Speaker volume protection - current sensing not microphone

...

Clamp-type sensor, probably current transformer, be aware that CT usually specified at 50/60 Hz, and may not be a good choice for 20-20000 Hz application. ACS712 seems for me a better option, though it requires to cut in line.

FantomT:
Clamp-type sensor, probably current transformer, be aware that CT usually specified at 50/60 Hz, and may not be a good choice for 20-20000 Hz application.

Hi FantomT, thanks for your response.
50/60Hz - Ahh, interesting. It wouldn't really need to be 20kHz, I would imagine 5kHz would be able for reliable detection.

Can anyone comment on signal degradation using in-line type of current sensor?

Yep, okay, I watch this video from the chip manufacturer and it seems any possible degradation would be negligibly minimal.

I've updated my original post to remove the 'which sensor?' question.

Thanks for pushing me to research that more deeply, FantomT! :slight_smile:

How about these options:

  1. make sure your amps can not damage the speakers simply because they can't produce that much power, or if visiting tech connects their amps:
  2. have them pay a deposit to cover any damage done to your equipment (or other way to make sure that you get compensated).

For the Arduino solution: of course this should be totally non-invasive, to not introduce any noise or distortion. A clamp-on current sensor may or may not work: those are normally designed for 50/60Hz AC currents, not an audio signal. Transformers are very sensitive to the frequency of the magnetic fields, so any current signal is likely to be way off. But there's another problem when measuring current: speakers are an inductive load, so the impedance (resistance) changes big time with the frequency. So for a fixed output voltage of your amps the current varies with frequency. It's all in all going to be a very complex relationship.

But imagine that there's indeed a sensor that can detect the current running through the cables. That I haven't heard of one doesn't mean there isn't one. Now you get your current signal, and you know the frequency, with it you can calculate the current, and then you detect the current is too high. What are you going to do with that information? Open a relay in the amp's power supply? Send a signal to the amp to lower the volume? What kind of signal would that be?

Hi wvmarle,
Thanks for your response. I've run a commercial studio for 18 years, so I have some ideas on this, though this is the first time I'll have a complex which will be available for dry hire.
As for insuring the damage; the speakers can't easily be replaced, so this is not really an option. Like a widow who gets £10,000,000 because their partner died - they're still dead.

Anyways, for the Arduino stuff, as I wrote in my pseudocode, the relays would break the circuit between the speaker and amplifier. Turning the amplifier off would be a bad idea because of 'spiking' the speakers and turning the input signal down would not be possible with the BOM I listed. (Besides, building what is essentially a volume control is outside the scope and necessity of this project.)

I just realised that after a threshold exceeded state has been met, I didn't turn the relays back on again, so I'll add that to the pseudocode now.

When you cut the current with the relays, there is no current left to measure. So you can not switch on those relays based on the output current of the amp.

Are any relays good enough to not cause any distortion of your precious audio signal? It sounds to me like it's a great source of noise, with their mechanical contacts and so, and this sounds like a seriously high-end audio setup.

As mentioned, speakers are an inductive load, and inductive loads are prone to causing pretty big voltage spikes by themselves when switched this sudden (inductors will do what they can to keep the current flowing - so your 19V can become a couple hundred volts, maybe even >1kV, depending on what else is in the circuit and the size of the inductor - the speaker coil - itself). I would expect speakers to behave in the same manner, it's definitely something to look into.

Another thing, out of curiousity: 105W sounds like a very low power rating for such a speaker to me. I have to admit I don't know much about it, but the ratings that I encountered in the pop music scene started at 500W or so, same for the amplifiers. That's probably a peak power vs. continuous power (I've seen tiny computer speakers being rated at more than your 105W...).

Anyway, if you have an amp that can produce only 80-90W continuous, isn't that enough of a safeguard for your speakers?

Hi wvmarle,
I'm sorry to say, but there are all sorts of logical flaws and assumptions in your reply.

  1. If you read the pseudocode, you'll see it doesn't remeasure the current before closing the relays after an incidence of threshold exceeded.

