@jean-paul The sequence of events I would worry about with 24 V DC rated relays is this:
1. Someone builds an amplifier with +/- 45 V supplies and an output relay with contacts rated for a maximum DC switching voltage of 24 V
2. The amplifier works fine for decades until one bad day an output transistor becomes a short circuit, putting 45 V on the output
3. 100 ms or so later, the DC protection tries to switch off the output relay
4. As the relay can only handle 24 V, it fails, and it happens to fail with contacts shorted
5. The loudspeakers burn out.
Unless a relay expert can assure me that step 4 can't happen because relays always fail open when they are subjected to too high DC voltages at switch-off, I would strongly prefer a relay with a maximum DC switching voltage greater than the supply voltage, like the one HRDSTL has chosen.
1. Someone builds an amplifier with +/- 45 V supplies and an output relay with contacts rated for a maximum DC switching voltage of 24 V
2. The amplifier works fine for decades until one bad day an output transistor becomes a short circuit, putting 45 V on the output
3. 100 ms or so later, the DC protection tries to switch off the output relay
4. As the relay can only handle 24 V, it fails, and it happens to fail with contacts shorted
5. The loudspeakers burn out.
Unless a relay expert can assure me that step 4 can't happen because relays always fail open when they are subjected to too high DC voltages at switch-off, I would strongly prefer a relay with a maximum DC switching voltage greater than the supply voltage, like the one HRDSTL has chosen.
In the above situation it will open the contacts which might draw an arc. If the PSU can not support this arc, it will extinguish. If it can, the relay will burn out.
99.999% of the time when an output relay is needed is at switch ON. Something has already upset the amp, and it comes up with the output stuck to the rail. Perhaps a shorted/open transistor, cold solder joint, or just a bad electrolytic cap somewhere. The DC protect relay NEVER closes in the first place and the amp comes up in “protect” mode with no sound. How many threads have we seen here where a newbie is asking what to do in this exact situation?
When an output transistor actually shorts, it won’t put DC to the speaker for very long. DC feedback will cause very high current in the “good” side. That much short circuit current will draw down the supply, or blow fuses or other components. In most cases the current gets interrupted somehow, pretty quickly.
You could short or open a VAS transistor or otherwise lose feedback - but those are less common failure modes while operating. If it does happen the DC protect relay will attempt to intervene. Might be successful, might weld shut. If the fault current is not ridiculously high - which would be the case when simply driving a speaker and you lose DC feedback, it stands a fairly decent chance of opening properly. The bigger that relay is the better.
When an output transistor actually shorts, it won’t put DC to the speaker for very long. DC feedback will cause very high current in the “good” side. That much short circuit current will draw down the supply, or blow fuses or other components. In most cases the current gets interrupted somehow, pretty quickly.
You could short or open a VAS transistor or otherwise lose feedback - but those are less common failure modes while operating. If it does happen the DC protect relay will attempt to intervene. Might be successful, might weld shut. If the fault current is not ridiculously high - which would be the case when simply driving a speaker and you lose DC feedback, it stands a fairly decent chance of opening properly. The bigger that relay is the better.
The loudspeaker resistance would be in series with the contact, so you would have a 45 V supply, a 6 ohm or so resistance and the relay contacts. Is that enough to sustain the arc?
If the loudspeaker burns out before the relay contact does, the attempt to protect the loudspeaker has been unsuccessful.
That's interesting. So even a relay with too low maximum DC switching voltage can reduce the chances of blown up loudspeakers substantially.
Good point, although it depends on where the fuses are. For example, suppose you have one in the positive rail and one in the negative rail, the amplifier develops a short to the positive rail, the negative side of the amplifier tries to correct it and the fuse of the negative rail blows.
It sounds to me like a relay with a DC switching voltage rating greater than the supply voltage is to be preferred, although depending on what exactly happens, there is a good chance that one with a lower rating can also protect the loudspeakers.
Knowing that relays are chosen wrongly often and we seem to choose high current rated versions with the right contact material chances are very small that the relay will go kaputt when the final breakdown of the device suddenly happens also while already playing. Of course it helps to find one that can disconnect 10A @100V DC but I wouldn't not worry too much. It will not fail the first time it would need to disconnect DC and high current as it has around 3 to 7 Ohm mainly resistive load. As said I use various Schrack types which don't have DC voltage mentioned in the datasheet (or a lowish value like 5A). A good habit is to use the 2 sets of contacts in parallel so 2 relays in total. Not a single one has failed after replacing the old relay since then. The gold alloy/gold plated ones did fail regularly as explained earlier on. Suffered from contact pitting from day 1. These will stick with DC disconnect at high current as the contact material is too weak. We call it "welding". BTW capacitive loads are killers as usual.
wg_ski sums it up pretty complete. The times I witnessed a sudden failure was indeed at power on (often a broken connection of one side of the rectifier or a MOSFET defective). Since it is custom to find complex solutions for simple task one can design complex stuff and extremely overthinking but it is not needed when the basic component is sturdy and well chosen. It is that simple.
