Konnichiwa,
Hmmm. I think I prefer NS doing what they did. As they did their homework and implemented the quasi complementary output stage properly, with correct thermal tracking and equal speed devices for both halve waves, plus minimised crossover distortion it is very hard to do better discrete.
You will do usually much worse due to problems with thermal coupling at low levels unless you run a significant class A current and should you be deluded enough to use complementary devices you will have added assymetrical timing in the output stage, which matters if there is no NFB but ballses up things royally when there is NFB around the Output stage.
I'd prefer to have the Output Stage of the NS Chip as Power chip and the ability to have a choice of drivers....
Sayonara
millwood said:
Hmmm. I think I prefer NS doing what they did. As they did their homework and implemented the quasi complementary output stage properly, with correct thermal tracking and equal speed devices for both halve waves, plus minimised crossover distortion it is very hard to do better discrete.
You will do usually much worse due to problems with thermal coupling at low levels unless you run a significant class A current and should you be deluded enough to use complementary devices you will have added assymetrical timing in the output stage, which matters if there is no NFB but ballses up things royally when there is NFB around the Output stage.
I'd prefer to have the Output Stage of the NS Chip as Power chip and the ability to have a choice of drivers....
Sayonara
Eric wrote:
You who do make a living of servicing equipment should be aware that it is impossible to make anything idiot proof since idiots are much too inventive !!!
A friend of mine is doing P.A. work for a living. He is in the sales and rental busines. He once had a customer (small music club) he had sold a processor P.A. (Audio Performance, something like a Swiss made Meyersound P.A.) to.
Some day one of the guys there found out that he can remove a specific cable and that he would have much more bass (it was the control line from the LF poweramp's output to the processor). It didn't last very long though.......
Regards
Charles
You who do make a living of servicing equipment should be aware that it is impossible to make anything idiot proof since idiots are much too inventive !!!
A friend of mine is doing P.A. work for a living. He is in the sales and rental busines. He once had a customer (small music club) he had sold a processor P.A. (Audio Performance, something like a Swiss made Meyersound P.A.) to.
Some day one of the guys there found out that he can remove a specific cable and that he would have much more bass (it was the control line from the LF poweramp's output to the processor). It didn't last very long though.......
Regards
Charles
Re: summarizing...
Konnichiwa,
This is a little too blanket. If your speakers are 97db/W/m and your amp's have > 6db dynamic headroom (subjectively not distorted with music) you need only 5 - 6W RMS. Now if your Speakers are 87db/W/m and there is NO or little dynamic headroom (most SS Amp's) you need around 200W....
So, the 25W gaincard with it's around 3db dynamic headroom (very soft supply, sustains only 25W RMS, but handles short peaks of > 50W RMS equivalent) needs at least (in my estimation) 93db/W/m speaker efficiency to allow sufficient dynamic range.
A 50W RMS Amp with no dynamic headroom BTW also needs 93db/W/m speakers....
BTW, this not to say that Gaincard/Gainclone are not good Amp's, HOWEVER, they do not like low impedance speakers or low sensitivity speakers (but then, who apart from Don D'Agostino does and he likes them only because they make him sell his Amp's).
Ahhhm. Actually, most chip amp's have many more parts than most discrete, but of course, you don't see these parts as they are inside the "black box".
For example, contrary to most discrete Amplifiers the NS LM38XX series Amplifiers use current source loads EVERYWHERE possible. Contrary to most dicrete Amp's the NS LM38XX series Amplifiers use Emitter followers (usually current-source loaded) in front of EACH AND EVERY amplification stage, killing (okay, drastically reducing) the effect of the nonlinear Beta (current gain) of the Bipolar trasistors. They also cascode a few things.... All measures that enhance the inherent circuit linearity NO END.
Few discrete circuits can afford that kind of attention to detail and consequent dealing with problems. Also, any "differential pair" in a chip amplifier tends to be tightly matched and tightly thermally coupled....
Practically too, for the reasons listed above.
