Yet another detector configuration. The symmetry should provide equal sensitivity to +ve and -ve DC error signals, and a small change enables the circuit to work with bridged amplifiers. The filter cap is a low-voltage non-polarized electrolytic e.g. Farnell 1236657. This is just a sim, I haven't built it.
Attachments
Thanks
This is actually quite good!! I like the idea that the gain in both polarity is the same unlike mine that one side is common base that has no current gain.
The problem is that your circuit won't trigger until the DC is over 1.4V as you need to turn on the NPN/PNP pair before current will flow. I did it my way to ensure I trigger the speaker relay when it is over 0.7V. Simulation show reliable trigger at as low as 1V DC.
Yours is definitely faster, so that make up the higher trigger voltage as most dangerous is when the output drive a lot of voltage. At 1.4V into a 4 ohm speaker, the current through the speaker is 0.35A. I guess it's not going to burn the speaker. AND hopefully you know soon enough the amp is no longer working and turn it off.
Mine can take a little over 1 second to trip at 1V, but the current is lower to the speaker. I don't see a compelling reason to give up my design just yet.
Thanks
One other question, I saw both suggestion where input low pass filter compose of 47K and 5uF/10uF. I am using 10K and 200uF that gives much lower cut off frequency.
I got my numbers from simulation using a 40Vpeak 10Hz continuous sine wave at the input and make sure it won't trigger the circuit. Am I over killing this, that I should lower the value of the capacitor?
I got my numbers from simulation using a 40Vpeak 10Hz continuous sine wave at the input and make sure it won't trigger the circuit. Am I over killing this, that I should lower the value of the capacitor?
If equal gain on both polarities is important enough, then it could be achieved with just one Vbe drop if you're willing to use dual supply voltage - see attached cct.
As far as the filter values are concerned, I didn't think too much about them and my values may be way out. My approach would be to design the PCB to accommodate a range of resistors and capacitors, and choose what works best during testing.
The other thing which may be worth considering is the behavior which AndrewT described in post #575, as oscillation. This type of detector is always going to produce an intermittent signal when the input is very close to the trigger point. If your relay is fast enough to follow it (a solid state relay definitely would), it would produce intermittent connection/disconnection of the speaker to the amp, and could be quite unpleasant.
You may consider that this will happen over a very small range of inputs, so it can be ignored or you may prefer to handle it in your design. In the attached circuit it's handled reasonably well by Q6, R7, R8 and C4. The capacitor can be discharged quickly by the transistor, but takes a relatively long time to charge through R8. A more robust approach might use a monostable.
As far as the filter values are concerned, I didn't think too much about them and my values may be way out. My approach would be to design the PCB to accommodate a range of resistors and capacitors, and choose what works best during testing.
The other thing which may be worth considering is the behavior which AndrewT described in post #575, as oscillation. This type of detector is always going to produce an intermittent signal when the input is very close to the trigger point. If your relay is fast enough to follow it (a solid state relay definitely would), it would produce intermittent connection/disconnection of the speaker to the amp, and could be quite unpleasant.
You may consider that this will happen over a very small range of inputs, so it can be ignored or you may prefer to handle it in your design. In the attached circuit it's handled reasonably well by Q6, R7, R8 and C4. The capacitor can be discharged quickly by the transistor, but takes a relatively long time to charge through R8. A more robust approach might use a monostable.
Attachments
One thing I don't understand, we have two circuits here that shows the two voltage sources ( assuming they represent the output of the power amp to the speakers) stack in series. You cannot stack the output of the two power amps like this!!! They are all ground referenced!!! You have to have separate individual frontend to detect DC fault condition of each side of the stereo amp.
This is the best suggestion so far, I think this one and mine are very close, both trip at 0.7V. Both have about the same number of gain stage. We both have 3 gain stages on the low gain side and 4 gain stages on the high gain side. Plenty of gain.
I don't think it is avoidable for a certain window of uncertainty. I gave a lot of thoughts on using IC opamp, comparators, 555 etc. The idea you have to generate a low voltage supply( supplies) prevent me from pursue this route. Whether it produce an annoying sound is not the end of the world. when this happens, your amp is screwed. As long as it protect the speaker, it's all good.
The higher the gain of the circuit, the smaller the uncertainty window. As I said, both circuit have plenty of gain even though the gain is not the same for both polarity.
The two voltage sources represent a single (faulty) amp output. One is producing a sine wave and represents the intended output of the amplifier, the other produces a small DC offset. When the DC offset is active, the detector should trigger. So the two sources are just the way I chose to construct my test signal. I am a beginner at spice, and I expect there may be better ways to do it.
