In preparation to replace my excellent B&K 200.2 amplifier powering my two Martin Logan ReQuest left/right speakers, with a DIY amp that has been in progress for a while (just for fun/bragging rights), I decided to measure the B&K's output swing across a speaker in various movie and musical passages, and see how big of a peak I might run across.
I hooked up my Rigol oscilloscope to the speaker +/- terminals, after checking the black terminal's resistance to ground to ensure I wasn't doing something bad by hooking up the O-scope's ground terminal (appears it was either floating or close enough to ground not to matter).
First I just turned on the TV and watched a movie, with the scope set to measure VV-max and VV-top, and a cursor set at +/- 32V rails, my originally planned amp rails. Even though it was a loud action movie, I never saw the signal go above +/- 18V. No problem!
Second, I decided to try a blue ray. I selected The Hobbit 1, because the beginning with the Dragon attack (spoilers!) has some very loud and impressive audio sequences. Then I really cranked up the volume, louder than I typically make it, but at a level I have heard before, many times for parties or demonstrations to friends.
Unfortunately I started seeing some huge peaks during this sequence, one of which was about 100Vpeak to peak! I managed to capture on screen a +/- 49V transient.
Worse, the Martin Logan reQuests have a very low impedance at high frequencies, nearing 1 Ohm.
So, needless to say I'm now having 2nd thoughts if I should follow my original plan to swap from the B&K to a DIY amp, especially since my original plan was to use 2 or more LM3886s-based amps in parallel. I was thinking 2 or more LM3886 should be able to handle the current, plus I could remove the shunt between the tweeter and panel, and amp each separately to reduce current further... but now I'm thinking even if I can source enough current, there will be some slight clipping at very high listening levels.
Any advice? Any chance since I'm using a digital O-scope with edge triggering it's somehow lying to me/I've measured it incorrectly? Is clipping at those levels going to be audible or missing the edge of that peak actually won't matter much?
(BTW for the home theater I also have 6 other speakers powered by a Proceed AMP5)
I hooked up my Rigol oscilloscope to the speaker +/- terminals, after checking the black terminal's resistance to ground to ensure I wasn't doing something bad by hooking up the O-scope's ground terminal (appears it was either floating or close enough to ground not to matter).
First I just turned on the TV and watched a movie, with the scope set to measure VV-max and VV-top, and a cursor set at +/- 32V rails, my originally planned amp rails. Even though it was a loud action movie, I never saw the signal go above +/- 18V. No problem!
Second, I decided to try a blue ray. I selected The Hobbit 1, because the beginning with the Dragon attack (spoilers!) has some very loud and impressive audio sequences. Then I really cranked up the volume, louder than I typically make it, but at a level I have heard before, many times for parties or demonstrations to friends.
Unfortunately I started seeing some huge peaks during this sequence, one of which was about 100Vpeak to peak! I managed to capture on screen a +/- 49V transient.
Worse, the Martin Logan reQuests have a very low impedance at high frequencies, nearing 1 Ohm.
So, needless to say I'm now having 2nd thoughts if I should follow my original plan to swap from the B&K to a DIY amp, especially since my original plan was to use 2 or more LM3886s-based amps in parallel. I was thinking 2 or more LM3886 should be able to handle the current, plus I could remove the shunt between the tweeter and panel, and amp each separately to reduce current further... but now I'm thinking even if I can source enough current, there will be some slight clipping at very high listening levels.
Any advice? Any chance since I'm using a digital O-scope with edge triggering it's somehow lying to me/I've measured it incorrectly? Is clipping at those levels going to be audible or missing the edge of that peak actually won't matter much?
(BTW for the home theater I also have 6 other speakers powered by a Proceed AMP5)
Yes, fine for the B&K. But not really my question, which was about switching to an LM3886 based amp.
Loud but not unpleasant. My room is pretty over the top, and big. 35x20 non rectangular, with ~40% upholstered walls, stadium seating, stage, etc. mix of absorbtive and diffraction treatment, staggered stud construction, blown in acoustic foam on all sides 1 foot thick etc.
Loud but not unpleasant. My room is pretty over the top, and big. 35x20 non rectangular, with ~40% upholstered walls, stadium seating, stage, etc. mix of absorbtive and diffraction treatment, staggered stud construction, blown in acoustic foam on all sides 1 foot thick etc.
LM3886 will not give you enough swing at the output.
You mentioned paralleling them for higher current capability. Having constant R load - in order to make the load demand more current - you need to swing higher voltage (Ohm's law
).