Challenge: check standard relays in various amplifiers of A brands, older ones and recent ones. Then compare specifications with what is chosen for the device by the OP....
Severe OCD solution: have a second tungsten contact relay shorting the loudspeaker itself at DC detection with 0.5 second delay. When the speaker relays contact are welding nicely the second relay will short the loudspeaker and take the current, fuses will definitely blow and output stages will be executed but the loudspeakers survive. No those fuses you know, they might not burn. Can we find a complex solution for those? And what about when the regulator for the protection board fails? Two PSU's in redundant configuration? Best battery system?
wg_ski sums it up pretty complete. The times I witnessed a sudden failure was indeed at power on (often a broken connection of one side of the rectifier or a MOSFET defective). Since it is custom to find complex solutions for simple task one can design complex stuff and extremely overthinking but it is not needed when the basic component is sturdy and well chosen. It is that simple.
Challenge: check standard relays in various amplifiers of A brands, older ones and recent ones. Then compare specifications with what is chosen for the device by the OP....
Severe OCD solution: have a second tungsten contact relay shorting the loudspeaker itself at DC detection with 0.5 second delay. When the speaker relays contact are welding nicely the second relay will short the loudspeaker and take the current, fuses will definitely blow and output stages will be executed but the loudspeakers survive. No those fuses you know, they might not burn. Can we find a complex solution for those? And what about when the regulator for the protection board fails? Two PSU's in redundant configuration? Best battery system?
Last edited:
Well, the solution I propose is as simple as using a relay that's designed to switch the DC voltage it needs to switch - as the OP is doing.
@jean-paul I'm amazed that you have witnessed so many amplifiers failing that you use the word "often". Were you professionally involved in amplifier lifetime testing, quality assurance or failure analysis?
@jean-paul I'm amazed that you have witnessed so many amplifiers failing that you use the word "often". Were you professionally involved in amplifier lifetime testing, quality assurance or failure analysis?
Hi yes/no. I repaired Japanese stuff as a profession and since then I am hardcore hobbyist in numbers if that was not clear. I am not exxagerrating when I state that I have replaced several tens (think of hundreds, never counted them) of relays of Japanese better quality audio stuff. It turned out to be a persistent design failure at Technics, Akai, Pioneer, Sony etc. that lasted many years. In fact you can count on it when buying a regular Japanese eighties/nineties device that you need to replace the relay as it is very likely a wrongly chosen type. Of course audio buyers defend brand names and designers (human behaviour 🙂) and a hobbyist well what does he/she know? But devices break down, one just does not know when. Normally repair guys simple replace defective parts quickly by the same parts, the second way of looking is to try to improve stuff so that it does not breakdown again. The third way is to copy the devices PCBs and redesign it without the design imperfections. Working like this one develops a quick eye to repeated and copied design errors.
Of course there also used to be the famous cottage industry British stuff with sometimes very dangerous constructions and no protection at all. Since I was one of those guys in the once large audio enthusiast scene who could explain what had happened I saw many burned woofers and developed a rigid and stubborn attitude towards unprotected higher power DC coupled amplifiers regardless of brand and designer: No protection? No entrance in my home nor a positive recommendation to audio friends.
Accepting no muting relays, no DC protection, severe power on/off phenomena and no RF stray in filtering in expensive devices in 2022 simply is silly as most of it is proven and tested technology already standard more than 40 years ago.
Of course there also used to be the famous cottage industry British stuff with sometimes very dangerous constructions and no protection at all. Since I was one of those guys in the once large audio enthusiast scene who could explain what had happened I saw many burned woofers and developed a rigid and stubborn attitude towards unprotected higher power DC coupled amplifiers regardless of brand and designer: No protection? No entrance in my home nor a positive recommendation to audio friends.
Accepting no muting relays, no DC protection, severe power on/off phenomena and no RF stray in filtering in expensive devices in 2022 simply is silly as most of it is proven and tested technology already standard more than 40 years ago.
Last edited:
For high DC voltage switching try Finder 45.31 or 45.91 series. They can break 16A/30V DC or even 4A/110V DC.
Most points were already mentioned, but if a relay-based protection circuit should work reliably we have to consider all possible conditions (even the "unthinkable"). For example, I've seen (attending a bi-amplification experiment) that the operator didn't connect the loudspeakers, but the outputs of amplifier one to the outputs of amplifier two, due to a cable mess. As usual, the bigger amplifier won. Or imagine a short circuit (faulty loudspeaker or cable).
