bear said:That's why you ought to use fuses when testing, and before you are confident that your amp/speaker combo is stable over time...
Or have an outboard "protection" scheme, if that is what you prefer...
I still prefer big CONTACTORS that have a manly KER-THUNK on pull in... no tiny "tink" sounds here!
_-_-bear
The 8 amp fuses didnt blow because it took less than 8 amps DC to fry the speakers. 8 amps AC is fine for the speakers because it is an inductive load. The inductive load always gives back most of the current into the amp a DC load does not.
Thats why when you watch class b distortion it is often offset on the scope. Read up on power factors if your not sure what I mean.
I like a relay with delay-on so that you never get a pop & other noises out of the speakers. As long as the relay can handle the current, and the coil lasts, then that's all I need. A relay couldn't make enough difference in the sound to not have them. 240V relays do the job nicely.
Now on a 12V BTL amp, you just ground the mute pin at startup/shutoff, and you get no pop and no relays needed
It would be cool if that mute function could be put into a discrete amp.
Now on a 12V BTL amp, you just ground the mute pin at startup/shutoff, and you get no pop and no relays needed
It would be cool if that mute function could be put into a discrete amp.
EWorkshop1708 said:I like a relay with delay-on so that you never get a pop & other noises out of the speakers. As long as the relay can handle the current, and the coil lasts, then that's all I need. A relay couldn't make enough difference in the sound to not have them. 240V relays do the job nicely.
Now on a 12V BTL amp, you just ground the mute pin at startup/shutoff, and you get no pop and no relays needed
It would be cool if that mute function could be put into a discrete amp.
I use a microcontroller to control a relay on my amp.
On power up it waits 3 seconds for the amp to settle down before turning on the relay.
The micro also looks at the amp output voltage.
If it stays DC for more than half a second, i.e. power output transistor blown, it switches the relay off to save the speakers.
G.Kleinschmidt said:Have those who worry about relay contact distortion ever considered using a relay to switch the supply rails of the power output stage instead of the speaker output?
Does anyone do this? It makes awfully good sense.
I've also toyed with putting a signal relay on the driver gates, mainly as a mute, such that when there's no power applied, the gates are pulled to ground. Then you'd only need a little bitty relay. It'd be inside the feedback loop too.
Of course it'd be no use for DC protection.
suzyj said:
Hi Suzy
I've got an Elektor magazine with a design that switches the rails to the output devices with (I think) a pair of MOSFET's. But they are turned on gradually (I think with the audio input muted also) so that the feedback loop can track the output voltage to prevent the power on thump.
But I guess shorting the VAS output / OPS gates to ground would be another way to prevent the power-up thump if used in conjunction with relay or MOSFET switching of the output device rails for DC fault protection.
Cheers,
Glen
This is my own experience :
Most relays I have come across have a minimum contact resistance of around 50 mohm when new. They do deteriorate with time, depending what they are switching and how many times. The one exception is probably mercury wetted relay, and you can get 20A+ mercury wetted relays in the USA.
The parallel-switch relay that Jan was referring to used to be quite common for automobile, e.g. Bosch 0 332 002 250 which will switch 50A in its main contact. These are not use very often these days, as the relay contact design has improved making the low-current pre-contact not necessary. But I think Jacco is expert on this.
I have tried both. But these days I use 2x 70A automotive relays in parallel shunting the speaker output. If you have a DC at the output larger than say 2V, the output devices are most likely to be defect, and I want to replace them anyway. So I only care about the speaker and not transistors. Plus that the protection circuit I use switches the speaker (shunt) protection on and the power supply off at the same time. Only need to drain the caps then, as I do not want the amp to see the power supply switch directly.
Because I am using a real switch rather than a triac or any semiconductor device, I do not expect them to add any distortion. The output from amp goes directly yo speaker cables.
And no, I have no distortion measurements to prove. Am still building Bob's distortion analyser.
