Not always. The 1/sqrt(2) factor applies to sine waves. For a pulse waveform the RMS value is sqrt(D), where D is the duty cycle.
Tom
Tom
Sorry, lots of mistakes, but there is a Plan B solution 😉
It can take more for minutes, definitely not for an hour or more.
It makes no sense.
This is a big conceptual error.
All Math after this makes no sense, including +/-22V rails
All of your post #5 is wrong, including Math and assumptions.
Post #7 is also wrong, all of it. Sorry.
Fine for sinewaves and "not that bad on oher waveforms, while Peak can be extremely misleading, specially on noisy/dirty signals or narrow peaks.
Averaging won´t be exact, of course, but error will be less significant.
VTVMs read peak to peak because they don´t rely on meter inertia (average) but diode rectify into capacitors (peaks).
Modern Multimeters do some kind of signal processing to get RMS,although in the old days there used to be a clever trick: mixing average and peak signals in a certain ratio, to offer very close RMS values with many different signals.
Plan B:
let´focus on what you have available in the Real World.
You have a 4 ohm 50W RMS Woofer? ... check.
An LM3886? ..... check.
Then follow datasheet suggestions:
Straight from the Neurochrome site:
https://neurochrome.com/pages/output-power
Use either:
* single LM3886, +/-28V rails for 68W into 4 ohms
or
* 2 parallel LM3886, +/-35V rails for 100W into 4 ohms.
this is your answer, look no further.
Post results and puctures when you finish your build 🙂
😳 it´s a 50W speaker!!!!
It can take more for minutes, definitely not for an hour or more.
That´s exactly what it is.
And what does that mean??????
It makes no sense.
No, not Peak to Peak, just plain peak.
This is a big conceptual error.
All Math after this makes no sense, including +/-22V rails
Maybe they have a point 😉
All of your post #5 is wrong, including Math and assumptions.
Post #7 is also wrong, all of it. Sorry.
Sorry but Analog (Galvanometer/needle) multimeters and cheap Digital ones measure Average but multiply by 1.11 to display RMS
Fine for sinewaves and "not that bad on oher waveforms, while Peak can be extremely misleading, specially on noisy/dirty signals or narrow peaks.
Averaging won´t be exact, of course, but error will be less significant.
VTVMs read peak to peak because they don´t rely on meter inertia (average) but diode rectify into capacitors (peaks).
Modern Multimeters do some kind of signal processing to get RMS,although in the old days there used to be a clever trick: mixing average and peak signals in a certain ratio, to offer very close RMS values with many different signals.
Closer but still no cigar.
Plan B:
let´focus on what you have available in the Real World.
You have a 4 ohm 50W RMS Woofer? ... check.
An LM3886? ..... check.
Then follow datasheet suggestions:
Straight from the Neurochrome site:
https://neurochrome.com/pages/output-power
Use either:
* single LM3886, +/-28V rails for 68W into 4 ohms
or
* 2 parallel LM3886, +/-35V rails for 100W into 4 ohms.
this is your answer, look no further.
Post results and puctures when you finish your build 🙂
There is a difference between peak and peak to peak you are maybe missing. Take Voltmeter RMS and multiply by 2.83 for peak to peak. Divide by two and this gives you one rail the other is the opposite polarity. For 100 Watts into 8 Ohms 28 Volts RMS times 2.83 = 79.24 divide by 2 for 39.6. With no loss you need + and - 40 Volt rails. Of course there will be a couple volt loss.
That's a good point. I was trying to find a definitive answer to how AC voltage is measured, and to my surprise, I was having a really hard time finding any reference where it's actually spelled out. If I had thought about squaring and its effect on negative numbers, that might've been a clue. But then, a lot of that is really an artifact of the inside baseball of how the meter works, rather than the standard it has to meet.
Yeah, absolutely -- noted, and aware. The goal is not to play a sine wave at 100W for indefinite periods. Merely to optimize for all of the driver's limits -- peaks that drive the speaker to its physical excursion limit, and the assumption that the crest factor of ... most music (hah) is probably still such that the long-term average will be half that power (or less).
