I am using a DSP with limiter (some live situation are too dangerous) and I am using a "rise up" from Ecler alma 26, and the amplifier class D I think it has an integrated limiter. all are very discreet, and when the clip red led is blinking I know that is fine, the limiter is working, speakers are safe.
Party was a great success. Ended up using peak limiters on all output channels set to just before clip and an input compressor set to medium knee with 300ms attack and release set to the point where pink noise was just triggering the peak limiters. On live music the input compressor basicly didn't activate but transients from the kick where hitting the midbass and sub limiting. On dj material the compressor was invaluable, I found the rearly good DJs just instinctively rode the edge of the compressor which changed the tone of the bass heavy passages in a rearlly pleasing way. I'm thinking that a possibility for improvement could be a split band input compressor to avoid the sparkles getting squashed during bass heavy pasaages but overall fantastic sound was achived for a relatively modest system.
The input compressor works a treat I've found. The release time Ive used is far shorter than yours which I think makes it less transparent to a point, but then sounds overly crushed and perhaps a bit nasty when really pushed. Hopefully the DJ will hear this and turn it down a bit, although at this point they would have to be lighting up 2 or more reds which would eat up 1 of their 3 strikes (and they are off, or the system goes off).
Several things:
Speakers are damaged in two ways: overheating and over-excursion. Tweeters are mostly damaged by overheating, as excursion is generally small. Woofers are generally damaged by over-excursion where the voice coil slams agains the magnet.
Heat damage is an RMS function, not a peak function. Pushing a signal above clipping limits the peak value (to the clipping point) but allows the RMS value to rise in a way that is not dissimilar to what you'd have if you raised the signal but did not clip, at lest for the first few dB above clip threshold.
Excursion damage is a peak function, and the voltage threshold of the excursion limit drops with rising frequency until it intersects with the heat damage point, then you have an RMS function.
Peak limiting with a zero attack is not the same as clipping because of the release time. Both clipping and peak limiting with fast release limit peaks but raise RMS. Peak limiting can result in less distortion than clipping, depending on release time and frequency. The slower release time, the less RMS increase above threshold. Taken to extreme, a zero attack and very long release will cause instantaneous downward gain change on a peak, but no other change (no distortion, no RMS increase) until the limiter recovers enough to attack the next peak above threshold. Taken to the opposite extreme, zero attack and extremely short release, the effect approaches that of clipping. So you dial in recover time to minimize (or maximize) audible artifact balanced against desired peak control. Remembering that once over-excursion has been dealt with, speakers are damaged by RMS (heating value) level, thus time is a factor.
A compressor with slower attack and release should not be assumed to use an RMS detector. Many use simple average detectors, others just us a peak detector that is very slow. Neither is an accurate RMS processor. Make sure you are getting true RMS response.
Separate limiters post crossover is a nice way to get away with increased limiting action with less audible artifact. However, maximum power handling of tweeters is generally about 1/10th as high as woofers and mid-range. Thresholds and response times for each band would vary significantly, as does the spectral density in music for each band.
Ideally a tweeter protection processor would be based on the thermal damage time vs RMS level curve of a specific driver. Have fun finding that on anything but pro drivers.
The ideal woofer protection processor would be based on both peak limiting and RMS limiting. The peak limiter would use a frequency variable threshold, lower frequencies cause higher excursions, thus should limit sooner. A straight, flat threshold peak limiter would be hyper-active for upper bass frequencies, and inadequate for very low ones. The RMS limiter would follow the heat damage curve of the driver.
Peak limiter distortion is a function of release time, faster results in higher distortion. Some hardware peak limiters use a complex release time where small amounts of gain reduction are released more quickly than large ones, as a distortion reduction method. Any release time stated in dB/ms is not doing this, and will produce the same amount of distortion regardless of how far above threshold it is operating.
A slow attack peak limiter isn't actually a peak limiter at all, as peaks above threshold will get right through.
Speakers are damaged in two ways: overheating and over-excursion. Tweeters are mostly damaged by overheating, as excursion is generally small. Woofers are generally damaged by over-excursion where the voice coil slams agains the magnet.
Heat damage is an RMS function, not a peak function. Pushing a signal above clipping limits the peak value (to the clipping point) but allows the RMS value to rise in a way that is not dissimilar to what you'd have if you raised the signal but did not clip, at lest for the first few dB above clip threshold.
