual Op-Amp for outputs and one and halve dual for inputs.
Cost driven would not have added additional Op-Amp's, performance driven would not have added them, reliability driven would not have added them.
Only hamfisted ejits would add them anyway to make Sturgeons Law is observed.
As the Audio Board is separate from the digital section with the DAC's etc. and has a nice ribbon cable running across from one PCB to the other.
So it would be easy to make a replacement PCB with RCA Jacks and discrete, feedback optional (feedback for those who think gear without feedback sounds bad, no feedback for those who think that gear with feedback sounds bad) and using TH Parts so our generation can still see the bleedin buggers...
Ciao T
It's not a buffer. It's basically a 2-channel uncommitted Tube gain stage with selectable local feedback for 6, 12 and 30dB gain (no feedback with 30dB gain). Resistor I/V conversion for +/-1mA and +/- 2mA full scale current DAC's is also on board.
It's basically a building block for a range of audio designs. Diyhifisupply use the same module for various DAC's, there are line Stages and even Phono Stages and Mike Pre's.
No. Hamfisted. The original DEQ8024 (which I reviewed ages ago for TNT-Audio and which may have contributed to bringing Behringer gear to the attention of Audiophiles and DIY Folks) used a by far more elegant circuit with only one dual Op-Amp for outputs and one and halve dual for inputs.
Cost driven would not have added additional Op-Amp's, performance driven would not have added them, reliability driven would not have added them.
Only hamfisted ejits would add them anyway to make Sturgeons Law is observed.
As the Audio Board is separate from the digital section with the DAC's etc. and has a nice ribbon cable running across from one PCB to the other.
So it would be easy to make a replacement PCB with RCA Jacks and discrete, feedback optional (feedback for those who think gear without feedback sounds bad, no feedback for those who think that gear with feedback sounds bad) and using TH Parts so our generation can still see the bleedin buggers...
Ciao T
Thorsten,
You bring up a good point I don't understand. We have a world where just about everything produces 1 to 1.5 volts output. All our amps product max output with about the same. Why so much gain in preamps? It seems like all we are doing is amplifying it just to burn it off in a pot to amplify it again. Instead of a gain of 10, why not a gain of 4? Use all that extra for something constructive like negative feedback? What good is a preamp that swings 10 volts? Would one that swings 4 very well be better? When I modeled the Rotel RA 840, I noticed it produced maximum output at 150mv in. So I am throwing away a gain of ten in the volume control which raises the source impedance on the input stage etc. I like your idea of selectable feedback. I might extend that to a step of 2.
I have seen old school dissection microscopes of about 20 power at auction. If I had space, that would help.
So, with proper analog and power, do you consider the DCX digital section to be suitable for high end? I was quite surprised how big a difference a decent external TI or Wolfson DAC made on my CD players. None of mine would be considered high end, but my "Muse"/NAD is at least as good as my Rotel 1070 and way ahead of any of my other CD's with internal DAC.
You bring up a good point I don't understand. We have a world where just about everything produces 1 to 1.5 volts output. All our amps product max output with about the same. Why so much gain in preamps? It seems like all we are doing is amplifying it just to burn it off in a pot to amplify it again. Instead of a gain of 10, why not a gain of 4? Use all that extra for something constructive like negative feedback? What good is a preamp that swings 10 volts? Would one that swings 4 very well be better? When I modeled the Rotel RA 840, I noticed it produced maximum output at 150mv in. So I am throwing away a gain of ten in the volume control which raises the source impedance on the input stage etc. I like your idea of selectable feedback. I might extend that to a step of 2.
I have seen old school dissection microscopes of about 20 power at auction. If I had space, that would help.
So, with proper analog and power, do you consider the DCX digital section to be suitable for high end? I was quite surprised how big a difference a decent external TI or Wolfson DAC made on my CD players. None of mine would be considered high end, but my "Muse"/NAD is at least as good as my Rotel 1070 and way ahead of any of my other CD's with internal DAC.
Hi,
Tradition.
There are many good reasons why a lot of Poweramp's sound better with an active preamp (maybe better to be precise - one with low output impedance), and if gain, it seems people are convinced it gain's much (pun intended).
