Hi,
First, I said 4KW peaks.
Second, in the 1980's I build a ridiculously powerful Amplifier in a kind of borg cube design. Actually, several.
It had two huge forced air cooling tunnels with massive industrial darlington transistors on them filling up the upper section. The fans driving air through them where marginally quiter than a jet at takeoff (okay, I am joking, but they sure where noisy).
Powersupply was using three huge UI transformers and the power-input was three phase of course. Many big PSU Cap's as well.
I never really found a load that allowed me to sinewave test this design, so I do not known how much power it could put out.
It was a Circlotron type circuit (all NPN - not a single PNP Transistor) with 100V output rails, so around 450W RMS into a single 8 Ohm Driver of a large group.
Even with 12 pcs of 18" EVM Pro-Line drivers hanging off the end of these I did not get any "early clipping" indication (the Amp's had both level meters and true clipping indicators), suggesting undistorted peaks of over 5KW with music (around 10dB crest factor). I suspect that in sinewave operation I could have gotten a bit under 4KW, as each transformer was rated for around 2KW.
Even with softstart and timed sequencing between individual amp's, on some stages that did not have 3-Phase power the only way to power up the Amp-stact was to replace the ceramic fuse inserts in the stage fusebox with solid turned brass and if necessary even those in the breaker panel further down the line.
I eventually changed the softstart to phase cutting (phasenanschnitt steuerung), so even on poor power systems they could ramp up real soft and slow, without blowing fuses or tripping breakers.
Ciao T
First, I said 4KW peaks.
Second, in the 1980's I build a ridiculously powerful Amplifier in a kind of borg cube design. Actually, several.
It had two huge forced air cooling tunnels with massive industrial darlington transistors on them filling up the upper section. The fans driving air through them where marginally quiter than a jet at takeoff (okay, I am joking, but they sure where noisy).
Powersupply was using three huge UI transformers and the power-input was three phase of course. Many big PSU Cap's as well.
I never really found a load that allowed me to sinewave test this design, so I do not known how much power it could put out.
It was a Circlotron type circuit (all NPN - not a single PNP Transistor) with 100V output rails, so around 450W RMS into a single 8 Ohm Driver of a large group.
Even with 12 pcs of 18" EVM Pro-Line drivers hanging off the end of these I did not get any "early clipping" indication (the Amp's had both level meters and true clipping indicators), suggesting undistorted peaks of over 5KW with music (around 10dB crest factor). I suspect that in sinewave operation I could have gotten a bit under 4KW, as each transformer was rated for around 2KW.
Even with softstart and timed sequencing between individual amp's, on some stages that did not have 3-Phase power the only way to power up the Amp-stact was to replace the ceramic fuse inserts in the stage fusebox with solid turned brass and if necessary even those in the breaker panel further down the line.
I eventually changed the softstart to phase cutting (phasenanschnitt steuerung), so even on poor power systems they could ramp up real soft and slow, without blowing fuses or tripping breakers.
Ciao T
Have you ever built one before , any multi chip design ? How does a parallel setup sound vs a single ?
There is a sonic character to these chip amps , no different from other genre i guess, for me there is very much a hardness in the top and a lack of good bass. If running on limited bandwidth speakers i guess the sound could be as good as a good discrete amplifier.
Interesting to hear how a multi chip design compares to the more popular
daniels single chip setup for eg...
Sounds like a "Hill" amplifier design , never liked the sound of the bass on the hill amps from the 80's . Crest did a much better job back then ...IMO
The Focal/JM-Lab SM11 pro-audio monitor use a pair of bridged LM3886 on the tweeter only, and they sound absolutely phenomenal with silk, 3D and real life high frequency response. I hope that helps
SM11 line. Focal Professional.
With kind regards,
Bas
Thank you all for your input, but may I ask you to stick to the original question.
One thing I never intended to do is bridging and/or paralleling. From my understanding the only benefit from this is doubling the power/current capacity (of which a single chip has plenty enough at normal listening levels) at the expense of added complexity, worse symmetry and higher cost. I don't like it. I want pure, not loud.
