diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Class D (http://www.diyaudio.com/forums/class-d/)
-   -   DC protection (http://www.diyaudio.com/forums/class-d/87304-dc-protection.html)

Pierre 28th September 2006 10:18 AM

DC protection
 
I am trying to determine what is the best strategy for effective but not too costly / complicated DC protections for my loudspeakers in case of amplifier failure.
Once a DC detection circuit is designed, there are three ways to react, using relays.
a) Cut both supply rails to the amplifiers.
b) Cut Speaker outputs
c) Cut PSU mains input (or turn supply off by other means)

Alternatives a) and b) are quite safe, but what about (c)? In my case it would be easy to implement and requires a single and smaller relay, and easier cabling. BUT I need to know if discharge of the PSU capacitors (say, 60V rails with 10000uF/rail) is enough to cause any damage in typical speaker/s in case of module shortcircuit of any rail to the speaker output. What do you think?

JohanH 28th September 2006 10:29 AM

Best would probably be to cut the speaker outputs.
Also cheaper because you would only need 2 contacts on the relay.

Cheers,
Johan

PS. I would not rely on the PSU caps to discharge quick enough

Pierre 28th September 2006 10:34 AM

Yes, you need a SPST relay PER CHANNEL. So for a stereo or multichannel amplifier it is more expensive. Also, the relay ratings are huge for high power amplifiers.

If you cut the rails (also very common approach), you need two relays (one per rail) or a double-circuit one, but also with high current ratings. This solution is still very safe but cheaper.

And the simpler one is to turn the supply off. You can do it by simply removing mains or, if it is a SMPS supply, shutdown the controller. This will produce that the output caps have to discharge and will do it through the speaker. Don't know how harmful this would be to the speakers.

ssanmor 28th September 2006 10:57 AM

Pierre, if you turn the SMPS off, in the worst case (shortcircuit of speaker output to a rail) you are effectively discharge a 10.000uF capacitor through a 8 ohm speaker, (voltage will drop exponentially from initial rail value to 0, and will take only 80ms to reach 33% of initial voltage), I doubt this is enough energy to cause any speaker damage, but would like to hear other opinions!

(You must also account for the time passed from the detection of the DC level to the supply cutoff, of course! But that should be a few ms max...)

JohanH 28th September 2006 10:58 AM

Quote:

Originally posted by ssanmor
I doubt this is enough energy to cause any speaker damage, but would like to hear other opinions!



That would depend on the speaker.
If one of the output devices fails or shorts, my guess is there´s speakers out there that would not like it.

Cheers,
Johan

ghemink 28th September 2006 02:29 PM

Quote:

Originally posted by ssanmor
Pierre, if you turn the SMPS off, in the worst case (shortcircuit of speaker output to a rail) you are effectively discharge a 10.000uF capacitor through a 8 ohm speaker, (voltage will drop exponentially from initial rail value to 0, and will take only 80ms to reach 33% of initial voltage), I doubt this is enough energy to cause any speaker damage, but would like to hear other opinions!

(You must also account for the time passed from the detection of the DC level to the supply cutoff, of course! But that should be a few ms max...)


The DC detection itself also takes time, if the detection is too fast, it will trigger on low frequency heavy bass signals. So you have a certain time where you have the DC at the output but where it is not yet detected, this could actually be longer than the time that it takes to discharge those caps. In case you want it to be safer with an SMPS, you could decide to use less capacitance at the SMPS rails (less stored energy).

Just a simple thought experiment to see if what I say here is correct.

I think most DC detection circuits consist of a simple 1st order low-pass filter that is connected to the amps output. If the output of that low-pass filter becomes too high (say higher than 0.6V in either direction, + or -) the DC detection circuit triggers. If we have an amp (like UcD400) that can give close to 60V peak output voltage (close to 40V RMS that is needed for 400W in 4 Ohm) at say 20Hz, it means that that low pass filter should attenuate the signal with a a factor 100 or more at 20Hz otherwise, that 20Hz signal could trigger the DC detection circuit. A factor 100 is 40dB, this means that with a 1st order low pass filter, the corner frequency of that filter should be at 0.2Hz. Knowing that f=1/(2*pi*RC) is the corner frequency of a first order RC filter, the RC time constant would be (crunch crunch crunch) about 0.8s. Knowing that, we can guestimate that it takes about 50ms to detect 10V DC (0.6V*0.8s/10V) (assuming the DC comes in in 0 time, meaning basically a step function, from 0 to 10V in no time). For 60V DC at the output, it would take then only 8ms to be detected, so pretty fast, actually faster than I had first thought. So the delay due to the DC detection circuit itself is actually not very long. If you want the DC detection circuit to be a bit more robust, i.e. you do not want it to trigger on a 16Hz pipe organ (how often do you listen to that at 400W level :-)) you have to lower that low-pass filter corner frequency a bit more, however, even then, the 60V DC would be detected in 12ms or so, far less than what it takes to discharge the caps via the speaker.

Gertjan

Pierre 28th September 2006 07:04 PM

I agree, any typical DC protection circuit will take some decens of ms to react. The question is... are that 80ms of the caps discharge really a problem?

Bgt 28th September 2006 07:15 PM

Quote:

Originally posted by ghemink
The DC detection itself also takes time, if the detection is too fast, it will trigger on low frequency heavy bass signals.
I use a double relay protection(2 contacts each side) in the output with a fast detection circuit. And yes, if I am testing 20hz at full output my relay will switch off but to me this is no problem. I never had a situation where the relay would switch off on music. If I just touch/put DC on the open input of the amp. with the speakers connected it would switch off without a click in the speakers. Protection input amp. is a differential pair with 1st order cutoff. Very sensitive.
Sound degradation....don't think so, anyway I can not hear it.
Safety is fantastic this way. I do quite a lot of testing with the amps. and tease them sometimes very badly, but the relays are so fast:-) No substitute for speed safetywise.

VEC7OR 28th September 2006 07:56 PM

I think that it would be best to implement cycle-by-cycle overcurrent protection, and DC protection which grounds the load on fault condition, and rails that are protected by fuses. IMO its not worth to cut the supply off - if you have a dead mosfet, then most likely damage is already done.

fredos 29th September 2006 04:18 PM

Simply short the speaker output with a TRIAC, and add fuse on supply rail! No relay on speaker line, a cheap 30A TRIAC will do the job and blow fuse on power supply if their any DC! That's the way I protect the speaker from my amplifier and that's work very well!

Have a nice day!

Fredos

www.d-amp.com


All times are GMT. The time now is 04:17 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2