Yes, thanks to both MF Audio for making my ideas into products and to HFW for accepting a "hairshirt" passive pre.
Sayonara
Konnichiwa,
carlosfm said:
This is a little too blanket. If your speakers are 97db/W/m and your amp's have > 6db dynamic headroom (subjectively not distorted with music) you need only 5 - 6W RMS. Now if your Speakers are 87db/W/m and there is NO or little dynamic headroom (most SS Amp's) you need around 200W....
So, the 25W gaincard with it's around 3db dynamic headroom (very soft supply, sustains only 25W RMS, but handles short peaks of > 50W RMS equivalent) needs at least (in my estimation) 93db/W/m speaker efficiency to allow sufficient dynamic range.
A 50W RMS Amp with no dynamic headroom BTW also needs 93db/W/m speakers....
BTW, this not to say that Gaincard/Gainclone are not good Amp's, HOWEVER, they do not like low impedance speakers or low sensitivity speakers (but then, who apart from Don D'Agostino does and he likes them only because they make him sell his Amp's).
Originally posted by carlosfm Anyway, thinking of an ideal amplifier (when are we going to have a wire with gain?), I will point out two main advantages of chip amps:
1. Short signal path.
2. Much less components.
[/B]
Ahhhm. Actually, most chip amp's have many more parts than most discrete, but of course, you don't see these parts as they are inside the "black box".
For example, contrary to most discrete Amplifiers the NS LM38XX series Amplifiers use current source loads EVERYWHERE possible. Contrary to most dicrete Amp's the NS LM38XX series Amplifiers use Emitter followers (usually current-source loaded) in front of EACH AND EVERY amplification stage, killing (okay, drastically reducing) the effect of the nonlinear Beta (current gain) of the Bipolar trasistors. They also cascode a few things.... All measures that enhance the inherent circuit linearity NO END.
Few discrete circuits can afford that kind of attention to detail and consequent dealing with problems. Also, any "differential pair" in a chip amplifier tends to be tightly matched and tightly thermally coupled....
Practically too, for the reasons listed above.
Originally posted by carlosfm Anyway, thinking of an BTW, Kuei, your pre was reviewed in Hi-Fi World.
They say it's very good.[/B]
Yes, thanks to both MF Audio for making my ideas into products and to HFW for accepting a "hairshirt" passive pre.
Sayonara
This cracks me up...
As was pointed out, there are many times more components internally in the "little black box".
Minimalist chip amps are a contradiction of terms.
In my opinion, there are two schools of thought when it comes to amplifier design:
1- True minimalist- as in the fewest series gain devices possible, and minimal feedback.
The point being that the fewer gain devices (which, as Nelson Pass says, are most responsible for distortions in the amp) the least amount of damage to the signal.
2- Lowest measured distortion- This school of thought uses all the tricks- current sources, mirrors, active loads, etc. If another stage will lower the distortion, go for it! Feedback in whatever amount gets the job done is acceptable.
Op amps fall into this category, no matter how much (or little) wire you use to hook them up.
I won't be the one to say which is the right way. I guess it depends on who you are, and your preferences. Maybe the truth is somewhere in between?
I fall into the first category, which is inherently discrete by nature. My reasoning is:
1. I believe that greater amounts (but not ridiculous amounts, of course) of low order distortion is preferable to smaller amounts of higher order distortion.
2. I believe that feedback, which is a subtractive process, tends to subtract more than just distortion when overused.
With these "I believes" it sounds a little like religion, and that might not be far from the truth.
Anyway, if I were to build a "type 2" amp, I would probably not bother using discretes, because it is pointless. If you're going to use multiple gain devices, etc. you might as well go with those with the best matching (and all the other benefits mentioned before).
So, my vote is- discretes are better for minimalist techniques, and chips are better when the numbers matter more than the sound.
Just my opinion, of course.
Yeah, who was it that said there is an idiot born every minute !!!.phase_accurate said:Eric wrote:
You who do make a living of servicing equipment should be aware that it is impossible to make anything idiot proof since idiots are much too inventive !!!