As for the sonic signature of chip amps, lack of bass etc. - I have found that the pre section is vitally important and also the capabilities of the source. The bass issue in particular became simply non-existent after I've installed a decent op-amp in my CD player's output stage.
@ lanchile - I haven't tried big power supply caps myself. I decided that if Peter Daniel says it sounds worse this way, he must have reasons to say so.
One thing I never intended to do is bridging and/or paralleling. From my understanding the only benefit from this is doubling the power/current capacity (of which a single chip has plenty enough at normal listening levels) at the expense of added complexity, worse symmetry and higher cost. I don't like it. I want pure, not loud.
As for the sonic signature of chip amps, lack of bass etc. - I have found that the pre section is vitally important and also the capabilities of the source. The bass issue in particular became simply non-existent after I've installed a decent op-amp in my CD player's output stage.
@ lanchile - I haven't tried big power supply caps myself. I decided that if Peter Daniel says it sounds worse this way, he must have reasons to say so.
Last edited:
But paralleling chips isn't only for getting louder, but you can decrease the distortion significant also with average listening levels. Like Mr. Self point out in his book, that adding more pairs of transistors reduce distortion, the same is truth with the chips.
Personally I like the Burr Brown OPA549 very much but that is a subjective personal preference. I like it especially in a nested feedback-loop with a good input opamp (OPA627)
With kind regards,
Bas
Hi,
Yes.
If done right and if away from clipping, protection circuitry cutting in etc. the same.
That is not the results I get, even with single chip designs. Clearly I'm doing something wrong.
I have not been using speakers much more limited in bandwidth than most, all with sub 30Hz LF response and using ribbon HF going up to around 40KHz.
Ciao T
Yes.
If done right and if away from clipping, protection circuitry cutting in etc. the same.
That is not the results I get, even with single chip designs. Clearly I'm doing something wrong.
I have not been using speakers much more limited in bandwidth than most, all with sub 30Hz LF response and using ribbon HF going up to around 40KHz.
Ciao T
Hi,
First, it is nothing like Hill.
No-one in those days in the west still used as NPN Amplifiers or AC coupled circlotrons. In the east we used this kind of stuff because we were much more limited in what we could get in semi-conductors.
So you actually have no idea what the Amplifiers I build where like, sounded like or where capable off.
Ciao T
First, it is nothing like Hill.
No-one in those days in the west still used as NPN Amplifiers or AC coupled circlotrons. In the east we used this kind of stuff because we were much more limited in what we could get in semi-conductors.
So you actually have no idea what the Amplifiers I build where like, sounded like or where capable off.
Ciao T
high current peaks can happen with certain waveforms, even with BiAmping single drivers - basically "pump" the resonance and try to reverse the phase and the amp will need to supply up to 2.4x the nominal resistive rating current peak
paralleling also enables Class A output bias if wanted - or few 100mA AB
http://www.diyaudio.com/forums/chip-amps/163385-so-just-how-good-can-chip-amp-4.html#post2134459
Self shows measured improvements in distortion over his "optimum bias" AB with his new "Class-XD" added current bias - likely the power chip amps are similar in output bias level to his Blameless
a power output ccs could be done with LM317 to keep integration level high/parts count low
Last edited:
Guys, can you please go and make yourself a separate thread for pointless discussions? I was really expecting more from this forum... So far the only helpful answers I got were from Sebastiaan. [edit: and now also jcx]
@Sebastiaan - judging by the posts I've read so far the difference between LM3875 and OPA549 is very small and may well come down to the matter of personal taste. And LM3875 has the distinct advantage of being offered in Peter's fantastic kit.
I'm not sure I would want to go down the nesting alley either... Mainly because I lack the necessary knowledge, but also because it again sounds like unnecessary complication of the design. But I may well be wrong - I would love to hear opinions of people who actually did have the chance to compare single bi-amped vs bridged vs parallel.
@ jcx - Thanks, that's very interesting, I'll give it a thorough read.
@Sebastiaan - judging by the posts I've read so far the difference between LM3875 and OPA549 is very small and may well come down to the matter of personal taste. And LM3875 has the distinct advantage of being offered in Peter's fantastic kit.