There is also the 12 O'Clock Dynamics brigade, they claim that if you have to turn the volume above 12 O'Clock the system has "bad dynamics"...
Why stop there? Why not unity?
In my experience using "average efficiency" speakers (so 87dB/2.83V/1m) gives and "industry standard" output levels of 2V at full scale gives normal "audiophile" listening levels with around 16dB overall gain and "realistic levels" at around 26dB overall gain.
Seeing any gain actually used is rare.
I would suggest use all that extra for something de-constructive, like removal of gainstages and negative feedback leaving a simple buffer with ideally J-Fet input (to not load the wiper of the attenuator) and either direct output (Z-Load >= 10KOhm) or a beefy diamond Buffer Z-Load < 10KOhm.
Yes, that is a bit on the low side, i prefer around 30 -50V RMS before hard clipping...
USB based high magnification digital cameras are now common, cheap and tiny. I can still work on SMD down to 0603 size with naked eyes. It is the shaking of the hands that becomes a bit of a problem (not that my hands shake when you look at them, they seem rock steady.
Just the fine motor skills ain't what they used to be when I used to (occasionally) shoot the ace out of the card at 500m with a Dragunov on a tripod. Now postioning a chip cap 0.06in * 0.03in cap with tweezers to a lot less less than 1/10th of dimensions becomes a challenge.
I am a little weary of Behringers DSP programming, but the answer is "probably". Certainly there is a lot of flexibility in the Behringer that you cannot get in analogue domain at all or just with incredibly complex stuff, at the press of a button and twist of a knob.
The ADC's and DAC's are Delta Sigma and not really my favourites, but I heard them in quite expensive recording gear where they did okay.
I'd probably go digital in to cut out the A2D.
Ciao T
Tradition.
There are many good reasons why a lot of Poweramp's sound better with an active preamp (maybe better to be precise - one with low output impedance), and if gain, it seems people are convinced it gain's much (pun intended).
There is also the 12 O'Clock Dynamics brigade, they claim that if you have to turn the volume above 12 O'Clock the system has "bad dynamics"...
Why stop there? Why not unity?
In my experience using "average efficiency" speakers (so 87dB/2.83V/1m) gives and "industry standard" output levels of 2V at full scale gives normal "audiophile" listening levels with around 16dB overall gain and "realistic levels" at around 26dB overall gain.
Seeing any gain actually used is rare.
I would suggest use all that extra for something de-constructive, like removal of gainstages and negative feedback leaving a simple buffer with ideally J-Fet input (to not load the wiper of the attenuator) and either direct output (Z-Load >= 10KOhm) or a beefy diamond Buffer Z-Load < 10KOhm.
Yes, that is a bit on the low side, i prefer around 30 -50V RMS before hard clipping...
USB based high magnification digital cameras are now common, cheap and tiny. I can still work on SMD down to 0603 size with naked eyes. It is the shaking of the hands that becomes a bit of a problem (not that my hands shake when you look at them, they seem rock steady.
Just the fine motor skills ain't what they used to be when I used to (occasionally) shoot the ace out of the card at 500m with a Dragunov on a tripod. Now postioning a chip cap 0.06in * 0.03in cap with tweezers to a lot less less than 1/10th of dimensions becomes a challenge.
I am a little weary of Behringers DSP programming, but the answer is "probably". Certainly there is a lot of flexibility in the Behringer that you cannot get in analogue domain at all or just with incredibly complex stuff, at the press of a button and twist of a knob.
The ADC's and DAC's are Delta Sigma and not really my favourites, but I heard them in quite expensive recording gear where they did okay.
I'd probably go digital in to cut out the A2D.
Ciao T
USB camera. Great idea. I'll look into that. So far my hands seem steady enough. I am sure it is only time.
While waiting on my Digi-Key order, I went back to pick up where I left off on my last set of speakers. A Seas 27TBFC/Zaph budget tower set for my Nephew. Got the first prototype crossover mocked up on my terminal board so time to build the second one and do final voicing. Gasp do the inductors keep going up. A pretty simple 4th LR effective crossover, BSC, and a couple of notches, costs more than an electronic crossover! This is using Jensen coils and dayton caps. Nothing exotic. I went looking for bulk wire and it costs more than pre-made coils. One more reason to quit trying to make metal cone mids to behave. If I get back to paper, I can push the crossover a bit higher and save half on the crossover. Spend that on a better driver in the first place.