I don't know about the Amplimo relay (I've never seen a detailed data sheet), but tend to think that it could be a similar design as the "Schrack", that Jean-Paul already mentioned. These are fine for (very) high inrush currents, but the tungsten contact does not help for switching off (being in an intermediate open position ensured the patented construction), therefore the switch-off capability is restricted to the somewhat normal range.
I think MarcelvdG's concerns are valid ones, although the arcing (when switching off a dc-load higher than rated) would not prevent the relay to switch off.
But is it still true, if a relay contact was already working under unfortunate conditions before? The minimum load was mentioned before, and it was good advice to exchange the relay after any emergency switch-off, or at least to carefully inspect the relays (including measurements, if possible). Using a socket helps a lot.
It seems, for the utmost safety (protecting the loudspeakers from an amplifier failure and vice versa and any human mishandling), a combination of two relays at the output (one carrying the switching load, while the other is used for normal operation) and a mains relay (switching off the power supply) is needed. For critical applications, even some redundancy should be included.
I don't know about the Amplimo relay (I've never seen a detailed data sheet), but tend to think that it could be a similar design as the "Schrack", that Jean-Paul already mentioned. These are fine for (very) high inrush currents, but the tungsten contact does not help for switching off (being in an intermediate open position ensured the patented construction), therefore the switch-off capability is restricted to the somewhat normal range.
I think MarcelvdG's concerns are valid ones, although the arcing (when switching off a dc-load higher than rated) would not prevent the relay to switch off.
But is it still true, if a relay contact was already working under unfortunate conditions before? The minimum load was mentioned before, and it was good advice to exchange the relay after any emergency switch-off, or at least to carefully inspect the relays (including measurements, if possible). Using a socket helps a lot.
It seems, for the utmost safety (protecting the loudspeakers from an amplifier failure and vice versa and any human mishandling), a combination of two relays at the output (one carrying the switching load, while the other is used for normal operation) and a mains relay (switching off the power supply) is needed. For critical applications, even some redundancy should be included.
This one will switch 25A@400VDC
https://www.mouser.ca/ProductDetail...ABs40A22FtnVR5o423W6oEXRu_bY3ODRoCii0QAvD_BwE
NOW you are getting somewhere. But no amplifier manufacturer uses one that is going to cost $20 apiece in 1,000,000 piece lots. Hence, you get the little PCB mount thingies that the contacts get oxidized and go partially rectifying after a few years (sound like **** and need replacement) or get welded shut when/if something bad does happen. They just count on the fact that MOST failure modes are more benign statistically. Of course you can spend the 50 bucks for something that will do the job building your own amplifier. The ice cubes I buy for $5-10 apiece on the surplus market don’t have trouble in the real world opening on 80 volt power supplies with 4 ohm loads, even if not rated for it. This should be able to do more.
I had to locate mosfets for the SSR of a large amp prototype earlier this year. In order to get the on resistance down low enough at 165 volt rail, it takes 8 TO-220 mosfets and 4 driver optos - per channel. Of course the $3 mosfets weren’t going to be in stock for A YEAR - so I settled for $9 equivalents. That adds up to more than the required mechanical relay - even at onsey twosey prices - which I would have used but will not fit in the Z-height of the board. I’m already stacking PCBs and it’s got to fit UNDER the daughterboard so TO-220’s it is. One of those cheap little ones they use in most amps will, but…..
I had to locate mosfets for the SSR of a large amp prototype earlier this year. In order to get the on resistance down low enough at 165 volt rail, it takes 8 TO-220 mosfets and 4 driver optos - per channel. Of course the $3 mosfets weren’t going to be in stock for A YEAR - so I settled for $9 equivalents. That adds up to more than the required mechanical relay - even at onsey twosey prices - which I would have used but will not fit in the Z-height of the board. I’m already stacking PCBs and it’s got to fit UNDER the daughterboard so TO-220’s it is. One of those cheap little ones they use in most amps will, but…..
But only in one direction, apparently (see the note that the contacts are polarized in the datasheet).
relay Vs mosfet
op, consider this:
switches, connectors and relays are prone to intermittent failure sooner or later
do you want to introduce this possibililty?
op, consider this:
switches, connectors and relays are prone to intermittent failure sooner or later
do you want to introduce this possibililty?
The reason why relays are used is because of possible blowing sensitive semis that "usually" short and relays are sturdy and can take abuse. You see, there is a reason some don't use any protection at all 🙂 Too many choices for simple matters.
https://www.diyaudio.com/community/...le-community-pat-quilter.387291/#post-7049962
https://www.diyaudio.com/community/...le-community-pat-quilter.387291/#post-7049962
Last edited:
It is rated and built for DC. The magnetic field tends to blow-out the arc. While some AC contacts are also magnetic blow-out, the geometry is different. I would not try this on audio (AC plus accidental DC) without months of testing.