I'm not sure I would want to go down the nesting alley either... Mainly because I lack the necessary knowledge, but also because it again sounds like unnecessary complication of the design. But I may well be wrong - I would love to hear opinions of people who actually did have the chance to compare single bi-amped vs bridged vs parallel.
@ jcx - Thanks, that's very interesting, I'll give it a thorough read.
Is there anything to be gained in running LM38xx balanced? I mean, if you have a balanced source you run the hot/non-inverted and cold/inverted signals through two identical (say non-inverted) lm38xx power amps with the speaker connected across their outputs instead of between output and ground as in a conventional application.
Is this effectively the same thing as bridging?
Is this effectively the same thing as bridging?
Last edited:
Stick to the original question, indeed. I answered your original question, and you dismissed it as too complex. I have built several of these amps with different cost levels of parts. Complete kits with selected premium parts are for sale right now, and they are easy to build. No adjustments are necessary. The power supply and a muting circuit is right on the PCB. Solder, add a transformer, and play.
These are the quietest of all the amps I've owned, including Class D designs. They provide more spaciousness, inner detail, and dynamics than some expensive discrete designs (such as Aragon, Classe, ARC), without any harshness or grain. They are efficient. The design responds very well to using better parts, especially the input cap.
Ignore my suggestion if you will, but it's your own loss. Please continue with putting silver wires between pieces of wood. Talk about clumsy design!
Peace,
Tom E
These are the quietest of all the amps I've owned, including Class D designs. They provide more spaciousness, inner detail, and dynamics than some expensive discrete designs (such as Aragon, Classe, ARC), without any harshness or grain. They are efficient. The design responds very well to using better parts, especially the input cap.
Ignore my suggestion if you will, but it's your own loss. Please continue with putting silver wires between pieces of wood. Talk about clumsy design!
Peace,
Tom E
Dear Leon,
I am sure we can all help you with Nested. I can try to make some simulation tonight and post it here as help. What I like more from the OPA549 in comparison with the LM3875, is the smoother highs and the more in depth placement of instruments. Of course depending on the implementation. I would't run it on to high power supplies. In my opinion the single chip sounds better on lower voltage, (25 volt the absolute max, but I found it sound smoother on +/- 20 volt with the downside you loose of course power).
What I like about nested is, that you increase the output impedance and gives far better control over the low-end. I come back later today with some suggestion simulated. Maybe you feel you are not skilled enough, but we will help you and sometimes you just have to jump into the deep water
With kind regards,
Bas
The benefit is you don't refer with the negative input to a maybe dirty ground if you feed it with a truth balanced source. This can make a difference. I prefer to feed them balanced.
With kind regards,
Bas
Is there anything to be gained in running LM38xx balanced? I mean, if you have a balanced source you run the hot/non-inverted and cold/inverted signals through two identical (say non-inverted) lm38xx power amps with the speaker connected across their outputs instead of between output and ground as in a conventional application.
Yes, I agree, and balanced power amps seem to be widely used for high end headphone amps. Odd they're not that common for speaker amps. Must be the cost I guess.
My questions remains though, is this electrically the same as bridging from an unbalanced source?
Yes, I agree, and balanced power amps seem to be widely used for high end headphone amps. Odd they're not that common for speaker amps. Must be the cost I guess.
My questions remains though, is this electrically the same as bridging from an unbalanced source?
Last edited:
That depends. If you bridge the chips direct from an unbalanced source, one chip is inverted and the other is non-inverted, still each chip has one leg with reference to ground. If you feed both chips balanced but each in opposite phase (to make it bridged) you don't have that problem. One of my secret's to get such a silent chip amp designs, was use them always balanced with no leg connected to ground.
With kind regards,
Bas
OK, I've read a bit about MyRef amp, and it does indeed look interesting. Way more complex than Peter Daniel's kit though. Can anyone point me to a picture of a finished MyRef amp and its PCB? It may sound silly, but I want the finished amp to look good too, not just sound good.
@Gopher - I have nothing against your questions (and would gladly answer them if I could), but I would like to point out that it is a simple courtesy not to hijack other people's threads.
@Gopher - I have nothing against your questions (and would gladly answer them if I could), but I would like to point out that it is a simple courtesy not to hijack other people's threads.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.