If you'd like to be able to drive kind of the same 100Vpp peaks into your speakers, you need to have at least +/-60V rails in your DIY amplifier. And enough output current capability for providing at least 200W per channel with no artifacts.
You mentioned paralleling them for higher current capability. Having constant R load - in order to make the load demand more current - you need to swing higher voltage (Ohm's law
).If you'd like to be able to drive kind of the same 100Vpp peaks into your speakers, you need to have at least +/-60V rails in your DIY amplifier. And enough output current capability for providing at least 200W per channel with no artifacts.
> B&K 200.2 amplifier
That's 350W in 4 Ohms.
It would take a silly stack of 68W LM3886 to equal it.
> My room ...big. 35x20 ... ~40% upholstered walls, ...absorbtive ..treatment, ...blown in acoustic foam on all sides 1 foot thick etc.
That's a big hole to fill with sound. At a guess, it is 10X more sound-suck than a typical living room; and modern cinema sound may peak louder than many musics. So for hi-fi speakers it would be quite reasonable to supply 1,000 Watts minimum. Way past LM3886 turf.
That's 350W in 4 Ohms.
It would take a silly stack of 68W LM3886 to equal it.
> My room ...big. 35x20 ... ~40% upholstered walls, ...absorbtive ..treatment, ...blown in acoustic foam on all sides 1 foot thick etc.
That's a big hole to fill with sound. At a guess, it is 10X more sound-suck than a typical living room; and modern cinema sound may peak louder than many musics. So for hi-fi speakers it would be quite reasonable to supply 1,000 Watts minimum. Way past LM3886 turf.
I would say actually that this is not a may but a will.
I refer to the THX cinema volume levels as a reference for what you might want if you might aim for a cinema experience. It peaks at 105dB at the listening position per each front main channel. To put that in perspective the total (incl surround and LFE) system would supposedly be able to replicate the reality of a small petrol whipper snipper being used in the same room. Whether you want the whipper snipper or the sound in the room, is up to you
Now that you built your LM3886, use it.
It will play good, just not that loud.
And for your next project, build a discrete mp with +/-70V rails
Do not waste time parallelling, bridging or both, with chipamps.
The beauty of chipamps is simplicity, plus they sound good ... within their limitations.
Once you need 2 or 4 of them, you start approaching discrete amp complexity, so, what´s the point?
It will play good, just not that loud.
And for your next project, build a discrete mp with +/-70V rails
Do not waste time parallelling, bridging or both, with chipamps.
The beauty of chipamps is simplicity, plus they sound good ... within their limitations.
Once you need 2 or 4 of them, you start approaching discrete amp complexity, so, what´s the point?
BTL time
Thanks for all the help!
I guess I never really thought about (or maybe I did and forgot!) exactly how bridging an amp (especially a fully balanced one) increases max potential across the speaker, effectively doubling the rail voltage. In my case bridging would basically double the rail capability, which would be inside my (observed) test (with ~28% margin vs clipping on transients at very high volume). I was thinking (incorrectly) referenced to ground that the most any topology could do without a fast switching-mode would be a Vp limited by the rails. But, of course with BTL the "ground" (negative) side of the speaker is moving too WRT amp "ground".
I agree a discrete design would be more traditional and... maybe cheaper? I may go that route later, although what with so many assets in the Class D world now available and so much commercial effort in that direction maybe I would go that way. However I had originally planned to BTL/parallel the device anyway and I already bought a crap-ton of LM3886's (I have 8 more unused isolated and... wow just counted, um 17 non-isolated! I'm not sure where all those came from).
Thanks for all the help!
I guess I never really thought about (or maybe I did and forgot!) exactly how bridging an amp (especially a fully balanced one) increases max potential across the speaker, effectively doubling the rail voltage. In my case bridging would basically double the rail capability, which would be inside my (observed) test (with ~28% margin vs clipping on transients at very high volume). I was thinking (incorrectly) referenced to ground that the most any topology could do without a fast switching-mode would be a Vp limited by the rails. But, of course with BTL the "ground" (negative) side of the speaker is moving too WRT amp "ground".
I agree a discrete design would be more traditional and... maybe cheaper? I may go that route later, although what with so many assets in the Class D world now available and so much commercial effort in that direction maybe I would go that way. However I had originally planned to BTL/parallel the device anyway and I already bought a crap-ton of LM3886's (I have 8 more unused isolated and... wow just counted, um 17 non-isolated! I'm not sure where all those came from).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.