Patrick
Most relays I have come across have a minimum contact resistance of around 50 mohm when new. They do deteriorate with time, depending what they are switching and how many times. The one exception is probably mercury wetted relay, and you can get 20A+ mercury wetted relays in the USA.
The parallel-switch relay that Jan was referring to used to be quite common for automobile, e.g. Bosch 0 332 002 250 which will switch 50A in its main contact. These are not use very often these days, as the relay contact design has improved making the low-current pre-contact not necessary. But I think Jacco is expert on this.
I have tried both. But these days I use 2x 70A automotive relays in parallel shunting the speaker output. If you have a DC at the output larger than say 2V, the output devices are most likely to be defect, and I want to replace them anyway. So I only care about the speaker and not transistors. Plus that the protection circuit I use switches the speaker (shunt) protection on and the power supply off at the same time. Only need to drain the caps then, as I do not want the amp to see the power supply switch directly.
Because I am using a real switch rather than a triac or any semiconductor device, I do not expect them to add any distortion. The output from amp goes directly yo speaker cables.
And no, I have no distortion measurements to prove. Am still building Bob's distortion analyser.
Patrick
With all due respect to everyone who has posted, I think we're missing a very important point here. The relay serves two functions. Turn on/off pulses and speaker protection from amplifier failure.
To state the obvious, in order for the protection circuit to work it usually senses a DC component on the output. The not so obvious is that a filter is used to ignore the audio, say above 20Hz. This represents a minimum time delay of 50ms. Add in another 10-20 ms for the relay to drop out and you have a minimum of 60-70 ms time delay. So the speaker sees 80VDC before the relay ever trips. Fuses don't help much because they're not rated for 80VDC. 250VAC no problem. When the fuse starts to blow it arcs initially. With AC the arc extinguishes every half cycle so the fuse clears. Not so with DC. It arcs until everything is vaporized. That's why glass fuses turn black. Look at your fuses, they're not rated for the DC in you're amplifier.
Here is the real problem. If the amplifier failure mode is such that there is 80VDC on the output, the relay has to break 80VDC at about 10 amps. There isn't a relay made, that will fit in most any amp, that was designed to break DC. Check you're contact ratings. They're only rated for 24VDC. The relay will arc and weld under these conditions until the fuse blows. The fact that many amps have never had welded contacts upon failure says nothing about the relay. It just means the relay didn't have to break DC.
About a year ago I did some testing to determine failure modes of an amplifier. I used a Phase Linear 400 as the test amp because I intended to design an output relay for it. I used only resistive load for the test and had a hard time making it fail. I know all about power factor and phase shifts and so on, but it's not easy to add inductive loads on a bench at 250 watts. Any how I had to remove all but one of the outputs in the positive rail. Case temp reliably ran up to 145 degrees C before failure. I triggered the scope on collector current using a small current transformer. In each case when the amp failed there was no significant DC on the output. So the fault was through the output stage not the load. On most, but not all failures, a transistor in the negative rail let go also. If any one is interested I can post the scope photos if I can still find them.
If the amplifier failure mode is in the front end and sends DC to the output then the relay is in trouble. I'm sure that's why Peavey used the triac plus the relay. I've only repaired a few CS400's. One of them had a blown triac. It's likely the owner tried bigger fuses/foil hoping it would fix itself.
I saw one post about using relay contacts on the rails. That won't work for the same reasons I mentioned above. I don't like mosfets or BJT's as a switch here as I'm not convinced they can handle the fault current without shorting. Shorting out the gates with a relay would work for turn on/off thumps but give no protection against shorted outputs. I use a form C contact (one normally open and one normally closed contact) in series with the speaker. When the relay drops out it shorts across the speaker and disconnect itself from the amp. See attachment.