  2. A low output amplifier could potentially be pushed into clipping (square wave output) well before reaching the maximum input of the speakers they're connected to. This DC would damage the speakers, so under-rating the amplifier to protect the speaker is a fallacy. The converse is also (obviously) a bad idea. Education is the answer and then that's too risky, bring in the Arduino. :slight_smile:

  3. Greater than 1kV? I doubt that's possible in the scenario I described.

  4. I've seen computer speakers with 1000W ratings. :wink: Look at the hifi world and you'll see lots of low power amplifiers that are highly regarded and widely used. For reference, the amp is 45+45W RMS into 4R (and it sounds amazing)

  5. The speakers are Duntech Marquis which are very high efficiency. They can get loud very easily with even a modest amplifier. The Duntech Crown Prince (Princess) is a very different beast and I have a pair of Hypex 700WRMS monoblocks that I built running those.

  6. "Your precious audio" came across as snarky. The Hypex amps have relay-operated protection boards so if relays are okay for Bruno Putzeys, they're okay for me. :slight_smile:

Thanks for giving it a shot, though.

Best regards,
Dax.

daxliniere:
2) A low output amplifier could potentially be pushed into clipping (square wave output) well before reaching the maximum input of the speakers they're connected to. This DC would damage the speakers, so under-rating the amplifier to protect the speaker is a fallacy. The converse is also (obviously) a bad idea.

This goes against intuition, if only as DC can be blocked readily with a capacitor, while passing the AC signal just fine. If you can have relays that don't disturb your signal, there must be capacitors that can do the same.

You can't underrate an amp you say (I'll take your word for it, references would be better), but overrating is obviously equally bad. That doesn't explain how a matched amp (one that can produce the power the speakers need, no more no less) can in fact damage a speaker.

Anyway, my basic comment is still: I'd try to make sure the input can never make the output exceed the limits. If you control the amp as well, put a physical limit on the volume knob so it can only go to 10, not 11, let alone 12. Limit whatever input signal the amp takes to a safe maximum. Whatever. It's always going to be more reliable than trying to measure the current that goes into your speakers and then just cutting them off.

Man, I don't want to argue about the application with you. I am asking the community for help in putting my idea together. If I wanted ill-informed opinions and intuition, I'd follow trump on twitter.

You cannot get a capacitor that doesn't affect your signal.

Regarding 'limiting the volume knob', take this example. You have your input potentiometer 'locked' to a maximum of 50% - what's stopping someone increasing their input signal? This happens all the time on live gigs where the DJ keeps turning himself up from his mixer and the FOH engineer keeps turning him down at the FOH desk. Eventually his output is so hot and distorted he's feeding up completely clipped audio which, no matter how low you turn it, WILL damage the LF drivers in a serious speaker system. A volume control isn't a limiter and putting an actual limiter between the source would definitely degrade signal, provide an non-accurate representation of the source audio and also lead to the same (approaching-)clipping problem I described above.

School's out for summer.

You're asking for ideas on how to handle an audio system on an Arduino forum. You get ideas from Arduino engineers, not from audiophiles.

Then don't complain if those ideas are not what you need, or are simply not feasible for your application. If you don't like the ideas you get here, you'll have to come up with your own, and make sure you research them thoroughly as otherwise you may very well damage your equipment. Asking on an audio forum that deals with high-end amplifiers is actually probably a better place than this for your question, as your real problem is how to sense you're reaching the limits. The Arduino part is truly trivial.

Or just supervise whoever rents your equipment.

wvmarle:
You're asking for ideas on how to handle an audio system on an Arduino forum.

This confirms you read less than half of what I wrote.
Please explain to me which of the 4 questions I posed is an audio question?

My questions:

  1. What would be the best current sensor to use? I think it makes sense to favour precision over power handling so I guess a 5A capacity sensor seems adequate. (The the amplifier's power specs are 19VDC 5.3A input (maximum 105W consumption), 2x45W output into 4R load.)

  2. Would CH340 be fast enough and have enough memory to store the data for long-term threshold? (AT328-compatible)

  3. How would I calculate the long-term threshold? I reckon this should be an average of about 20-30 secs.

  4. Any pitfalls? Questions I should have asked? :slight_smile:

daxliniere:
This confirms you read less than half of what I wrote.

