Oh, gosh, i had exactly the same idea 20 years ago, and that's how work my own protection since those years.
A working amplifier must have the same signal (scale apart) out and in.
And this gives a lot of advantages. it protect from everything: Clipping, HF, oscillations as well as DC and short circuits on the loudspeaker line. And it is instant.
The only problem is, because the complex charge witch is the loudspeaker and its motional signals , and because the phase switch that the amplifier can introduce, we need to adjust a gate for the relay does not switch-on at each drum kicks.
On my side, i do not introduce any temporal delay before protection switch on. Just my circuit authorizes the good amount of errors from the loudspeaker line. But i have a delay witch keep the protection on for 3s after the cause of the trouble had disappear .
By default, on my circuit, the Loudspeakers are not switched, and i have an additional circuit witch wait 5sec after power on operations before the relay can switch them on. And cut the relay instant when power off.
In fact, when you power off, first the relay cut the loudspeakers, then the power relay switch off the power.
When you power on, there is a resistance in serial with the PSU condensers to avoid huge transients. After 2sec, a relay short circuit this resistance. After 5 more seconds, the power is send to the loudspeakers protection relays. All that use only a quad OP AMP, 3 transistors, few resistances and condensers, and 3 LEDs shows the sequence and states (power on > condensers on > loudspeakers on).
Works like a charm, with not a single "clic" during all those years.
Member
Joined 2009
Paid Member
I'm building a new system, so i will post the shematic and circuit board design within the next month.
It will be here:
http://www.diyaudio.com/forums/digi...x-ways-enclosures-class-d-system-project.html
If you are in a hurry, i can open my actual amp to recover my original schematic with precise values, but i feel lazy ;-)
After building the DCB1 buffer (and not being able to go back due to the big improvements relative to my previous line buffer that uses a cap in the output) I am worried that a PSU failure might cause excess offset and kill my speakers (or worst).
I read a lot and only found designs and ideas of protections that use relays on the speakers.
Is it possible to implement one of these designs to control relays between the buffer output and the power amp ?
I am using a complex simetrical PSU and am afraid of possible failure (due to housekeeping damages, and not to the implementation or design).
I read a lot and only found designs and ideas of protections that use relays on the speakers.
Is it possible to implement one of these designs to control relays between the buffer output and the power amp ?
I am using a complex simetrical PSU and am afraid of possible failure (due to housekeeping damages, and not to the implementation or design).
you can fit a mute to the output of the DCB1,
or fit a mute to the input of your power amps.
or fit a cutoff in the power supply rails of the power amps,
or fit a crowbar to the output of the power amps.
All of these avoif placing any sound interfering relay in the speaker lead.
Finally a cutoff of amplifier output.
There is a recent discussion on replacing relays with FETs to improve the reliability of the cutoff and avoid some of the sound degrading properties of relays.
or fit a mute to the input of your power amps.
or fit a cutoff in the power supply rails of the power amps,
or fit a crowbar to the output of the power amps.
All of these avoif placing any sound interfering relay in the speaker lead.
Finally a cutoff of amplifier output.
There is a recent discussion on replacing relays with FETs to improve the reliability of the cutoff and avoid some of the sound degrading properties of relays.
The purpose of a protection is to... protect. An amplifier can broke itself for a lot of reasons, then burn your valuable speakers. So there is two places you can only put relays, it is between power supply and amp, or between amp and loudspeaker. The first one can be ok in case of faillure, but not protect against power-on/off noises and DC . Devinitively, the best place is between amp and Loudspeakers. Of course, you can add an other relay in the Amp input, but why ? A well designed amp is able to manage any kind of signal at the input up to peak +-15v if no charge. And it is the max that can feed your preamp. More, you need to continue to compare the signal when the protection is "on", that suppose your amp continue to amplify the input signal: the protection will stay "on" until the input signal returns to a normal level/frequency range.
Quad fitted a triac-based crowbar circuit to the Quad 405. It uses a Silicon Bilateral Switch to trip the triac at about 10V or so, with an RC delay. I don't know where to buy an SBS in the UK, but the data sheet for the device gives a transistor equivalent. I built it, and it seems to work fine. However, from what I can tell, I notice that Quad now doesn't bother with such a circuit, using a floating ground PSU topology that inherently protects against DC faults (although I think the energy in one of the caps has to discharge through the speaker).
