A simple speaker protection circuit

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,
do you have any comparators?
I think they would work here instead of an opamp.

The whole detection thing could be done with a pair of comparators set up as a window comparator. Just one dual 8pin chip and a switching transistor pair.

Worth breadboarding to see the difference (if any) from your proposal.
 
AndrewT said:
Hi,
do you have any comparators?
I think they would work here instead of an opamp.

The whole detection thing could be done with a pair of comparators set up as a window comparator. Just one dual 8pin chip and a switching transistor pair.

Worth breadboarding to see the difference (if any) from your proposal.

Interesting thought - in this context comparators are the same as op-amps, only their outputs are open-collector (not all, I know). So connect the outputs together for a wired-OR connection, use this to discharge the power-on delay capacitor...

The only "But" I can think of is that a single transistor following on from this delay-cap will get us back to an earlier problem, whereby if the transistor turns on slower than the relay power-saving capacitor discharges, the latter might not perform its function properly. But I've never played with this particular variation myself, so maybe I'm worrying about nothing...

Of course, you could add another dual comparator to do the power-on delay separately, and use the spare section as an over-temperature detector ;)

Russ, if you put the two resistors around the op-amp, you probably won't need an extra zener. Glad you're enjoy this :)

Mark
 
mhennessy said:
The only "But" I can think of

Just thought of another one - you have to create a -ve supply for the -ve threshold of the window comparator, not to mention the supply for the chip itself.

Not a big deal, but an extra diode/capacitor/dropper resistor...

There's also a chance that the max supply voltage the IC's run from could become an issue, requiring supply regulation. That's why I generally prefer transistors in these sorts of roles - no such limits if you choose the right device, and no cost-hikes either (unlike high-voltage op-amps!)

Mark
 
Nordic said:
I don't think the speed thing is such a big issue...

so haveing too fast a response is not nessecarily good is it?

Nordic,

As you know i know that circuit (MyRef) inside and out. This is actually significantly different in multiple ways.

The place we want things to happen very fast is when the relay switches when DC "is" detected. That is why the schmidt trigger is there as well as the R and C around the relay coil. The faster the relay witches to GND the faster the speaker is relieved from the DC.

The speed at which the DC is detected is determined by adjusting the RC filter. That would be adjusted to be reasonable for the application with some general starting value which should work for most amps. The triggering of the detector will be determined by the voltage and frequency of the signal coming from the amp's output.


Cheers!
Russ
 
Prototype working! :)

I made a simple PCB (will post later)

Here is the circuit I settled on.

Since the opamp will not swing all the way to GND I added the diodes and resistor at the output.

I test the circuit with 35V on input and it takes about 5ms for it to kick off the relay. :) Better than I hoped.

Cheers!
Russ
 

Attachments

  • sp1.png
    sp1.png
    10.9 KB · Views: 568
Hi,
now test it with AC signals.
try full voltage 20Hz and progressively lower the frequency until the filter lets through enough signal to trip the relay.

If satisfactory, then try to see how little DC is needed to trip the relay.

My targets would be
1. pass 10Hz full voltage and stay untripped.
2. detect 1.5Vdc and trip.
 
AndrewT said:
Hi,
now test it with AC signals.
try full voltage 20Hz and progressively lower the frequency until the filter lets through enough signal to trip the relay.

If satisfactory, then try to see how little DC is needed to trip the relay.

My targets would be
1. pass 10Hz full voltage and stay untripped.
2. detect 1.5Vdc and trip.

Hi Andrew. :)

Seems to easily pass those tests.

Been running only with 35V rails though, will try higher rails later. I can actually get it below 3hz without issue at these rails.

1.5Vdc does indeed trip it, after a half second or so. (simple AA battery enough to test this).

The nice thing is that the Schmitt Trigger with positive feedback does its job nicely keeping the relay from chattering in the gray areas.

Cheers!
Russ
 
Hi Russ,
now for a bit of tuning.

In my opinion, we want FAST detection with minimum false triggering on music and sound effects.

The DC detect will get faster as you reduce the input RC time constant.

Reducing the cap will alter both the DC time delay AND increase the minimum frequency that passes without tripping.
Your 500mS and 3Hz is maybe a good compromise, but try pushing this to become a little faster.

Finally, what about latching?
What if a semi goes short circuit and passes DC current to the speaker? The relay pulls in? What next? It must not reconnect the speaker while the fault still exists.

Will the circuit hold the speaker relay open at amp start up and delay connection until the amp output levels have settled (in those first few tens of mS)?
 
Russ White said:
Here is the supply circuit I am thinking of using. :)

Easily tuned for 110 or 220V mains.

No dependency on amp transformer at all. :)

The circuit will be adjust to provide a constant load even when the speaker relay is off.