Excursion damage is a peak function, and the voltage threshold of the excursion limit drops with rising frequency until it intersects with the heat damage point, then you have an RMS function.
Peak limiting with a zero attack is not the same as clipping because of the release time. Both clipping and peak limiting with fast release limit peaks but raise RMS. Peak limiting can result in less distortion than clipping, depending on release time and frequency. The slower release time, the less RMS increase above threshold. Taken to extreme, a zero attack and very long release will cause instantaneous downward gain change on a peak, but no other change (no distortion, no RMS increase) until the limiter recovers enough to attack the next peak above threshold. Taken to the opposite extreme, zero attack and extremely short release, the effect approaches that of clipping. So you dial in recover time to minimize (or maximize) audible artifact balanced against desired peak control. Remembering that once over-excursion has been dealt with, speakers are damaged by RMS (heating value) level, thus time is a factor.
A compressor with slower attack and release should not be assumed to use an RMS detector. Many use simple average detectors, others just us a peak detector that is very slow. Neither is an accurate RMS processor. Make sure you are getting true RMS response.
Separate limiters post crossover is a nice way to get away with increased limiting action with less audible artifact. However, maximum power handling of tweeters is generally about 1/10th as high as woofers and mid-range. Thresholds and response times for each band would vary significantly, as does the spectral density in music for each band.
Ideally a tweeter protection processor would be based on the thermal damage time vs RMS level curve of a specific driver. Have fun finding that on anything but pro drivers.
The ideal woofer protection processor would be based on both peak limiting and RMS limiting. The peak limiter would use a frequency variable threshold, lower frequencies cause higher excursions, thus should limit sooner. A straight, flat threshold peak limiter would be hyper-active for upper bass frequencies, and inadequate for very low ones. The RMS limiter would follow the heat damage curve of the driver.
Peak limiter distortion is a function of release time, faster results in higher distortion. Some hardware peak limiters use a complex release time where small amounts of gain reduction are released more quickly than large ones, as a distortion reduction method. Any release time stated in dB/ms is not doing this, and will produce the same amount of distortion regardless of how far above threshold it is operating.
A slow attack peak limiter isn't actually a peak limiter at all, as peaks above threshold will get right through.
Hey Jaddie,
Good to see you back around this side of the forum. I'd been looking for your posts a little while ago - found them here: Calculating power required for bi-amped 2 way system
Realised I never said thanks for taking the time to present the data. So, thank-you for that. I can see it was a lot of work.
Chris
Good to see you back around this side of the forum. I'd been looking for your posts a little while ago - found them here: Calculating power required for bi-amped 2 way system
Realised I never said thanks for taking the time to present the data. So, thank-you for that. I can see it was a lot of work.
Chris
Thanks.
When I looked over that thread I felt again that every time I post the clipping/spectral analysis, it's like tilting at windmills. It always generates a fight with people who have swallowed the "other" idea. When I went after the clipping-burning-drivers problem I expected to find clear evidence contrary to what I found. If there was even a little bit of vagueness about the result, I'd have backed off a long time ago. It's a case where the popular understanding isn't really an understanding, its a belief.
Don't worry its pretty obvious when you consider real power is the integral of the current voltage product W.R.T time and clipping is just chopping the peaks off. Considering a drivers complex impedance the only real limiting would be voice coil temprature sensing and peak limiting to prevent over excursion. Powersoft have amplifiers which can measure the real power delivered to the load and perform limiting based on that which is most of the way there.
For a custom sub amp design it would be possible to introduce a small DC offset and use that that measure the voice coil temperature but the amp would have to be a plate amp as the composition of the wiring to the VC would have to be known (also you would have to know the voice coil wire temperature coefficient - which would probably require testing on a re-cone kit).
I set my limiters to about 5% below the maximum output of the amp. The limiter will not completely stop signal from going louder, as that would also be practically clipping. Using a very high ratio compression like limiter (15 to 1) means that it functions more like an analog rounding clipper.