At least fumbling around on the WEB I found my old wire supplier when I worked on my cars. KJ Co. 18 ga. 19 strand GXL or TXL wire looks to be a pretty decent generic hook-up wire. (Not for tube B+, you need thicker insulation) What I can get around town is more like 7 to 12 strand with cheap vinyl insulation. KJ sells it by the foot in dozens of colors. They were quite decent to deal with. Now I need to get a spool of something more suitable for internal signal hookup. I have just about used up my supply of "free" interconnect cables. I had some fantastic HP spec mini coax but used it up. If anyone knows some Beldon numbers they have had good results with, it would be appreciated. Even with wire, it is getting expensive to guess wrong. I have a 50' s-video cable I longer use, maybe I'll just start using that. It is just a pair of miniature 75 Ohm coaxes.
While waiting on my Digi-Key order, I went back to pick up where I left off on my last set of speakers. A Seas 27TBFC/Zaph budget tower set for my Nephew. Got the first prototype crossover mocked up on my terminal board so time to build the second one and do final voicing. Gasp do the inductors keep going up. A pretty simple 4th LR effective crossover, BSC, and a couple of notches, costs more than an electronic crossover! This is using Jensen coils and dayton caps. Nothing exotic. I went looking for bulk wire and it costs more than pre-made coils. One more reason to quit trying to make metal cone mids to behave. If I get back to paper, I can push the crossover a bit higher and save half on the crossover. Spend that on a better driver in the first place.
At least fumbling around on the WEB I found my old wire supplier when I worked on my cars. KJ Co. 18 ga. 19 strand GXL or TXL wire looks to be a pretty decent generic hook-up wire. (Not for tube B+, you need thicker insulation) What I can get around town is more like 7 to 12 strand with cheap vinyl insulation. KJ sells it by the foot in dozens of colors. They were quite decent to deal with. Now I need to get a spool of something more suitable for internal signal hookup. I have just about used up my supply of "free" interconnect cables. I had some fantastic HP spec mini coax but used it up. If anyone knows some Beldon numbers they have had good results with, it would be appreciated. Even with wire, it is getting expensive to guess wrong. I have a 50' s-video cable I longer use, maybe I'll just start using that. It is just a pair of miniature 75 Ohm coaxes.
On a different subject, here's what's nagging me: at what level of measured performance would you say that you're in no trouble? When is enough - enough?
For many years, 0.1% was considered to be "the magic line", menaing that once you get your THD, IM, etc values below 0.1%, and assuming your are not going wild with NFB, you're in the clear.
Perosnally, I prefer that "magic line" to be more like 0.05% into 4 Ohms, because they are more difficult, so I figure if an amp can cope nicely with 4 Ohms, it should be a clean one (but still not necessarily great sounding, but that's several other matters).
I've never investigated this in any organized way, but experience teaches me that this can be obtained with no more than 26 dB of overall NFB and without breaking the bank, and it should sound relatively clean and trouble free.
For many years, 0.1% was considered to be "the magic line", menaing that once you get your THD, IM, etc values below 0.1%, and assuming your are not going wild with NFB, you're in the clear.
Perosnally, I prefer that "magic line" to be more like 0.05% into 4 Ohms, because they are more difficult, so I figure if an amp can cope nicely with 4 Ohms, it should be a clean one (but still not necessarily great sounding, but that's several other matters).
I've never investigated this in any organized way, but experience teaches me that this can be obtained with no more than 26 dB of overall NFB and without breaking the bank, and it should sound relatively clean and trouble free.
Well, the DH-120 is rated .006 THD at 20K, IM less than .005 at 60W into 8 ohms, DF of over 100 @ 20K It fails my wife's golden ears. I confirmed less than .003 at 4KHz, 1W.
The Rotel 840 is rated .03 at any level to rated 40W, for THD and IM at 20K. DF ( no freq specified) is 30, but I measured 40 at 1K. It passes her test.
I am sure there are numbers that would achieve what you ask, but THD and IM would not seem to be them. I believe this is the very subject of this thread, if I would quit hijacking it to take advantage of the expertise available to understand amp design.
