Clay idols
Even if I just raised this same question in another thread, I think it belongs here. It started with the muting function but I think it goes deeper.
The LM3886 has a muting function, as well as the LM3876. I have read many claims that this muting circuit is bad for the rest of the circuit. But I don't see quite why yet. Can anybody risk an opinion?
Halojoy has also raised the question that he doesn't quite believe that an inverted gainclone will necessarily sound better than a non-inverted, and I think I am with him.
We seem to be rejecting certain opinions from objectivists that raise measurements as THE LAW to be worshipped, but we are creating new idols apparently, with no tangible proof they do exist indeed.
About the inverted vs non-inverted question I've read reports from DIYers that built both and it was really a matter of preference, not audio quality.
And I'm afraid this muting option (which can be practical if used) became a new idol.
And idols are not something I really like...
Carlos
Even if I just raised this same question in another thread, I think it belongs here. It started with the muting function but I think it goes deeper.
The LM3886 has a muting function, as well as the LM3876. I have read many claims that this muting circuit is bad for the rest of the circuit. But I don't see quite why yet. Can anybody risk an opinion?
Halojoy has also raised the question that he doesn't quite believe that an inverted gainclone will necessarily sound better than a non-inverted, and I think I am with him.
We seem to be rejecting certain opinions from objectivists that raise measurements as THE LAW to be worshipped, but we are creating new idols apparently, with no tangible proof they do exist indeed.
About the inverted vs non-inverted question I've read reports from DIYers that built both and it was really a matter of preference, not audio quality.
And I'm afraid this muting option (which can be practical if used) became a new idol.
And idols are not something I really like...
Carlos
Re: Clay idols
I suggest, that a best solution to your problem is to built both circuits and then decide which one is better, instead of playing with words.
carlmart said:Even if I just raised this same question in another thread, I think it belongs here. It started with the muting function but I think it goes deeper.
The LM3886 has a muting function, as well as the LM3876. I have read many claims that this muting circuit is bad for the rest of the circuit. But I don't see quite why yet. Can anybody risk an opinion?
Halojoy has also raised the question that he doesn't quite believe that an inverted gainclone will necessarily sound better than a non-inverted, and I think I am with him.
We seem to be rejecting certain opinions from objectivists that raise measurements as THE LAW to be worshipped, but we are creating new idols apparently, with no tangible proof they do exist indeed.
About the inverted vs non-inverted question I've read reports from DIYers that built both and it was really a matter of preference, not audio quality.
And I'm afraid this muting option (which can be practical if used) became a new idol.
And idols are not something I really like...
Carlos
I suggest, that a best solution to your problem is to built both circuits and then decide which one is better, instead of playing with words.
pinkmouse said:
I am using the MBR20100 schottkys from OnSemi parralleled up, and I like them,(although I have not tried the MUR devices yet, I have tried most others). Like Fedde stated, they produced an amp that was very hard to stop listening to, you always want to just try another track to see how it sounds
Did you compare single diodes and paralleled? I had only 8 of them, so I couldn't use them in parallel, but if it makes a difference I'll try as well.
Can I use this diode to build gainclone?
http://www.cpu.com.tw/kh/m/sd/st-1.gif
It's rating 5A 200V 30ns.
I want to build with LM3875 based on Peter's schem.
How to use this? It looks like two diodes inside.
Using only one ? (from A1 to K)
any other suggestion ?
thanks..
http://www.cpu.com.tw/kh/m/sd/st-1.gif
It's rating 5A 200V 30ns.
I want to build with LM3875 based on Peter's schem.
How to use this? It looks like two diodes inside.
Using only one ? (from A1 to K)
any other suggestion ?
thanks..
Peter Daniel said:
I suggest, that a best solution to your problem is to built both circuits and then decide which one is better, instead of playing with words.
Well, I certainly don't think I'm playing with words. And I also don't think it's my problem.
What I'm raising here is an attitude, an issue. I will certainly do what I think should be tried on the pcbs I am now designing and report my findings here. But the more of us that try different alternatives instead of "playing safe" the more we may get to "rules" that may apply, or at least what we can expect.