A friend of mine is doing P.A. work for a living. He is in the sales and rental busines. He once had a customer (small music club) he had sold a processor P.A. (Audio Performance, something like a Swiss made Meyersound P.A.) to.
Some day one of the guys there found out that he can remove a specific cable and that he would have much more bass (it was the control line from the LF poweramp's output to the processor). It didn't last very long though.......
Regards
Charles
Live sound guys are usually pretty responsible so long as the drinks rider is after the show (most live engineers know not to drink on the job - bad for the rep) and DJ's are the worst - if there is a way to overdrive and break a system they will find it - cloth ears and too much chemical enhancement is the cause as I understand it.
Eric.
Re: Re: summarizing...
Don't let the specs fool you.
I was very sceptic too, before doing my GC, that they would drive my Epos 11 speakers properly. (8 ohms, 86~87db).
I read the datasheet of the LM3875 and wasn't convinced.
I can tell you I never thought it was possible that these speakers could deliver such an impressive dynamics.
My Nad amp (about the same power) wasn't man enough.
My LM3875 GC is.
I think that more important than power is current delivery.
But even than, the LM3875 is nothing special in this field.
Strange animal...
How could he drive some tiny Tangent 3 speakers (10cm woofers, 87db, 8 ohm) on a 25 square meter room in a way that it made us (I was not alone) preffer their sound to a pair of Tannoy (91db) and a pair of Allison (90db) speakers?
I can't explain this by maths or physics, you really have to listen.
I repeat: it's a strange but beautiful animal.
I know
.
But still, it's a shorter signal path, isn't it?
It would be impossible to make today a Pentium 4 processor as they did CPUs until about 30 years ago: discrete, on big boards.
And today CPUs are much faster.
Impossible to get with discrete components.
Kuei Yang Wang said:
Don't let the specs fool you.
I was very sceptic too, before doing my GC, that they would drive my Epos 11 speakers properly. (8 ohms, 86~87db).
I read the datasheet of the LM3875 and wasn't convinced.
I can tell you I never thought it was possible that these speakers could deliver such an impressive dynamics.
My Nad amp (about the same power) wasn't man enough.
My LM3875 GC is.
I think that more important than power is current delivery.
But even than, the LM3875 is nothing special in this field.
Strange animal...
How could he drive some tiny Tangent 3 speakers (10cm woofers, 87db, 8 ohm) on a 25 square meter room in a way that it made us (I was not alone) preffer their sound to a pair of Tannoy (91db) and a pair of Allison (90db) speakers?
I can't explain this by maths or physics, you really have to listen.
I repeat: it's a strange but beautiful animal.
Kuei Yang Wang said:
I know
.But still, it's a shorter signal path, isn't it?
It would be impossible to make today a Pentium 4 processor as they did CPUs until about 30 years ago: discrete, on big boards.
And today CPUs are much faster.
Impossible to get with discrete components.
nobody special said:
I agree. my experience with gainclones has not been exactly stellar but I wasn't expecting more to begin with.
However, I do think as a DIYer, discret amps are much fun to work with and give more freedom to the builders.
But if you don't need tons of power, I don't think you can tell the two apart.
mrfeedback said:
Funny you should mention that...
I had a really interesting experience when my band was playing out one night. The lead singer's friend's band asked us to open for them, as kind of a favor. We showed up, and there were no soundguys anywhere! They had a massive system, made up of mostly JBL stuff, and it was a huge system! We got everything set up, and finally the sound guys show up- drunk off their *****, of course.
We had NO monitors... it sounded completely AWFUL! They were able to make great equipment sound like the worst garbage. It was so bad that the other bands packed up their stuff and left before they were supposed to go on!
I'll never forget that, hehe.
This is a really good way to ensure that you don't get any live mixing work !.
BTW, it is quite normal for the support act to be intentionally made to sound inferior to the main act.
