steenoe said:
How do you expect the heatsink to work in the illustrated amp? The only way in which heat is transmitted out of the design is via the walls of the chassis.
Ah.. Another amusing thread. To this point, only about half the posts are regarding the topic. The rest are taking that spiraled out thread nose-dive of subjectvism vs. objectivism.
My interpretation of Jan's posts weren't that the snubber doesn't work, just that it wasn't optimized. It appeared that carlos came to his values via trail and error, rather than try to understand why the whole thing works in the first place, and possibly using the correct algorithm to optimize it.
And once again, it looks like Moving Electron is struggling valiantly to keep the thread on topic. Keep at it man! (seriously, it's apprecaited.. I have a similar question: would a regulated power supply benefit from a snubber before and/or after the regulator?).
Now mind you, I'm not an EE, so those of you that actually do know this may laugh at me.. I'm just trying to understand some of the SCIENCE behind this, not any pseudo-science ****. It's not like a snubber/zobel/carlosinator is some mystical new invention in electronics -- it's already documented in several pdf files and electronics books and so on.
I think this is a continuation of a previous thread we had, in which what some of us were looking for is the reason why this works, and the formula for arriving at the optimum values.
Thorsten brought up a good point about the PSRR @ 250khz on the datasheet, so it would seem that maybe optimizing the snubber/zobel/"whatever folks want to call it today" around this frequency. Anyone care to toss up a set of values for the snubber based on a ballpark guess of say a 200VA toroid, MUR 860's and 10,000 uF of capacitence on each rail? Maybe then those who already have these parts can give it a try and see if there's a difference between this and carlos's specs.
It'd also be nice to see some test results from a scope on both the PSU and the amp output with no PSU zobel, with carlos's values, and maybe with the optimized values.
Who knows, maybe I'm an idiot smoking crack here, but maybe what I posted here makes sense to someone 😀
My interpretation of Jan's posts weren't that the snubber doesn't work, just that it wasn't optimized. It appeared that carlos came to his values via trail and error, rather than try to understand why the whole thing works in the first place, and possibly using the correct algorithm to optimize it.
And once again, it looks like Moving Electron is struggling valiantly to keep the thread on topic. Keep at it man! (seriously, it's apprecaited.. I have a similar question: would a regulated power supply benefit from a snubber before and/or after the regulator?).
Now mind you, I'm not an EE, so those of you that actually do know this may laugh at me.. I'm just trying to understand some of the SCIENCE behind this, not any pseudo-science ****. It's not like a snubber/zobel/carlosinator is some mystical new invention in electronics -- it's already documented in several pdf files and electronics books and so on.
I think this is a continuation of a previous thread we had, in which what some of us were looking for is the reason why this works, and the formula for arriving at the optimum values.
Thorsten brought up a good point about the PSRR @ 250khz on the datasheet, so it would seem that maybe optimizing the snubber/zobel/"whatever folks want to call it today" around this frequency. Anyone care to toss up a set of values for the snubber based on a ballpark guess of say a 200VA toroid, MUR 860's and 10,000 uF of capacitence on each rail? Maybe then those who already have these parts can give it a try and see if there's a difference between this and carlos's specs.
It'd also be nice to see some test results from a scope on both the PSU and the amp output with no PSU zobel, with carlos's values, and maybe with the optimized values.
Who knows, maybe I'm an idiot smoking crack here, but maybe what I posted here makes sense to someone 😀
jackinnj said:
How can you see that on steen's amp?
It doesn't show the top cover...
http://www.diyaudio.com/forums/attachment.php?s=&postid=547254&stamp=1105135028
By the way, my post wasn't meant to be a slap or disparraging towards ANYONE.. I'm just trying to steer this thing back in the direction it originally was going.. Getting the answers as to if/how/why this snubber stuff works and/or how to optimize it seems like a much more worthwhile effort than everyone bashing each other!
I would still be making the maths instead of listening to the music with immense pleasure.😎
Don't worry, it was a good listening session, starting with 560nf and going down with several values, listening to the system on each change of values.
The point is also that I've put in my head the idea that unregulated high capacitance PSU had to sound good with chip amps.
I didn't rest until I did it, and I tested many things.
Because the typical low capacitance GC can't drive most normal speakers (and I'm talking 8 ohms
).If you guys didn't notice it before, you will, when you try this PSU.😀
Anyway, would the "optimized" circuit with the "correct algorithm" sound better?
Maby yes, maby no.😉
EDIT and PS: When I (and others) said long ago that high capacitance didn't sound good with these chips some didn't believe. Most of them don't believe in the snubber too, and I see some of them again here.🙄
grege said:
The second link has helpful info related to regulated supplies. It indicates that a typical voltage regulator could be modeled as having 1uH to 2 uH of inductance is series with the output.