For instance and to illustrate my point, if the LM3875 is "better" than the 3886, why is it that Jeff Rowland used the "inferior" 3886 on his highly regarded big buck amps?
Carlos
Speaking for myself, I never said that 3875 is better or inverted config is better. It's just what I tried first and I simply like the results. I'm trying to improve on simple issues like removing the unnecessary parts and choosing the ones that stay to my sonic preferences. And that's all. If somebody can gain on my experience here, I don't look at this as emerging of a trend or creating idols.
This is a simple circuit and there is not too many options that can be tried, unless you want to bridge them or use elaborate balanced setups. It is mostly about fun and satisfaction with what can be achieved and learned in a process.
Your suggestions and subsequent doubts about this or that setup being better is simply spoiling our fun.
If you have doubts and you suspect that something else is better, built it and then share your experiences and subjective impressions, instead of polemizing on a subject without trying much first.
This is a simple circuit and there is not too many options that can be tried, unless you want to bridge them or use elaborate balanced setups. It is mostly about fun and satisfaction with what can be achieved and learned in a process.
Your suggestions and subsequent doubts about this or that setup being better is simply spoiling our fun.
If you have doubts and you suspect that something else is better, built it and then share your experiences and subjective impressions, instead of polemizing on a subject without trying much first.
UrSv said:
I agree with what you're saying -- but in a 50/60 Hz rectifier application, these problems typically do not occur. This is what plain old boring rectifiers are made for. It seems unlikely to me that they'd design a $2.20/1K qty part to be optimal sounding only with the use of super fast recovery diodes.
A plain jane diode has a recovery time of around 2uS. If the input signal being rectified is 60 Hz, that's over 16,000 times the speed that you need for full wave recovery time.
Just out of curiosity, I went and wired up a bridge this morning out of 1N4007s. I looked at them on the scope both with and without bypass. In neither case did I observe what looked like any significant amount of ringing.
Fair enough. The point that I'm trying to make is that I don't see where the ultra fast diodes are going to make a difference here. It's not a switchmode supply running at 100KHz+. To me, it seems like debating about the virtues of putting in 101 octane in a 1977 Grenada.
Load is significant.
schmad,
Under what conditions did you perform your test?
I would tend to believe that your observations are correct when the power supply for an amplifier is properly designed--neither too little nor too much capacitance. Or in a stituation such as a class A amplifier that draws a constant current from its power supply.
However, with amplifiers that have dynamically changing current requirements, the significance of fast diodes or at least RC snubber circuitry can not be overlooked. Here is a link if you haven't seen the article already; Calculating Optimum Snubbers. At least it is some interesting reading and full of objective analysis of which we need more. Don't get me wrong, I use objective analysis as a starting point and then let my ears perform the subjective testing after that. Being objective is always a good starting point.
schmad,
Under what conditions did you perform your test?
I would tend to believe that your observations are correct when the power supply for an amplifier is properly designed--neither too little nor too much capacitance. Or in a stituation such as a class A amplifier that draws a constant current from its power supply.
However, with amplifiers that have dynamically changing current requirements, the significance of fast diodes or at least RC snubber circuitry can not be overlooked. Here is a link if you haven't seen the article already; Calculating Optimum Snubbers. At least it is some interesting reading and full of objective analysis of which we need more. Don't get me wrong, I use objective analysis as a starting point and then let my ears perform the subjective testing after that. Being objective is always a good starting point.
Diodes
Hi I'm playng with gain-cloning for one year now.
Ordinary bridge doesn't sound very good for me.
I'm using 2x1000uF Fa (Fc) Panasonic. I've tried also some ordinary 'noname' caps as Hitano or Jamicon: tends to be forward sounding (maybe a bit agresive). Coupling caps are 1uF polycarbon (better than polypro wima or polyester wima).
I'v tried doble schottky bridge: using SR390, SB5100, and paralleled
MBR20100. The sound have less presence, this may be an advantage with forward sounding speakers.
What I like the best is ultra fast recovery BYV28-200 and in the single bridge (4 diodes only for whole amp).