The Rotel 840 is rated .03 at any level to rated 40W, for THD and IM at 20K. DF ( no freq specified) is 30, but I measured 40 at 1K. It passes her test.
I am sure there are numbers that would achieve what you ask, but THD and IM would not seem to be them. I believe this is the very subject of this thread, if I would quit hijacking it to take advantage of the expertise available to understand amp design.
the GedLee metric has some perceptual validation - heavily weights distortion at low levels
so it seems a good way to use THD, IMD is to see if they continually decrease with decreasing signal level, down into the noise - harmonic spectrum envelope shouldn't shift to higher orders as level goes down either
THD vs power is a pretty conventional plot for amp distortion – just pay more attention to the left end of the plot
the poor reputation of "conventional" distortion measurements probably comes from marketing depts searching for impressive #, touting super low THD at higher powers when stuck with an amp that has annoying levels of crossover distortion
how does your wife rate psychoacoutic compression – even 320k/-V0 is “tossing out” 75% of RedBook CD Shannon-Hartley “Channel Capacity” information
the compression algorithms only use ~6-7 bits mantissa per critical band
could be interpreted as humans seldom hearing anything much below 1% - possibly including “distortion” – that tracks signal level, fit our auditory processing "filters" structure
so it seems a good way to use THD, IMD is to see if they continually decrease with decreasing signal level, down into the noise - harmonic spectrum envelope shouldn't shift to higher orders as level goes down either
THD vs power is a pretty conventional plot for amp distortion – just pay more attention to the left end of the plot
the poor reputation of "conventional" distortion measurements probably comes from marketing depts searching for impressive #, touting super low THD at higher powers when stuck with an amp that has annoying levels of crossover distortion
how does your wife rate psychoacoutic compression – even 320k/-V0 is “tossing out” 75% of RedBook CD Shannon-Hartley “Channel Capacity” information
the compression algorithms only use ~6-7 bits mantissa per critical band
could be interpreted as humans seldom hearing anything much below 1% - possibly including “distortion” – that tracks signal level, fit our auditory processing "filters" structure
Last edited:
Hi,
Distortion audibility is a complex and multidimensional topic, SPL and Frequency feature heavily. THD is completely meaningless WRT distortion audibility.
So, if you want to be absolutely sure no distortion can be heard at any SPL and any frequency AND we insist on using THD as metric we need to set a minimum specification of better than around -20dB SPL for any harmonic at any SPL genertaed up to 120dB or so, so all distortion products would need be below -140dB at full power.
Clearly, that way doth madness lieth and the Map is marked in medieval script "than way be monsters"...
Ciao T
Distortion audibility is a complex and multidimensional topic, SPL and Frequency feature heavily. THD is completely meaningless WRT distortion audibility.
So, if you want to be absolutely sure no distortion can be heard at any SPL and any frequency AND we insist on using THD as metric we need to set a minimum specification of better than around -20dB SPL for any harmonic at any SPL genertaed up to 120dB or so, so all distortion products would need be below -140dB at full power.
Clearly, that way doth madness lieth and the Map is marked in medieval script "than way be monsters"...
Ciao T
@jcx
If I understood you properly, you are referring to the phenomenon knows as "distortion comeback", which causes distortion to rise, sometimes dramatically, as power output decreases below say 1W?
Right or wrong, I always thought that to be a typical class B phenomenon, which occurs when the idling (bias) current of the output stage is set at very low values, say below 20 mA (no hard value, just an example). The Japanese used to do that quite a lot in the early 70ies.
I wondered about it - why do highly regarded say Sansui models, with nominally better specs, almost always sound worse than say Grundig (for those who don't know them, a once vast German popular electronics brand, catering mosty in the El Cheapo sector, later bougth out by Philips), which had a whole lot less power and worse specs? As far as I could determine, it was due to three factors: 1) The boys in Grundig knew a whole lot more than they were credited for, 2) their amps were initially designed to drive 2 pairs of 4 Ohm speakers in parallel, so despite their much lower nominal power outputs, they were initally way more load tolerant than the Jap fare, and 3) their quisecent current levels never went below 60 mA, and were often at 70-80 mA. Thorsten may have something to add to this list.