A muso friend of mine did a support act for a big international touring show and sounded so good that the main act production guys cut him off early half way through his show because he was sounding too good.
Eric.
I had an idea that was the way it was, but the sound was so bad- to the point of being unbearable.
The lead singer of the band we were opening for knew what we were capable of as musicians, because he had just finished recording our demo... I think he was afraid they would sound as bad as we did.
What a night!
The lead singer of the band we were opening for knew what we were capable of as musicians, because he had just finished recording our demo... I think he was afraid they would sound as bad as we did.
What a night!
Professionalism...
In this town turning up late and blind drunk and pulling a bad mix is a guaranteed way to blow your rep bigtime - reports like that travel faster than sound and are pretty much impossible to undo.
A hot babe coming up to the desk and giving you a drink, or going outside for a smoko break during intermissions are fine, but pulling a bad mix because you had too much is considered VERY unprofessional and gets your **** kicked.
Getting sozzled on the job and blowing drivers is unforgivable and the worst rep possible.
Eric.
In this town turning up late and blind drunk and pulling a bad mix is a guaranteed way to blow your rep bigtime - reports like that travel faster than sound and are pretty much impossible to undo.
A hot babe coming up to the desk and giving you a drink, or going outside for a smoko break during intermissions are fine, but pulling a bad mix because you had too much is considered VERY unprofessional and gets your **** kicked.
Getting sozzled on the job and blowing drivers is unforgivable and the worst rep possible.
Eric.
It's pretty obvious that the more components in a design, the more opportunities your average constructor has for getting it wrong.
Many components in an amp have a secondary function (such as improving stability, or PSRR, or open-loop linearity) which means they can be wrongly fitted in some way, and the amp will still function, but it just performs poorly. Unless you have decent test equipment and an understanding of each component's function, you'll have a hard time telling apart bad design from bad construction.
If you're asking because this is the first amp you've built, going for a Gainclone or similar is a really sensible idea. It will let you play around with volume pots, layout, wiring, power supplies, connectors etc, safe in the knowledge that most of the hard-to-test bits have been done for you.
If you don't like it, for whatever reason, you can always keep the expensive bits like the PSU and case/heatsink, and replace the LM3875s with something discrete.
Cheers
IH
Many components in an amp have a secondary function (such as improving stability, or PSRR, or open-loop linearity) which means they can be wrongly fitted in some way, and the amp will still function, but it just performs poorly. Unless you have decent test equipment and an understanding of each component's function, you'll have a hard time telling apart bad design from bad construction.
If you're asking because this is the first amp you've built, going for a Gainclone or similar is a really sensible idea. It will let you play around with volume pots, layout, wiring, power supplies, connectors etc, safe in the knowledge that most of the hard-to-test bits have been done for you.
If you don't like it, for whatever reason, you can always keep the expensive bits like the PSU and case/heatsink, and replace the LM3875s with something discrete.
Cheers
IH
Re: Re: Re: summarizing...
Konnichiwa,
Note, I wrote db/W/m and NOT db/2.83V/m. Many modern speakers that show a higher "paper" sensitivity do so by lowering impedance. The ES11 is a pretty Bona Fida 8 Ohm Speaker. However, on any amp I could think on (including NAP135 and NAP 500) it stands no chance to match my Tannoys (Monitor Red 15", 230 Liter Corner Reflex Box) driven by 5 - 6 Watt in terms of maximum loudness (never mind dynamics and LF).
NAD Amp? Is that Short for "Not A Decent Amplifier"? The 3020 in it's day may not have been exactly bad but other than a cheap favourite was not much to cop. What came after was poorer sound for more money in many cases.
Maybe because the Tangents have a much higher impedance and are less reactive?
Agreed.
As in physical length of the involved current loops? Yes, of course and that is usually a good thing as well, something I agree upon with Kimura San wholeheartedly.