Bigger BYV29-200 are not as good as 28-200.
I can also confirm that diodes have the very long break-in time (~1000 hours).
v.
Hi I'm playng with gain-cloning for one year now.
Ordinary bridge doesn't sound very good for me.
I'm using 2x1000uF Fa (Fc) Panasonic. I've tried also some ordinary 'noname' caps as Hitano or Jamicon: tends to be forward sounding (maybe a bit agresive). Coupling caps are 1uF polycarbon (better than polypro wima or polyester wima).
I'v tried doble schottky bridge: using SR390, SB5100, and paralleled
MBR20100. The sound have less presence, this may be an advantage with forward sounding speakers.
What I like the best is ultra fast recovery BYV28-200 and in the single bridge (4 diodes only for whole amp).
Bigger BYV29-200 are not as good as 28-200.
I can also confirm that diodes have the very long break-in time (~1000 hours).
v.
Re: Load is significant.
It was a pretty simple setup:
+6/CT/-6 transformer into the rectifiers, no snubbers.
.01uF, .1uF, 470uF caps off of both rails. 470uF is an electrolytic, .01/.1uFs are polyprops.
Measurements were taken against a resistive load of 100 ohms using a Tektronix TDS 3032.
I was looking at the output of the diodes directly and then the rails with the added capacitors.
Specifically, I was looking for any sort of noise or extraneous harmonics.
Now since reading your post, I decided to gave have a look at a more "real world" application. I tested an amplifier that I'd built (a version of Randy Slone's OptiMOS amplifier) that can do upto ~800W/chan with a 4 ohm nominal load. The power supply begins with a 1500VA 60-0-60 transformer, two plain old 400V/35A full bridges with snubbing caps, 6 160V 22,000uF caps,
2 .1uFs, and 2 .01uFs.
I ran the amp, monitoring the supply rails with the scope. They stayed pretty much consistent at ~85V. I didn't notice any sort of harmonics making their way through.
I would agree with you here. If you are having dropout problems going all the way back to the rectifiers as a result of dynamic power requirements, you probably don't have enough capacitence in the supply rails.
Again, given that you're rectifying a 50/60 Hz wave, I don't see how rectifiers that are already 16,000 times faster than needed are going to be the source of the problem.
I did see that, but thanks for the reminder.
I agree entirely. Further to that point, I'm not trying to be beligerent here, just raising some discussion on the topic of objective analysis.
I'm with you here again. When I sit down to design something, I like to have a very good idea of what to expect based off of modeling and theory. If I make a change to a circuit, I like to have some idea of what to expect. If I observe something other than what I expect, I like to be able to understand why.
stadams said:
It was a pretty simple setup:
+6/CT/-6 transformer into the rectifiers, no snubbers.
.01uF, .1uF, 470uF caps off of both rails. 470uF is an electrolytic, .01/.1uFs are polyprops.
Measurements were taken against a resistive load of 100 ohms using a Tektronix TDS 3032.
I was looking at the output of the diodes directly and then the rails with the added capacitors.
Specifically, I was looking for any sort of noise or extraneous harmonics.
Now since reading your post, I decided to gave have a look at a more "real world" application. I tested an amplifier that I'd built (a version of Randy Slone's OptiMOS amplifier) that can do upto ~800W/chan with a 4 ohm nominal load. The power supply begins with a 1500VA 60-0-60 transformer, two plain old 400V/35A full bridges with snubbing caps, 6 160V 22,000uF caps,
2 .1uFs, and 2 .01uFs.
I ran the amp, monitoring the supply rails with the scope. They stayed pretty much consistent at ~85V. I didn't notice any sort of harmonics making their way through.
I would agree with you here. If you are having dropout problems going all the way back to the rectifiers as a result of dynamic power requirements, you probably don't have enough capacitence in the supply rails.
Again, given that you're rectifying a 50/60 Hz wave, I don't see how rectifiers that are already 16,000 times faster than needed are going to be the source of the problem.
I did see that, but thanks for the reminder.
I agree entirely. Further to that point, I'm not trying to be beligerent here, just raising some discussion on the topic of objective analysis.