Thanks, Occam and grege.
carlosfm said:
Columbus said to his generals that he could put an egg standing vertical on top of a table.
After the guys scratched their heads for some time discussing it was impossible, it was demonstrated: splash!😀
He was actually supposed to find India and went around in circles until he bumped into America in 1492 and thought he had found India! Ever wondered why the native Americans are referred to as "Indians" ?
Vasco Da Gama then in 1498 discovered the correct route to India.
Thought I'd give you guys a lesson in history since anyway we are talking about everything else other than Chip Amps 😀
Interesting that it's impossible to discuss _only_ instead of "if you don't like, don't eat it".
At the moment Carlos and Thorsten have seen some light. Thorsten has visualized the problem a bit through simulation but no real measurements have taken place so we don't know exactly what the problem is nor how to fix it. I'm not the guy who tests everything I hear, like put CD's in the freezer, peel capacitor jackets off, etc. I want to hear some trustworthy technical explanantions from those who get it. I'm afraid I'm one of those who don't get it.
Let me ask the gathered audience: All Peter Daniel/Brian Bell Gainclone builders: Do you experience any remarkable about your LM3875 amps? I'll gather you all miss this snubber.
Regardless of what some people think I'm listening _and_ I haven't made my mind up.
Notice also that I assume that all caps are very close to the LM3886, like in my case. We are talking about one inch or two for the big caps and directly at the pins for the smaller caps.
At the moment Carlos and Thorsten have seen some light. Thorsten has visualized the problem a bit through simulation but no real measurements have taken place so we don't know exactly what the problem is nor how to fix it. I'm not the guy who tests everything I hear, like put CD's in the freezer, peel capacitor jackets off, etc. I want to hear some trustworthy technical explanantions from those who get it. I'm afraid I'm one of those who don't get it.
Let me ask the gathered audience: All Peter Daniel/Brian Bell Gainclone builders: Do you experience any remarkable about your LM3875 amps? I'll gather you all miss this snubber.
Regardless of what some people think I'm listening _and_ I haven't made my mind up.
Notice also that I assume that all caps are very close to the LM3886, like in my case. We are talking about one inch or two for the big caps and directly at the pins for the smaller caps.
Holes are drilled underneath the middle section of fins.
Dosn,t show on the pic though. The topplate is made from
"holeplate" not sure if it is the right word in english, but I am sure
you get the meaning. Two red ultrabright LED's are fitted inside
the chassis, so they are redhot when playing, but not because of the heat🙂 Looks kinda cool, glowing form the inside
Steen.
steenoe said:
now I understand 🙂
vertical heat sink fins (as in your design) are better than horizontal ones == at least if you are going to rely on ambient movement of air.
i have run my bridged 4780 amp > 120 watts when the heat sink is cooled with a fan, btw.
percy said:
The portuguese king at the time refused his proposal to make that trip with portuguese chips.
The reason was we already knew the existence of Brazil, but kept it secret, because of the negotiations with the spanish.
Columbus traveled with spanish flag, chips, and support.
The objective was to find India, traveling to the opposite direction.
There was America in the middle...😱
Anyway, these trips at the time were something like dreaming in landing on the moon in the 1950's. Risky business...
percy said:
Yes.
Before Columbus, no?😀
Percy, this is world history, but also the history of my country.😎
Back to audio...
peranders said:
For the 1000th time:
You can put the big caps on the chip's pins, bypassed or not.
Without the snubber, the sound is no better than a cheap Rotel amp.
And I wouldn't loose my time and money if it was to make a "supermarket" amp.
Upupa Epops said:
Is the manufacturer of a component the one who knows how to use it better?
Does Toshiba make good amps just because they make some good transistors?
Does Panasonic (Matsushita) make good amps or cdps just because they make some "good" caps?
I see... you read the datasheet of an LM3886 and you convince yourself that the best you can do with it is written in there.
Stick to what's there, it works.
Just...
To Carlos : Look at my schematic - there is one thing, which you don't have in your PS. Try to think why is there.
jeff mai said:
I understand what you are saying, but I didn't say there can be no difference. I have made my ideas clear a zillion times, but I understand that a single post in isolation is just that. My strong feelings are that a simple, sighted listening test by someone who is obviously involved in the results (like the guy who made the mod in the first place) is totally unreliable. I don't care whether his results or reports happen to be in line with what I think it should be or not. The result, in this discussion, is irrelevant. The point is that the result is totally unreliable. It is unreliable for the party reading it as well as for the party reporting it.
Jan Didden
- Status
- Not open for further replies.