No, I'm just more than aware of the XY problem. So it's always a good idea to start by figuring out whether the proposed solution is a suitable solution for the problem at hand, then whether it's the best solution. Most problems have multiple solutions available, all with their pros and cons. In your case the human problem seems to be a major part of the whole issue.

daxliniere:

  1. What would be the best current sensor to use? I think it makes sense to favour precision over power handling so I guess a 5A capacity sensor seems adequate. (The the amplifier's power specs are 19VDC 5.3A input (maximum 105W consumption), 2x45W output into 4R load.)

The nature of the signal is audio. Very different than the clean, single frequency, usually 50 or 60 Hz, sine wave that these sensors are designed for. You may ask an audio engineer how to non-invasively measure the current of a speaker signal.

  1. Would CH340 be fast enough and have enough memory to store the data for long-term threshold? (AT328-compatible)

That's a USB to TTL converter. Whether it's fast enough depends on the amount of data you try to push through it, it has a handful of bytes for buffer, that's it. Don't know what you plan to do with it in this project. Agreed, no audio issue here.

  1. How would I calculate the long-term threshold? I reckon this should be an average of about 20-30 secs.

Depends on your signal sampling frequency, and the nature of the signal. Audio is fluctuating all the time, has lots of superimposed frequencies, and is generally a very dirty signal. So your long-term threshold depends on the limits of your speaker system - audio technology. What current can it have on average for 20 seconds? For 1 second? For 1 ms? How do subsequent peaks affect your speakers? How fast should it react? What peak/series of peaks is acceptable? Do you want to detect clipping? (another for the audio engineers).

  1. Any pitfalls? Questions I should have asked? :slight_smile:

Frequency response of current sensor. Voltage spikes when cutting current in speakers (an inductive load - but what is its response?), and potential issues with cutting current on the amplifiers side - ask audio amplifier engineers on how amps react on such switching. Current vs frequency (it's an audio signal and audio speakers). Audio engineers may come up with more issues with these ideas.

This actually a very difficult problem and the best solution is probably to use a lower power amplifier that’s less-likely to blow the speakers.

See [u]Speaker Power Requirements[/u]. It says with clean music you can use an amplifier with twice the speaker rating. With guitar amplifiers that are likely driven into distortion the recommendation is flipped-around and the speaker should be rated for twice the amplifier.

If crazy engineers (or drunk people) are in charge of the volume control, you should probably go with the 2nd option and use an amplifier with half the speaker rating. Or, don’t allow anyone else to use your irreplaceable speakers.

I’ve read stories of frequently-blown speakers at the old Motown studios (I think they were blowing the midrange in AR speakers) and Phil Spector was famous for monitoring at insane levels but I don’t know if he blew any speakers/monitors.

…For reference, the amp is 45+45W RMS into 4R (and it sounds amazing)

  1. The speakers are Duntech Marquis which are very high efficiency. They can get loud very easily with even a modest amplifier. The Duntech Crown Prince (Princess) is a very different beast and I have a pair of Hypex 700WRMS monoblocks that I built running those.

Are you using a 45W amp or a 700W amp? What’s the power rating of the speakers?

As you probably know, the peak-to-average ratio (crest factor) of music is about 20dB. Speakers are usually burned-out from the average power (perhaps short-term average) so you can fry a 100W speaker with continuous 100W test tones. And, that’s if you can trust the power rating of the speaker. And of course, the woofer has a higher power rating than the tweeter so it’s even easier to blow a tweeter with test-tones. With a 100W speaker the tweeter is designed to handle the high-frequency part of a signal peaking at 100W.

…Given that ~20dB ratio, probably the only “foolproof” solution is to use an amplifier with 1/10th the speaker rating, but that’s simply not practical or economical (it would take a huge speaker) and you still might be able to fry the tweeter with sine waves or square waves.

There is a common myth that you’re more-likely to fry a tweeter with a low-power amplifier driven into distortion (because of the harmonics) but that’s not true. It’s true that you can blow a tweeter that way… The average power continues to go-up when the peaks are clipped and the clipping does generate more high-frequency energy. But, a cranked-up higher-power amplifier will put-out even more power, including more power in the high frequencies. (And, most people will turn-down the lower-power distorted amp, but possibly turn-up the undistorted amp.)