SOOO much more to it than that. A limiter is not practically clipping, even at infinite ratio, primarily when you consider release time. A good technical peak limiter will fully limit on the leading edge of a peak, then not distort because its release time doesn't allow the gain increase to track the waveform, only the envelope. Additionally, attack time can actually be slowed to permit a slightly smoother sound so long as the threshold is appropriately lowered as well. Release time can also be made complex, as a function of peak duration, or gated to freeze during the peak to effectively reduce gain change induced distortion to zero. Then there's the DSP based peak limiter with look-ahead, which totally eliminates attach envelope distortion.
By definition a peak limiter has an infinite slope and will completely stop peaks over threshold. However, that's not a loudness function, that's a peak function. Very different, and not completely related. Advancing the input to a peak limiter with fast release will result in a range of proportionally increased loudness at the expense of short-term dynamic range. If the peak limiter isn't stopping peaks, it's a terrible peak limiter, or is deliberately slowed for other reasons, at which point the threshold can be lowered to maintain maximum peak control.
Nah, nothin' like clipping. At all.
If your hitting a limiter then the amp/speakers simply aren't loud enough.
I worked in an audio shop for a while and the amount of overdriven stuff we got in was amazing. Just people not buying the right gear in the first place.
You wouldn't flog your car at full revs all day long so why do the same with an audio system ?
I worked in an audio shop for a while and the amount of overdriven stuff we got in was amazing. Just people not buying the right gear in the first place.
You wouldn't flog your car at full revs all day long so why do the same with an audio system ?
In an ideal world we would all have the rig for the gig per say. However I'm not sure your aware of the kind of gear I regularly see limiting. I see things like 14 FLH subs running on MA5002VZ's hitting limiters on bass heavy passages (large lossy indoor space). If you have the capability in the sub bass people will use it. Having correctly set limiters is essential to protect the system.
My system always sounds great and nothing like some wedding DJ clipping the amps, I keep an eye on levels and adjust if needed, correctly set limiters just reduce the effort and provide protection.
I'm always working on making my system better since this gig I have another P7000s which will allow me to run my subs with 4* the input power as I can run each P7000 bridged. Admittedly I will have to peak limit at around 90V due to running out of excursion on these particular subs so the actual dB increase will be small. However the amps will now not clip before the excursion protection limiter has to kick in. I have also replaced the tops amp with a CTS4200 which should give a few more dB.
current projects are:
1) 4 new subs using 15TBX100, these will have ~4dB more output per sub than my current subs if given enough amp. I will probably use all 8 though in order to get more directivity from my sub woofer array.
2) New top amp, I have been fixing amps to save up spare cash for a high end quad channel amp ~500W/channel, I don't expect to put more average power through the tops but this will eliminate any clipping.
3) entirely new large system using front loaded horns for outdoor use (see Mega midbass straight horn 139dB)
My system always sounds great and nothing like some wedding DJ clipping the amps, I keep an eye on levels and adjust if needed, correctly set limiters just reduce the effort and provide protection.
I'm always working on making my system better since this gig I have another P7000s which will allow me to run my subs with 4* the input power as I can run each P7000 bridged. Admittedly I will have to peak limit at around 90V due to running out of excursion on these particular subs so the actual dB increase will be small. However the amps will now not clip before the excursion protection limiter has to kick in. I have also replaced the tops amp with a CTS4200 which should give a few more dB.
current projects are:
1) 4 new subs using 15TBX100, these will have ~4dB more output per sub than my current subs if given enough amp. I will probably use all 8 though in order to get more directivity from my sub woofer array.
2) New top amp, I have been fixing amps to save up spare cash for a high end quad channel amp ~500W/channel, I don't expect to put more average power through the tops but this will eliminate any clipping.
3) entirely new large system using front loaded horns for outdoor use (see Mega midbass straight horn 139dB)
Hi Chris, I hope that your also keeping busy, any projects?
I also have this mad subwoofer plate amp project which is actually taking all the time up at the moment. Basic idea is to store transient energy in an 10 mF capacitor bank to have an amp that can drive the largest 21" drivers to their limits while costing little money and running off a 13A plug: https://www.diyaudio.com/forums/pow...olated-amplifier-pfc-stage-6.html#post6162750
Its possible but I'm having to push the technology quite hard and the final cost might not be that low.