Anyway, I took my cues from Otala's work, and he postulated that the bias level should be set so as to be at around -17 dB of nominal power output. Well, that become quite a bit as your power level rises, but I still set my amps at 100...130 mA bias.
I've tried this on all production amps I own (from Sansui, Marantz and Harman/Kardon) and have found that by about 130 mA of quiescent current, differences in sound stop. You may be able to lift that up even more, depending on your heat sink real estate, but there is no additional gain to be had in terms of sound quality.
The paradox is that if you run an amp designed for say 70 mA at say 130 mA, your nominal THD specs deteriorate somewhat, but you do get a sweeter sound and you do all but eliminate any distortion comeback. At 2.83V/8 Ohms, equivalent to 1 Watt into 8 Ohms, classic forms of distortion tend to fall below the 0.001% mark, which is the limit of reliability of my equipment. The sound becomes audiably warmer and sometimes a little "rounded off", but it can also cause some loss of clarity, the overall becomes better but at the expense of some finer detail. This was particularly noticeable on my H/K 680 integrated (from 1999), while the older 6550 model was the happiest at its factory set value of 70 mA.
Not really surprised - increasing bias values makes them work at levels which they were not designed for, so anything can happen. But in general, Japanese products seem to benefit much from this, so long as you don't overdo it. In their case, you know you're overdoing it when their heat sinks become uncomfortably hot to touch (for models made after say 1975, before that time they weren't skimping on heat sinks) at even low nominal power levels.
That taught me an important lesson - there's no such thing as too much heat sinking.
If I understood you properly, you are referring to the phenomenon knows as "distortion comeback", which causes distortion to rise, sometimes dramatically, as power output decreases below say 1W?
Right or wrong, I always thought that to be a typical class B phenomenon, which occurs when the idling (bias) current of the output stage is set at very low values, say below 20 mA (no hard value, just an example). The Japanese used to do that quite a lot in the early 70ies.
I wondered about it - why do highly regarded say Sansui models, with nominally better specs, almost always sound worse than say Grundig (for those who don't know them, a once vast German popular electronics brand, catering mosty in the El Cheapo sector, later bougth out by Philips), which had a whole lot less power and worse specs? As far as I could determine, it was due to three factors: 1) The boys in Grundig knew a whole lot more than they were credited for, 2) their amps were initially designed to drive 2 pairs of 4 Ohm speakers in parallel, so despite their much lower nominal power outputs, they were initally way more load tolerant than the Jap fare, and 3) their quisecent current levels never went below 60 mA, and were often at 70-80 mA. Thorsten may have something to add to this list.
Anyway, I took my cues from Otala's work, and he postulated that the bias level should be set so as to be at around -17 dB of nominal power output. Well, that become quite a bit as your power level rises, but I still set my amps at 100...130 mA bias.
I've tried this on all production amps I own (from Sansui, Marantz and Harman/Kardon) and have found that by about 130 mA of quiescent current, differences in sound stop. You may be able to lift that up even more, depending on your heat sink real estate, but there is no additional gain to be had in terms of sound quality.
The paradox is that if you run an amp designed for say 70 mA at say 130 mA, your nominal THD specs deteriorate somewhat, but you do get a sweeter sound and you do all but eliminate any distortion comeback. At 2.83V/8 Ohms, equivalent to 1 Watt into 8 Ohms, classic forms of distortion tend to fall below the 0.001% mark, which is the limit of reliability of my equipment. The sound becomes audiably warmer and sometimes a little "rounded off", but it can also cause some loss of clarity, the overall becomes better but at the expense of some finer detail. This was particularly noticeable on my H/K 680 integrated (from 1999), while the older 6550 model was the happiest at its factory set value of 70 mA.
Not really surprised - increasing bias values makes them work at levels which they were not designed for, so anything can happen. But in general, Japanese products seem to benefit much from this, so long as you don't overdo it. In their case, you know you're overdoing it when their heat sinks become uncomfortably hot to touch (for models made after say 1975, before that time they weren't skimping on heat sinks) at even low nominal power levels.
That taught me an important lesson - there's no such thing as too much heat sinking.
- Status
- Not open for further replies.