I'm with you here again. When I sit down to design something, I like to have a very good idea of what to expect based off of modeling and theory. If I make a change to a circuit, I like to have some idea of what to expect. If I observe something other than what I expect, I like to be able to understand why.
Peter Daniel said:
I started off using one MBR20100 and two MBR10100 per side, as can be seen in this pic of an earlier iteration, but when I had to redesign the PSU when I moved the internals around in the case I had no MBR10100s left so I used the MBR20100s for the whole thing, and if it was the moving around of the internals or what, but the bass just seemed a little more extended- just a little, but noticeably different
Rookie said:Peter,
After so many tweaks you did, have you cosidered to try another op-amp? I think this is the only thing you didn't try. Maybe the LM3875 is not the ultimate choice...
Yeah, Peter, sometimes you surprise me -you are willing to spend hours testing caps ,resistors,directivity,diodes,wires etc and you don't audition other versions of the most crucial active part of the gainclone.I was thinking of sending you some OPA 549's just in the hope you would turn your head away from those voluptuous 3875's.
protos said:
Just tell me if it's available as samples and I'll get it. Some people say, they had good experiences with TDA7294. But also people said that attaching resistor from output to one rail improves the sound, but it didn't IMO.
If you have a good thing working, why looking for something else, which doesn't guarantee success?
Why am I testing all those passive components? Because whatever great chip you might have, you will never get a great amp, without taking care of those passive parts. There are only few of them, so don't worry.
Peter Daniel said:
If you have a good think working, why looking for something else, which doesn't guarantee success?
I cannot believe you mean this.
First of all, if people on this forum thought they had a good thing working and didn't think they should look for something else then they wouldn't be on this forum.
Secondly, even if you have a good thing working there is always a possibility that there might be something better. Theres no harm in researching other possibilities as long as you don't have to replace your "good thing" in doing so.
Thirdly, this statement does not seem to represent your (and most member's here) essentially inquisitive nature.
As for samples TI offers them but I am surprised you will "only" consider samples since non-availability of samples does not seem to have discouraged you at all in the past.
Furthermore I agree with you that passive components may make or break an excellent sounding amp but we should not exaggerate their effect either compared to a well engineered amp with resonably good but not hi-end resistors and caps.I think most people here would agree that circuit designs and topologies make a huge difference , probably more than passives.
pinkmouse said:
Like Fedde stated, they produced an amp that was very hard to stop listening to, you always want to just try another track to see how it sounds
That's exactly what I meant. And even more after a tweak. I can't stop until I have listened tens of well known tracks! Most of the time I use my pc now after tweaks, I have a lot of test mp3/waves on it. Though of course, cd's sound better on my modded Philips CD960 transport.
Another problem I am familiar with! Expensive hobby too...
Next project for me is to make a nice buffer... I have some ideas, let's boot the p-spice process. (bipolars, J-FETs, tubes).
I think that my Thor-amp has two important problems:
1. Lack of stability. I have to turn off my amp every 20-40 min. to stop oscillation. Hope to fix that with a 50 k att. Otherwise I have to improve the power supply filtering or so...
2. Very low input impedance at higher volume settings. Will be fixed by a buffer.
Fedde
protos said:
There are people on this forum who never built anything, yet they find pleasure in just talking about theoretical designs. I don't see nothing wrong with that.
For me the endless search for best op amp would prevent me from endevours into other audio projects. I have a CD transport design on hold as well non oversampling DAC, not to mention speakers. I just can't spent any more time pursuing amp design. The one I have right now is more than satysfying and this is where my search ends. You better believe it.
Since an op amp is my commercial product now, I think that I honestly qualify for a samples program now.
protos said:
Sure, of course the circuit is more important than the quality of the passive components. But sometimes you have to restrict some of your options, otherwise you will get crazy
=> 
I think that Peter would have to start over again if he would choose an other opamp. Likely, the character of the amp would have to be balanced with other components. So, I think that you have to make a choice for a component in the beginning to reduce the degrees of freedom. Though it's a good thing to continue to explore other options lateron.
Fedde
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.