No it's a better way. I never thought of this!! I just use the single voltage source and change it to DC to test speaker protection, then change to 10Hz 40V to test the low frequency. My bad.
I was expecting protection circuit for both channel as in my schematic.
You design these? You got some good ideas. I actually like yours slightly better. I am debating back and fore. The back end is the same. You use one extra transistor and has to have the -ve rail supply. So it's more complicated. But you do buffer the input with a CE stage to achieve higher impedance. I already finished the pcb layout already, I am debating whether I want to incorporate your front end.
One thing I don't understand, why are you using C4? That will slow the response down.
Last edited:
I think I discover a potential problem of your circuit. You use the -ve supply. If the -ve supply fails, Q2 will never comes on and trigger the protection. But you can argue when -ve supply fails, the output likely drift +ve, the circuit should still works in protecting the speaker.
In my circuit, I don't depend on the -ve supply. And if the +ve supply fail, the relay will loss power and open and protect the speaker.
In my circuit, I don't depend on the -ve supply. And if the +ve supply fail, the relay will loss power and open and protect the speaker.
Last edited:
I don't see it that way.
The circuit detects the offset and starts to trigger. This takes some time.
The trigger is sent to the relay. It starts to operate. This takes some time.
The relay starts to open. This takes some time.
In the meantime the offset has increased from 1V to ~43V and the relay is trying to break the DC current that is now flowing through the speaker.
It is this time delay and the high DC currents that need to be broken that has spawned the Threads looking at SS relays. Shorter turn off times and DC breaking capability.
I see the detection as differentiating between valid LF signals and faulty offset voltage.
It seems to me that this differentiation of Valid and Fault that is critical to the protection that must be triggered or not.
All the circuits become very fast if the DC is high. the trigger time was way way shorter than 1 second if I put a few volts as fault voltage.
But the point is you have to look at the current vs time. It's the heat that burn the speaker, if the DC of 1 or 2 volts only last one or two second, the speaker is not going to smoke.
I think you have missed my point. Must be the way I write.
If the offset is assumed to be changing at a rate of 10V/us and you detect the offset as "faulty" when it reaches 1V, then 2us later the offset will be 21V
What has responded in your detection and trigger and opening sequence in that 2us?
Will the relay be open yet?
What if your relay takes 4us after the trigger event to open?
Or start at assuming 10V offset as your fault event. Assume 2V/us of offset rate of change.
After 10us you have an offset of 30V. Is the relay open yet?
If the offset is assumed to be changing at a rate of 10V/us and you detect the offset as "faulty" when it reaches 1V, then 2us later the offset will be 21V
What has responded in your detection and trigger and opening sequence in that 2us?
Will the relay be open yet?
What if your relay takes 4us after the trigger event to open?
Or start at assuming 10V offset as your fault event. Assume 2V/us of offset rate of change.
After 10us you have an offset of 30V. Is the relay open yet?
Well, no circuit can response that fast, the input low pass filter will slow you way down. You can only do your best and hope for the best. Based on result, the speaker protection circuit works to save the speaker.
That is correct - the circuits I posted are not complete solutions. They only address the detector part of the circuit. There was no consideration of power-up delay, rail voltage failure, or mains failure.
The purpose of C4 is described in post #587, but it is intended to prevent the pulses, which the detector sometimes produces, from reaching the speaker relay. These pulses occur when the input signal is very near to the point where the detector discriminates between "DC on" and "DC off".
Here again, my circuits are incomplete, C4 produces a slow rising voltage when the detector state changes from "DC on" to "DC off". This slow rising voltage would not be suitable to operate the relay directly - it would need to be followed by circuitry which operates and releases the relay at chosen voltage(s).
I'd suggest that you consider the circuit used in the "Speaker Turn On Delay and DC Protector Board Set (V3)". Even if you don't want to use the PCB, you could still benefit from the fact that by now the circuit should have been thoroughly debugged.
http://www.diyaudio.com/forums/diya...ker-turn-delay-dc-protector-board-set-v3.html
The circuit is shown in post #5 of that thread, and C3 in that circuit performs the same function as my C4.
Personally, I would also replace the mechanical relay in that circuit, with a solid state relay - which is, after all, the subject of this thread.
Unless the protection circuit also is triggered by a rail fault.
A DC condition and a rail fault often go together. I actually touched my servo
with one of my faults , I'm glad the circuit was monitoring both (DC and
rail).
OS
That's exactly the point, I don't detect rail fault because it will show up as DC at the output. If it does not show up at the output, it won't matter to the speaker and don't need to be tripped. This is speaker protection, not amplifier protection.