Don't be sorry. This was entirely the point of creating this thread. There was obviously something that I didn't understand, because my calculations were at odds with the datasheet. Now, it's not unheard of that a datasheet might have errors, but given how popular this chip is, I'm fairly confident the error is on my end. Which it was. And it was indeed a conceptual error. I don't mind at all having to eat humble pie, if I come away from the experience knowing more than I did at the start.
Now, regarding advice that amounts to "just do what it says" and "here are some working designs, use those" .... ehhh, thanks, but no. 🙂 I checked the sign on the door, and it still says "DIY" Audio, yes? ;-) Seriously, though, the point of this -- and almost any project I take on -- is two-fold: 1) To get something that (hopefully) works for its intended purpose. 2) To hone all of the skills that go into that project.
I could easily buy proven PCBs and cobble together a functional system, but that's not what I want out of DIY. I want to understand what I'm doing, and I want to make something that meets my needs exactly, and preferably learn something from the process.
This one project is a potpourri of practices. Woodworking, finishing, CAD, passive XO design, active electronics design, PCB fab, plate amp panel fab (which in and of itself is a collab between PCB design, CAD, graphic design, and designing the whole thing to fit in some given 3D space.) I might even offload some of the system management stuff to a microcontroller, so that's digital electronics and programming thrown in for good measure. 🙂 Lots of skills. I'm not an expert in any of them, but it's empowering to have access and ability to dabble in all of it. And I enjoy the process.
I don't understand the hangup about how some voltage is measured.
It doesn't make any difference even when there's a tiny guy inside with a steam engine.
10V rms is 10V rms, however it is measured.
If you use a movimg coil meter, which reacts to average voltage, you correct the scale to make it show Vrms.
If you use a rectifier followed by a DC multimeter inside, you correct the scale to indicate Vrms.
What difference does it make what's inside the box?
Jan
It doesn't make any difference even when there's a tiny guy inside with a steam engine.
10V rms is 10V rms, however it is measured.
If you use a movimg coil meter, which reacts to average voltage, you correct the scale to make it show Vrms.
If you use a rectifier followed by a DC multimeter inside, you correct the scale to indicate Vrms.
What difference does it make what's inside the box?
Jan
Oh my. I thought we had cleared this up already.
It doesn't make a difference, Jan. I never cared about how the meter derives its answer. I was always concerned about what part of the waveform the measurement referred to. My hangup was analogous to "is a circle measured by diameter or by radius?" That got cleared up in post #9, and everything since then has been discussion. Mostly, me trying to explain what error I had been making (in past-tense), and people responding by explaining to me how to convert between peak and RMS.
It doesn't make a difference, Jan. I never cared about how the meter derives its answer. I was always concerned about what part of the waveform the measurement referred to. My hangup was analogous to "is a circle measured by diameter or by radius?" That got cleared up in post #9, and everything since then has been discussion. Mostly, me trying to explain what error I had been making (in past-tense), and people responding by explaining to me how to convert between peak and RMS.
For the measurement itself it isn't in the case of sine waveforms like mains. But if you are dealing with things like PWM motor drivers, then it is really important to know if the value your meter is showing is an average calibrated to sine rms or a true rms (and to what frequency and crest factor your meter is able to go).
Here however, it is important because obviously when trying to find out what mains voltage transformer you need to have a certain rms power into a given load means you have to understand how dc voltage of a rail (or positive and negative rail), peak-to-peak output swing, dc voltage after rectification and so on relate. This is where the thread starter struggled: he had an ac voltage of a transformer, didn't see how to relate that to a symmetric rail and how to get from there to rms output power.
Yes, it is easy for us who have grokked it at one point in time. But before you have it can be very confusing.
Here however, it is important because obviously when trying to find out what mains voltage transformer you need to have a certain rms power into a given load means you have to understand how dc voltage of a rail (or positive and negative rail), peak-to-peak output swing, dc voltage after rectification and so on relate. This is where the thread starter struggled: he had an ac voltage of a transformer, didn't see how to relate that to a symmetric rail and how to get from there to rms output power.
Yes, it is easy for us who have grokked it at one point in time. But before you have it can be very confusing.