Am I correct in thinking that series relays are not so good for this application because the contacts have a tendency to weld closed when attempting to open them during a high surge of DC?
Am I correct in thinking that series relays are not so good for this application because the contacts have a tendency to weld closed when attempting to open them during a high surge of DC?
I don't think so, if the relay is correctly dimensioned and their contact well protected. But you can think about a more complex circuitry: Mos fet for switching followed by the relay, relay first followed by the Mosfet when protection switch on. Here the relay will be there only for reducing the serial impedance of the MosFet and will not deal with noticeable current or tension. I had tried-it. Not a difference in sound for an increased complexity, and never a problem of relay during 2 years on my personal 2x140W amp with a lot of output short circuit tests: forgotten on my side.
A simple comparator between input and output signal of the amp does-it instant, as well as HF, clipping, switch-on " clocs" and DC errors. It is global !
few posts before in this thread, an other example :
www.diyaudio.com/forums/elektor/197552-current-feedback-mod-electktotvtimers-crescendo.html
As i said, i will publish mine as soon i get time to recopy my own (old) solution on the same principle.
Be careful which op amp you use, some invert the output if you go anywhere near the supply rails !!!
Here my schematic. Their is no risk the signal can go near the power rails because it is just atthe level of the input (near 1V). On my amp, as an example, if the amp goes to his rail level (70V) the signal will be reduced by a factor of 30, means 2V, far away from the 12v of the Op amps supplies.
http://www.diyaudio.com/forums/digi...sures-class-d-system-project.html#post2731758
So is the preferred approach to break the connection to the speakers, or to short them out? I am not convinced by the series relay approach.
Looking in detail at a relay data sheet, DC breaking capacity is specified for a resistive load. It seems to fall off rapidly as the voltage exceeds 30V. For a typical relay, at 50V DC, breaking capacity is <10% of the nominal current rating (so a 20A relay is rated for <2A at 50V DC).
Just what voltage and current will be present if the amp goes wrong? I'm presuming that it could be 50-60V for a powerful amp, and >10A through the speaker coil. Can you be really confident that you're not going 'tack weld' the contacts even with a large relay and snubber across the contacts?
Looking in detail at a relay data sheet, DC breaking capacity is specified for a resistive load. It seems to fall off rapidly as the voltage exceeds 30V. For a typical relay, at 50V DC, breaking capacity is <10% of the nominal current rating (so a 20A relay is rated for <2A at 50V DC).
Just what voltage and current will be present if the amp goes wrong? I'm presuming that it could be 50-60V for a powerful amp, and >10A through the speaker coil. Can you be really confident that you're not going 'tack weld' the contacts even with a large relay and snubber across the contacts?
A 100W into 8ohm amplifier runs from ~+-50Vdc supply rails.
If there is a serious output fault that sends DC to the speaker terminal, then expect most of that 50V to appear at the terminal and expect the DC resistance of the 8ohm speaker to be ~ 6r0.
The resistances between the PSU and output terminal and the return lead could be ~0r5
The current flowing till the smoothing caps start to discharge (just a us, 0.001ms or so) is ~7A @ 50Vdc.
It takes a mighty Power relay to reliably break that value of DC current at that voltage.
The more powerful the PSU the longer the caps stay recharged to a high (damaging) voltage. Result:- Relay contacts supremely welded or completely burnt out during that one incident.
If there is a serious output fault that sends DC to the speaker terminal, then expect most of that 50V to appear at the terminal and expect the DC resistance of the 8ohm speaker to be ~ 6r0.
The resistances between the PSU and output terminal and the return lead could be ~0r5
The current flowing till the smoothing caps start to discharge (just a us, 0.001ms or so) is ~7A @ 50Vdc.
It takes a mighty Power relay to reliably break that value of DC current at that voltage.
The more powerful the PSU the longer the caps stay recharged to a high (damaging) voltage. Result:- Relay contacts supremely welded or completely burnt out during that one incident.
Member
Joined 2009
Paid Member
The belt n braces approach would be to have something in series - either at the speaker output or the power rails to the output power devices. And have something that shorts the output to ground (e.g. simple Triac circuit) - which will destroy the amplifier if no series breaker is used.
And instead of relay, use power MOSFETs.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.