To state the obvious, in order for the protection circuit to work it usually senses a DC component on the output. The not so obvious is that a filter is used to ignore the audio, say above 20Hz. This represents a minimum time delay of 50ms. Add in another 10-20 ms for the relay to drop out and you have a minimum of 60-70 ms time delay. So the speaker sees 80VDC before the relay ever trips. Fuses don't help much because they're not rated for 80VDC. 250VAC no problem. When the fuse starts to blow it arcs initially. With AC the arc extinguishes every half cycle so the fuse clears. Not so with DC. It arcs until everything is vaporized. That's why glass fuses turn black. Look at your fuses, they're not rated for the DC in you're amplifier.
Here is the real problem. If the amplifier failure mode is such that there is 80VDC on the output, the relay has to break 80VDC at about 10 amps. There isn't a relay made, that will fit in most any amp, that was designed to break DC. Check you're contact ratings. They're only rated for 24VDC. The relay will arc and weld under these conditions until the fuse blows. The fact that many amps have never had welded contacts upon failure says nothing about the relay. It just means the relay didn't have to break DC.
About a year ago I did some testing to determine failure modes of an amplifier. I used a Phase Linear 400 as the test amp because I intended to design an output relay for it. I used only resistive load for the test and had a hard time making it fail. I know all about power factor and phase shifts and so on, but it's not easy to add inductive loads on a bench at 250 watts. Any how I had to remove all but one of the outputs in the positive rail. Case temp reliably ran up to 145 degrees C before failure. I triggered the scope on collector current using a small current transformer. In each case when the amp failed there was no significant DC on the output. So the fault was through the output stage not the load. On most, but not all failures, a transistor in the negative rail let go also. If any one is interested I can post the scope photos if I can still find them.
If the amplifier failure mode is in the front end and sends DC to the output then the relay is in trouble. I'm sure that's why Peavey used the triac plus the relay. I've only repaired a few CS400's. One of them had a blown triac. It's likely the owner tried bigger fuses/foil hoping it would fix itself.
I saw one post about using relay contacts on the rails. That won't work for the same reasons I mentioned above. I don't like mosfets or BJT's as a switch here as I'm not convinced they can handle the fault current without shorting. Shorting out the gates with a relay would work for turn on/off thumps but give no protection against shorted outputs. I use a form C contact (one normally open and one normally closed contact) in series with the speaker. When the relay drops out it shorts across the speaker and disconnect itself from the amp. See attachment.
> This represents a minimum time delay of 50ms.
Well, try to put a 80V step input to a simple RC filter of 20Hz, and I bet it does not take 50ms to get to 2V at the filter output. 2V is a typical level to trigger the protection circuits. You can always set it lower.
> The relay will arc and weld under these conditions until the fuse blows.
That is another benefit of using the relay to shunt the speakers. So it welds, so what. As long as the speakers are saved, the cost of new relays are negligible.
And relay on the rails is required to break rather than make in case of fault.
Patrick
Well, try to put a 80V step input to a simple RC filter of 20Hz, and I bet it does not take 50ms to get to 2V at the filter output. 2V is a typical level to trigger the protection circuits. You can always set it lower.
> The relay will arc and weld under these conditions until the fuse blows.
That is another benefit of using the relay to shunt the speakers. So it welds, so what. As long as the speakers are saved, the cost of new relays are negligible.
And relay on the rails is required to break rather than make in case of fault.
Patrick
EUVL said:
For the delay I'll check the test data. I used square, sine and triangle waves from 1/10hz to 100hz.
I completely agree on the relay shunting the speakers. I revised my previous post and added a pdf. I used a socket for the relay so that it could be replaced.
For a relay in the rails I don't know of any that (1) will break 80 VDC at 10 amps, and (2) is smaller than a grapefruit, and (3) costs less than $100. Please, somebody prove me wrong on this.
> For a relay in the rails I don't know of any that (1) will break 80 VDC at 10 amps, and (2) is smaller than a grapefruit, and (3) costs less than $100.
Tyco Power Relay F7 (Catalog 1308028) will shutdown safely at 40V 10A. I use bridged amps, so I do not need 80 VDC. I fact I barely need 40V.
It is MUCH smaller than a grapefruit, and cost much less than $20.