So it's a matter of not understanding how to get from sine wave to RMS value to peak value.
I get that.
But has noting to do with the innner workings of a meter.
But I'll leave now.
jan
I get that.
But has noting to do with the innner workings of a meter.
But I'll leave now.
jan
No, you're relating this to two very different questions.
RMS to peak to peak-to-peak is all very well documented anywhere, and I understand that completely. (Well, kind of. I was lost on whether RMS * 1.414 netted Peak or Peak to Peak, and just how often those two were inappropriately equated in shorthand.)
How a meter works is irrelevant, meter-dependent, and frankly not that interesting to anyone but people who design meters or just like knowing how they work.
My actual question was exactly what Havoc said.
Every voltage measurement is a potential difference between two points. So, what two points on a waveform graph does an AC RMS value represent? Just saying "RMS is RMS" is like saying "an inch is an inch" while ignoring the difference between a radius and a diameter, and therefore having a potential error margin of 100%.
I think this is such a basic concept that a lot of people here are having a really hard time comprehending that any other (admittedly, incorrect) possible way of measuring could even exist. As a self-taught electronics hobbyist, I missed that day in EE school, and have mostly dealt with DC measurements where there was no potential for confusion, so I've just never experienced a situation where this mattered before. Unfortunately, a lot of even "introductory" articles on AC measurements don't actually address this at all -- they're either too simplistic to bother with the nuances, or assume you know it already -- so it's quite possible to live in ignorance for quite some time.
RMS to peak to peak-to-peak is all very well documented anywhere, and I understand that completely. (Well, kind of. I was lost on whether RMS * 1.414 netted Peak or Peak to Peak, and just how often those two were inappropriately equated in shorthand.)
How a meter works is irrelevant, meter-dependent, and frankly not that interesting to anyone but people who design meters or just like knowing how they work.
My actual question was exactly what Havoc said.
Every voltage measurement is a potential difference between two points. So, what two points on a waveform graph does an AC RMS value represent? Just saying "RMS is RMS" is like saying "an inch is an inch" while ignoring the difference between a radius and a diameter, and therefore having a potential error margin of 100%.
I think this is such a basic concept that a lot of people here are having a really hard time comprehending that any other (admittedly, incorrect) possible way of measuring could even exist. As a self-taught electronics hobbyist, I missed that day in EE school, and have mostly dealt with DC measurements where there was no potential for confusion, so I've just never experienced a situation where this mattered before. Unfortunately, a lot of even "introductory" articles on AC measurements don't actually address this at all -- they're either too simplistic to bother with the nuances, or assume you know it already -- so it's quite possible to live in ignorance for quite some time.
Last edited:
An RMS measurement DOES NOT represent the difference between two points. It may be the same value as the amplitude at 45°, but RMS is root-MEAN- square, ie it is an AVERAGE over time, typically the whole repeated wave, whatever shape it is. The use of 0.71 * peak is a cheap hack because it is only valid for a pure sine wave and finding the actual RMS value is not easy,
None! That's the whole point! It gives the RMS value of that waveform!
The RMS value is the value of a DC voltage that generates the same heat in a load as the original waveform does.
That's the definition.
A meter is generally calibrated assuming the waveform is a sine wave. If it isn't, the meter indication is in error.
But the definition is still the same.
Edit - I see @steveu made the same point.
Jan
You know... you two are completely correct. But you can be correct and also miss the point entirely.
Here's the problem, guys: I tried abandoning RMS earlier, and just saying "hey let's pretend that meters just measure peak values, or that our waveform is a square wave." Because RMS is a whole ball of wax that really doesn't matter here, and it just gets in the way. In that case, my question would've just been as simple as, "does a meter show you 0V to Peak, or Peak to Peak?" And then, I understood that you could multiply that by 0.707 (assuming sine, symmetric, yada yada), and be back to the technically correct RMS figure. But that didn't help either.
I give up. It doesn't matter. I figured out the problem a page ago. The rest of this has been rehashing a long series of misunderstandings, and it doesn't seem to be getting any better, so I can't determine what point there is in continuing.