Sayonara
Konnichiwa,
carlosfm said:
Note, I wrote db/W/m and NOT db/2.83V/m. Many modern speakers that show a higher "paper" sensitivity do so by lowering impedance. The ES11 is a pretty Bona Fida 8 Ohm Speaker. However, on any amp I could think on (including NAP135 and NAP 500) it stands no chance to match my Tannoys (Monitor Red 15", 230 Liter Corner Reflex Box) driven by 5 - 6 Watt in terms of maximum loudness (never mind dynamics and LF).
carlosfm said:
NAD Amp? Is that Short for "Not A Decent Amplifier"? The 3020 in it's day may not have been exactly bad but other than a cheap favourite was not much to cop. What came after was poorer sound for more money in many cases.
carlosfm said:
Maybe because the Tangents have a much higher impedance and are less reactive?
carlosfm said:
Agreed.
carlosfm said:
I know
But still, it's a shorter signal path, isn't it?
As in physical length of the involved current loops? Yes, of course and that is usually a good thing as well, something I agree upon with Kimura San wholeheartedly.
Sayonara
Re: Re: Re: Re: summarizing...
Well, with a few tweaks, y ou can get excellent sound out of that amp, including that woeful NAD. start with gold-plated resistors, plentium coated caps, tungsgen reinforced heatsinks, 14K white gold jacks, and silver-plated cover, you don't have to worry about those sucky transistors anymore.
BTW, don't forget about using aluminum oxide insulation pads.
As they say in banking, there is no bad amp, just with the right resistors,
Kuei Yang Wang said:
Well, with a few tweaks, y ou can get excellent sound out of that amp, including that woeful NAD. start with gold-plated resistors, plentium coated caps, tungsgen reinforced heatsinks, 14K white gold jacks, and silver-plated cover, you don't have to worry about those sucky transistors anymore.
BTW, don't forget about using aluminum oxide insulation pads.
As they say in banking, there is no bad amp, just with the right resistors,
Re: Re: Re: Re: Re: summarizing...
Sounds like a good start for an amp! Only add some LM3875's and good 1000 uF caps!
Fedde
millwood said:
Well, with a few tweaks, y ou can get excellent sound out of that amp, including that woeful NAD. start with gold-plated resistors, plentium coated caps, tungsgen reinforced heatsinks, 14K white gold jacks, and silver-plated cover, you don't have to worry about those sucky transistors anymore.
BTW, don't forget about using aluminum oxide insulation pads.
Sounds like a good start for an amp! Only add some LM3875's and good 1000 uF caps!
Fedde
nobody special said:
Steve,
That's OK, as long as you realise that there is no reason why you should only "believe" this. Feedback is quite well known, and anything you want to know about it is either available or can be deduced with a little bit of intellectual effort.
But, as they say, have it your way.
Jan Didden
Re: Re: Re: Re: Re: summarizing...
Konnichiwa,
While I am the first one to admit to using very esotheric military surplus resistors (BTW - as surplus they are around $ 1 each - non inductive wirewound, endcapless construction with the resistor wire welded directly to the leadouts and the leadouts in most cases silver/silver alloy), in case of NAD and many other manufacturers I feel that some basic "common sence" design features (high inherent linearity prior to applying NFB, open loop bandwidth between 1 octave to 1 decade higher than the nominal signal bandwidth, no "sticking" after momentary overload) may actually pay more sonically dividends at much lower cost.
I prefer to get the basics right, before I start fishing at the edge.... BWTFDIK.
Sayonara
Konnichiwa,
millwood said:
While I am the first one to admit to using very esotheric military surplus resistors (BTW - as surplus they are around $ 1 each - non inductive wirewound, endcapless construction with the resistor wire welded directly to the leadouts and the leadouts in most cases silver/silver alloy), in case of NAD and many other manufacturers I feel that some basic "common sence" design features (high inherent linearity prior to applying NFB, open loop bandwidth between 1 octave to 1 decade higher than the nominal signal bandwidth, no "sticking" after momentary overload) may actually pay more sonically dividends at much lower cost.
I prefer to get the basics right, before I start fishing at the edge.... BWTFDIK.
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.