As per usual anything that meets the spec:
I notice Lab Gruppen CX series amps are sometimes going cheap second hand. Another option would be to put the P7000s's on tops and get a new sub amp but its tricky working out what the fan noise situation is with high power amps. Still waiting for the great flood of cut price amps from venues going bust. In a dream world I would just run everything off an X8 ha.
I also have this mad subwoofer plate amp project which is actually taking all the time up at the moment. Basic idea is to store transient energy in an 10 mF capacitor bank to have an amp that can drive the largest 21" drivers to their limits while costing little money and running off a 13A plug: https://www.diyaudio.com/forums/pow...olated-amplifier-pfc-stage-6.html#post6162750
Its possible but I'm having to push the technology quite hard and the final cost might not be that low.
As per usual anything that meets the spec:
- quiet (both electrically and acoustically)
- if using a DSP it needs to have digital input and use 40 bit or above floating point processing. This is due to having very large dynamic range requirements from my system and needing to control the gain upstream from the amps and not liking to change the attenuation settings on my amps. While also disliking hiss.
I notice Lab Gruppen CX series amps are sometimes going cheap second hand. Another option would be to put the P7000s's on tops and get a new sub amp but its tricky working out what the fan noise situation is with high power amps. Still waiting for the great flood of cut price amps from venues going bust. In a dream world I would just run everything off an X8 ha.
Hmmmm, looks like the Powersoft T-series only use 32-bit. I think that's enough, though, given the dynamic range available in a typical venue with an audience present.
Having a lot of storage for the big bass hits makes a lot of sense. FWIW, I'm looking to sell my Crown MA12000i. Quite a lot of energy storage in there - the caps are 220v 1200uF, and there's 20x of them. There's 10x 1KV 33uF at the primary side.
Needless to say, I didn't switch it on while I was poking around. The fan noise might be a problem for you, but I'm happy to dig it out and take some SPL measurements if you're interested.
Projects-wise, I spent yesterday in the back yard with the Powersoft rack, a measurement mic, REW and RePhase. Got the 10" coaxials and 2x10+HF speakers to have a flat phase response from about 600Hz upwards, and made the frequency response nice and flat, too.
Sounded exceptionally good - I actually spent a while just sitting on the grass and listening to music.
Apart from that, I'm fixing up an old McGregor mixer-amp from the 1980s. I'm nowhere near as good as you at electronics, so it's simple stuff for me. Also working on various mics - turns out I own over 100 of the things, so I'm fixing those that need work and will get them on eBay soon.
Hope you're keeping safe, and let me know on that Crown amp. I just checked, and there's the option to limit output power delivery according to the power ratings of the speakers. Might be handy.
Chris
Having a lot of storage for the big bass hits makes a lot of sense. FWIW, I'm looking to sell my Crown MA12000i. Quite a lot of energy storage in there - the caps are 220v 1200uF, and there's 20x of them. There's 10x 1KV 33uF at the primary side.
Needless to say, I didn't switch it on while I was poking around. The fan noise might be a problem for you, but I'm happy to dig it out and take some SPL measurements if you're interested.
Projects-wise, I spent yesterday in the back yard with the Powersoft rack, a measurement mic, REW and RePhase. Got the 10" coaxials and 2x10+HF speakers to have a flat phase response from about 600Hz upwards, and made the frequency response nice and flat, too.
Sounded exceptionally good - I actually spent a while just sitting on the grass and listening to music.
Apart from that, I'm fixing up an old McGregor mixer-amp from the 1980s. I'm nowhere near as good as you at electronics, so it's simple stuff for me. Also working on various mics - turns out I own over 100 of the things, so I'm fixing those that need work and will get them on eBay soon.
Hope you're keeping safe, and let me know on that Crown amp. I just checked, and there's the option to limit output power delivery according to the power ratings of the speakers. Might be handy.
Chris
Clipping can burn speaker coils. This is what I was warned about when I purchased my two Wharfedale Titan 12" speakers. Even if the speakers are large, it all depends on the voice coils' resistance, amplifier rail voltage and power supply current capabilities. At clips the thermal dissipation in a speaker's coil is equal to V^2/R, just like with any resistor. So, with a rail of 30V and a coil resistance of 2 Ohms, this will be 30^2/2 = 450W of power. A few seconds are enough to destroy a voil coil.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.