Cheers!
Russ

If I understand the proposed power supply right, the negative terminal is one diode drop away from the line voltage. The detector ties the output ground of the amp to the negative DC of the supply. Assuming the chassis of the amp is tied to the output ground, that will put the chassis of the amp one diode drop away from the line voltage and pose a serious health risk. Am I missing something?
 
d3imlay said:


If I understand the proposed power supply right, the negative terminal is one diode drop away from the line voltage. The detector ties the output ground of the amp to the negative DC of the supply. Assuming the chassis of the amp is tied to the output ground, that will put the chassis of the amp one diode drop away from the line voltage and pose a serious health risk. Am I missing something?


You are very right. :) It is probably not advisable at all as it is. I use this scheme for my soft start, but In that situation a relay and amp's transformer are isolating the circuit from the user. So using this circuit the same way may not work without some safety precautions, like fast blow 500ma fuse or something. The draw on the circuit is only about < 23ma.
 
AndrewT said:
Hi Russ,
now for a bit of tuning.

In my opinion, we want FAST detection with minimum false triggering on music and sound effects.

The DC detect will get faster as you reduce the input RC time constant.

Reducing the cap will alter both the DC time delay AND increase the minimum frequency that passes without tripping.
Your 500mS and 3Hz is maybe a good compromise, but try pushing this to become a little faster.

Finally, what about latching?
What if a semi goes short circuit and passes DC current to the speaker? The relay pulls in? What next? It must not reconnect the speaker while the fault still exists.

Will the circuit hold the speaker relay open at amp start up and delay connection until the amp output levels have settled (in those first few tens of mS)?

Hi Andrew,

The input filter design is always a compromise, and the best way to improve on this is to choose a 2 pole input filter - that's what the Velleman kit does. One has to balance this against the increase in complexity - this is one area where reliability is key!

I've already spoken about latching (post #10), and I would again suggest that it is not necessary. Can you think of a situation where the DC offset will dissapear once the speaker has been disconnected? I can't (that's not to say it's impossible, of course!). But, looking at the latest schematic, it would be reasonably straightforward to reconfigure Q1/Q2 as a bistable latch...

Yes, this circuit does provide power-on delay - R4/C2 do this. Much has been said about the need to quickly discharge C2 in the event of a fault - hence the PNP EF (Q3)

Best regards,

Mark
 
Hello Mark,

Thanks for articulating those points. I agree with you on these.

Have you looked at the optoisolated version? It has one key advantage that the relay PS is completely isolated from the audio signal(and GND). But it has one key tradoff, that is that the turn on voltage for the LEDs is around 1V as oppose to .6V for the transistors. Keep in mind, that is just the SPICE circuit, the real one will have some additional parts (like the discharging diode for the delay cap).

Cheers!
Russ
 
Russ White said:
OK, I think I have an idea, I could separate the protection PS from the AMP PS via an optoisolator. That should keep every thing well isolated. I will see what I can work out. Also the case should be connected to mains earth.

Hi Russ,

I've used capacitive-dropper power supplies before, and they're a mixed bag. Certainly not for the inexperienced constructor - so while this is fine for your own use, I'm not sure if I'd do it in a kit... The one published project on my site that uses one also has several warnings and disclaimers!

It would certainly be easier to use the AC from the transformer, as earlier suggested. If you're worried about accomodating a range of supply voltages, simply change the relay drive to a currect source. The rest of the ciruit can be powered from a zener and dropper resistor...

The main worry I have with the latest circuit is the detector sensitivity. The LEDs need around around 1-2mA (perhaps less because the collectors don't need to sink much). But 1-2mA through a 20K resistor suggests a voltage drop of 20-40V. So you might need to buffer the filter output, which creates more complexity.

Having said that, opto-couplers can be really usefuly building blocks, even when you don't need to isolation properties. I've seen this circuit used plenty of times before :)

What else? I like the connection of R1, and was going to suggest it myself earlier (if you wanted to save a 1n4148 when not using the optional DC fault LED). It makes the base look like a current source (0.6/500K - 1.2uA which is probably a bit low here for good noise immunity - earth those opto-bases!). FWIW, R10 can be omitted. R4 needs the parallel diode to discharge C2. Not sure why you have D5? Consider replacing those series diodes with a zener or LED to reduce component-count (if it matters). The two output transistors are obviously there to keep the supply voltage roughly constant, but this strikes me as wasteful. A consequence of the capacitive-dropper PSU. These could easily form Andrew's latch :)

Not sure why Q1/2 are different types - they both switch the same load. R12 is a bit high, probably... Never used an LT1224 before, but it looks massively over-spec'd to me. Was that selected as convenience in the simulator? Personally, a good old 741 is what I'd use here - basically anything that's cheap, as a circuit like this shouldn't make any demands of its components...

Don't know if this helps?

Mark
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.