Here's the problem, guys: I tried abandoning RMS earlier, and just saying "hey let's pretend that meters just measure peak values, or that our waveform is a square wave." Because RMS is a whole ball of wax that really doesn't matter here, and it just gets in the way. In that case, my question would've just been as simple as, "does a meter show you 0V to Peak, or Peak to Peak?" And then, I understood that you could multiply that by 0.707 (assuming sine, symmetric, yada yada), and be back to the technically correct RMS figure. But that didn't help either.
I give up. It doesn't matter. I figured out the problem a page ago. The rest of this has been rehashing a long series of misunderstandings, and it doesn't seem to be getting any better, so I can't determine what point there is in continuing.
I don't know why so much confusion. Everything is on the LM3886 datasheet, also you can read articles of neurochrome website. Everything is there. I made my first lm3886 without any problem by following everything stated above. I did some modifications & choose every part convenient to my purchased pcb's. The amplifier running well past 1 year or so without any oscillations or such thing. Traditionally chipamps are quite straightforward, you just need to implement it properly. Anyway if you want to play with oscilloscope for vanishing THD or last drop of so called "RMS" you need to know the math well.
Best Regards
Best Regards
Last edited:
Jeez, this is like pulling teeth. A newbie arguing with notable audio designers! LOL.
Let's just cut to the chase. A 3886 will work on transformer voltages anywhere up to and including 30V-0-30V, giving rails of +/- 44V and it will drive even sub 4R speakers without much trouble. (I know because I have one here.) It won't quite make the 100W mark into 8R as the rails will have sagged by then but it won't be far off. You could hardly have more scope for choice! Incidentally, designing an amp to start clipping just as the bass units start to struggle is probably NOT a great idea. What you should really be concentrating on is the cooling. Two 3886s in parallel will help no end with this by doubling the area of contact, and if you are planning on doing 5hr listening sessions then it might extend its life, not that that is usually a problem. If I were doing a commercial product, and pushing the rails that high, I would use two. For myself, I'd use one - all other things being equal. If you want a direct recommendation I would buy a 25V to 28V transformer and be done with it.
Let's just cut to the chase. A 3886 will work on transformer voltages anywhere up to and including 30V-0-30V, giving rails of +/- 44V and it will drive even sub 4R speakers without much trouble. (I know because I have one here.) It won't quite make the 100W mark into 8R as the rails will have sagged by then but it won't be far off. You could hardly have more scope for choice! Incidentally, designing an amp to start clipping just as the bass units start to struggle is probably NOT a great idea. What you should really be concentrating on is the cooling. Two 3886s in parallel will help no end with this by doubling the area of contact, and if you are planning on doing 5hr listening sessions then it might extend its life, not that that is usually a problem. If I were doing a commercial product, and pushing the rails that high, I would use two. For myself, I'd use one - all other things being equal. If you want a direct recommendation I would buy a 25V to 28V transformer and be done with it.
That's like saying, "I found gravity to be inconvenient so I just don't care about it. It keeps getting in the way".
If you want to use Ohm's Law (or just about any other equation) on a sine wave you need to use the RMS values for the current and voltage. That's pretty bloody fundamental. I do this daily. If this is inconvenient for you, I suggest you learn what RMS is.
There are three terms that you need to keep straight:
- RMS - Root Mean Square. As others have pointed out, this is a type of average. A sine wave with a voltage of 1 V RMS will dissipate 1 W in a 1 Ω resistor. P = E * I and I = E/R, remember?
- Peak. Often denoted with a subscript p or peak. That's measured from the middle of the sine wave (often 0 V) to the top (or bottom) of the sine wave.
- Peak-to-peak. Often denoted with a subscript pp or p-p. That's measured from the bottom of the sine wave to the top. Vpp = 2*Vp.
For a sine wave, Vp = sqrt(2)*RMS. It then follows that RMS = Vp/sqrt(2) = Vp*0.707. As I pointed out above, this is not the case for all waveforms but it does work for a sine wave. Thankfully, those are the most common in audio.
Don't make it harder than it is by inventing new words like "netted peak" or whatever.
Tom