Perhaps not a solution for your specific needs, but still a solution for many, I think. Certainly for me.
Patrick
Tyco Power Relay F7 (Catalog 1308028) will shutdown safely at 40V 10A. I use bridged amps, so I do not need 80 VDC. I fact I barely need 40V.
It is MUCH smaller than a grapefruit, and cost much less than $20.
Perhaps not a solution for your specific needs, but still a solution for many, I think. Certainly for me.
Patrick
suzyj said:
Hi Suzy,
This makes total sense as far as avoiding the disaster of shorted output transistors. The only concern, apart from a bit of rail voltage loss, is that this function alone does not provide the speaker muting function on power-up and power down. If other means are used to avoid thumps (or if the amplifier naturally does not create such thumps), then this approach makes a lot of sense.
Cheers,
Bob
EUVL said:
Hi Patrick,
These are all very good points. The use of a shunt device to prevent disastrous conditions for the speaker is a very good one, as long as the amplifier itself will survive a direct short across its output (it should be designed that way anyway).
This would seem to be the cleanest approach sonically, as long as any thump problems are handled by other means. The use of the Triac is very tempting, but I have not done measurements on such an approach.
Cheers,
Bob
d3imlay said:
This is an excellent and thoughtful post. The problems of protecting the speaker in a fault condition without compromizing sonics are truly vexing. Your reminder about a relay's limited ability to break dc is well-taken. If I was going to break the rails, I would definitely do it with high-current MOSFETs. They should have no problem breaking the rail DC.
Cheers,
Bob
"Let's assume you are right"
1) I am right, in the sense that the DC opening rating is not a magic thing, but is only of import IF there is inductance in the circuit.
Wherelse do you think an arc can come from?
2) Of course there is inductance in speakers, but nothing like electric motors and solonoids which is where the ratings for relay contacts are biased towards.
Cliff
1) I am right, in the sense that the DC opening rating is not a magic thing, but is only of import IF there is inductance in the circuit.
Wherelse do you think an arc can come from?
2) Of course there is inductance in speakers, but nothing like electric motors and solonoids which is where the ratings for relay contacts are biased towards.
Cliff
> The use of the Triac is very tempting, but I have not done measurements on such an approach.
With automotive relays at $6 a piece, mountable on sockets for quick change, and 70A rating, I would not bother with anything else (for the shunt application). If the contact welds, so much better.
As you quite rightly said, a good amp should be designed to be short circuit safe.
> If I was going to break the rails, I would definitely do it with high-current MOSFETs.
Yes indeed. One convenient way which I also use is the so-called cap multiplier. If the DC detection is also used to shunt the gate of the MOSFETs in a cap multiplier using a small BJT or even a relay, this will provide a convenient means to shut off the power with only one extra component. Any high current MOSFETs such as 2SK2955/2SJ554 will handle that with ease.
Patrick
With automotive relays at $6 a piece, mountable on sockets for quick change, and 70A rating, I would not bother with anything else (for the shunt application). If the contact welds, so much better.
As you quite rightly said, a good amp should be designed to be short circuit safe.
> If I was going to break the rails, I would definitely do it with high-current MOSFETs.
Yes indeed. One convenient way which I also use is the so-called cap multiplier. If the DC detection is also used to shunt the gate of the MOSFETs in a cap multiplier using a small BJT or even a relay, this will provide a convenient means to shut off the power with only one extra component. Any high current MOSFETs such as 2SK2955/2SJ554 will handle that with ease.
Patrick
O.K. Please help me understand something. The attached link is for the PCLH relay I'm using (p713).
PCLH relay
Why the different ratings for AC & DC
PCLH relay
Why the different ratings for AC & DC
AC current and voltage has zero crossings, which allows the arc during break to extinguish quickly. With DC the arcing only stops when the distance increases or voltage conditions change which might take a longer while, allowing a large amount of contact material to vaporize in the meantime... not good for the relay...
- Klaus
- Klaus
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.