As I’m sure you know, most modern studio monitors are active and bi/tri amplified. The amps are matched to the drivers with lower power for the tweeter (an midrange in a 3-way). I’m not sure if it’s impossible to fry one of these monitors, but it’s less likely and some may have built-in protection, maybe including thermal protection.

The mains electronics:
2x ACS712 (5 amp) hall-effect-on-chip current sensors (for left and right speakers)

You’re confusing me with “mains”… Usually “mains” refers to the AC power line.

I don’t see any advantage to measuring current. Voltage is easier to measure and speakers are tested/specified with applied voltage.

Power (Watts) = Voltage X Current.

And from Ohm’s Law (Current = Voltage/Resistance) we can derive:
Power = Voltage2/R or Current2 X R.

The impedance of speakers varies over the frequency range, but speaker power-rating specs assume the rated impedance. That means they use the actual applied voltage and the nominal rated impedance, not the actual current.

Hi DVDDoug,

You're confusing me with "mains"... Usually "mains" refers to the AC power line.

Yes, that's a gross typo, my apologies. I definitely meant to convey the BOM for the main components of the build.
(Corrected in post#1)

I don't see any advantage to measuring current. Voltage is easier to measure and speakers are tested/specified with applied voltage.

I want this protection system to be as isolated from the audio system as possible. My original thought was to use a clamp-style current sensor as I figured it would be unintrusive compared to an inline current sensor. However, having seen the inner-workings of the ACS712, I'm confident that would be fine.

Also, in my original post, I wrote: "I will set the thresholds manually using the serial read monitor in the programmer, so no power calculations are necessary."

Hi wvmarle,

In my updated post, I used the strike-through style on the first question, as referred to in post #3, but when quoting myself, the style is omitted by the forum software.

The nature of the signal is audio. Very different than the clean, single frequency, usually 50 or 60 Hz, sine wave that these sensors are designed for. You may ask an audio engineer how to non-invasively measure the current of a speaker signal.

Okay sure, so a capacitor needs to go on the output of the current sensor boards to smooth the analog data.

Would CH340 be fast enough and have enough memory...
That's a USB to TTL converter.

My gosh, you're right. Whenever I see these "Arduino AT328 compatible" boards, they always have CH340 in the title, so I incorrectly assumed that the Chinese were using a CH340 chip instead of a genuine Atmega AT328. Thanks for the correction, now I know it's just the USB chip.

Fuses used to be used in such an application. Is this not possible?

Paul

Fuses used to be used in such an application. Is this not possible?

I concur. Simple and reliable.

And they won’t upset the sound.

Allan

I will set the thresholds manually using the serial read monitor in the programmer, so no power calculations are necessary.

There is a misunderstanding, how automatic control system usually works. Power calculation is Must, because you need a real-time data compare to threshold (doesn't matter where it comes from, written in EEPROM, given over serial or radio) . than algorithm Does a decision if action (on/off) required.
BTW, you don't need to change threshold, data given in the speakers data sheet.
Googling a little, I discovered some standard that defines RMS and Peak power.
http://www.doctorproaudio.com/doctor/temas/powerhandling.htm

For myself, I'd be concerned with Peak power, and if it's not specified in the data sheet, take 2x RMS power. Time period should not be 10-20 seconds, rather one period of the lowest frequency, 50 milliseconds or even less.

I don't see any issue with using a relay, and almost every protection unit I've seen use them.

Though power calculation math is unavoidable, and here is simple logic: frequency of the signal defines you hardware selection. Sensor should be certified up to 20k, (acs712- 80k - O'k), ADC sampling rate has to be 40k Per channel, 80k for stereo. There is a way to push common arduino AtMega chip to sample about that rate, and I think it 'd able to do a math, but it's need to be verified. Otherwise arduino Due is better choice, and it could process 4x channels (2 - current, and 2 - voltages, to get most accurate results w/o frequency non-linear impedance of the load)

Or you could follow more simple path, using hardware (OPA based) AC peak detector, leaving for arduino DC measurements. It 'd greatly simplified software, but not as accurate.

allanhurst:
I concur. Simple and reliable.

And they won't upset the sound.

Allan

False assumption, fuse is variable resistance nuisance, never used in high quality equipment due to high level of the THD