I agree with @Bonsai the TO-220 mosfets are perfectly capable of handling tens of Amps of load if switched correctly. Normally when one switches an amplifier on no music should be playing so switch on is not a problem, then if the mosfets are fully on their resistance is so small that they shouldn't heat up at all during normal amplifier operation and in case a fault occurs the IC logic switches the mosfets off so again no problem, if the datasheet says they can handle for example 129A max at 25 c temperature then I'm sure they can handle close to that current for short time it will take for the IC to turn off.
Now one thing I don't understand, why one would need to put the photovoltaic couplers in series?
Diodes normally work in parallel very well , is there a reason I don't see here ? You think one of the LED'S will switch on before the other?
Now one thing I don't understand, why one would need to put the photovoltaic couplers in series?
Diodes normally work in parallel very well , is there a reason I don't see here ? You think one of the LED'S will switch on before the other?
If you just wire the LED diodes in parallel with a single currenct limiting resistor, if the one diode has a higher voltage drop (very common) it hogs all the current.
You can wire the photo voltaic coupler outputs directly in parallel - this will also switch the mosfet faster but I never personally tried this.
You can wire the photo voltaic coupler outputs directly in parallel - this will also switch the mosfet faster but I never personally tried this.
These optos only put out 8 or 10 volts into an open circuit. Most mosfets need a smidge more than this to get the lowest possible ON resistance. 100 amp mosfet is more expensive than the coupler, and you want all that you paid for in the mosfets (and you do need two of those).
If you’re replacing a 24V relay, you could easily use 8 couplers - four parallel groups of two in series. Wire all the LEDs in series. No reason not to wire all the LEDs in series if you have enough forward voltage headroom. You need only 20 mA and not some multiple of it - wasted power just makes a resistor hot somewhere. And then add a diode-transistor circuit at the output if you’re still jonesing for more turn off speed.
current capability of a mosfet when used as an SSR under "normal usage conditions"
just to clarify my earlier question:
when a mosfet is used as an SSR, the vast majority of its time will be spent being ON i.e. "normal usage conditions"
it is a link from the poweramp to the speaker and so needs to pass whatever current flows between the poweramp and the speaker
so, with the mosfet ON, what is its current capability?
just to clarify my earlier question:
when a mosfet is used as an SSR, the vast majority of its time will be spent being ON i.e. "normal usage conditions"
it is a link from the poweramp to the speaker and so needs to pass whatever current flows between the poweramp and the speaker
so, with the mosfet ON, what is its current capability?
Last edited:
Yes the mosfet is in series with the speaker/load, current in series circuit is equal through all parts.
The current capability of a mosfet in ON state is the current rating given in its datasheet, why you don't believe that?
Modern mosfet ON resistance is for example 10 miliohms. There is practically no voltage drop across it, heat dissipation is very minimal.
Usually datasheets have a "continuous drain current" rating.
This datasheet you provided only shows the rating for a pulsed current it seems, but others show it as continous drain current. Anyway I am sure any powerful mosfet will fit for your SSR, what is the amplifier power you want to use it on, rail voltage etc?
Before I offer any comments it should be noted that I designed a UPC1237 circuit for Phase Linear 400's and sell them on EBAY. Spare me the comments on P/L amps if you don't like them. They're still plenty popular for their intended application, that being home audio. So a person could question my motives here. My intent is to provide information only, mostly free of prejudice. You be the judge.
Anyone considering the merits of amplifier output protection circuits should read Rod Elliot's articles on relays, mosfet switches and amplifier failure detection circuits. There is a lot of wisdom there. Here is a link. Rod Eliott
In the early 80's I worked in pro sound and 9 out of 10 local R&R bands had racks of Phase 400's. Any amp can fail and if it does it can take out expensive speakers. The 400's had no speaker protection. I heard a lot of legitimate sob stories.
Most protection circuits disconnect the speaker after detecting DC on the output. The problem with using relays is any economical relay that would fit in an amplifier chassis has contacts rated for about 30 VDC and 250 VAC. The reason for the low DC rating is when the contacts open they arc. With DC the arc is sustained and the contacts weld. When a relay interrupts AC the current goes to zero each half cycle and the arc extinguishes. The contact opens.
As Rod points out, a successful application of a relay must not only disconnect the speaker from the amp, but also short out the speaker in the event the contacts weld. The circuit I designed uses a "Form C" contact which has a normally open and a normally closed contact. I used the highest current ratings I could find to meet the design goals. The normally closed contact shorts out the speaker. The normally open contact is in series with the speaker. Normal is defined as shelf state meaning the relay is de-energized.
When developing the circuit I had a sacrificial speaker and used a knife switch across the output stage of a P/L 400 to simulate a failure. This dumped 80 VDC into the circuit under test and the speaker. I ran the test about 30 times and captured scope photos of the output to optimize the circuit. The test speaker still works today. I have not experiencedd one relay failure.
At my full time job I've worked with industrial inverters for nearly 40 years. The early inverters used SCR's as switches. Todays inverters use IGBT's (insulated gate bipolar transistors). SCR's and diodes can be protected from short circuits with high speed fuses. IGBT's can not be protected with fuses and still make use of their max current. They require sophisticated monitoring circuits that don't always work. I don't trust a transistor to open under short circuit conditions. My only experience with mosfets was Hafler amps which were reliable.
Although it appears that there may be some merit in using mosfets to protect speakers from amplifier failures, I still prefer the advantage of the form C relay contact. The UPC1237 is a reliable detector and is still plentiful and I'm sticking with it. Anyone using mosfets should look at the short circuit capability of a mosfet and compare it with the time/current specs of fuse that is protecting it. The fuse has to open before the transistor fails/shorts. The detection circuit must be fast but not create nuisance trips at low frequencies.
It is my opinion all of the off the shelf relay boards on EBAY are misapplied and will result in disappointment when called upon to serve it's purpose. The relay contacts will weld. They do not use a form C contact.
Anyone considering the merits of amplifier output protection circuits should read Rod Elliot's articles on relays, mosfet switches and amplifier failure detection circuits. There is a lot of wisdom there. Here is a link. Rod Eliott
In the early 80's I worked in pro sound and 9 out of 10 local R&R bands had racks of Phase 400's. Any amp can fail and if it does it can take out expensive speakers. The 400's had no speaker protection. I heard a lot of legitimate sob stories.
Most protection circuits disconnect the speaker after detecting DC on the output. The problem with using relays is any economical relay that would fit in an amplifier chassis has contacts rated for about 30 VDC and 250 VAC. The reason for the low DC rating is when the contacts open they arc. With DC the arc is sustained and the contacts weld. When a relay interrupts AC the current goes to zero each half cycle and the arc extinguishes. The contact opens.
As Rod points out, a successful application of a relay must not only disconnect the speaker from the amp, but also short out the speaker in the event the contacts weld. The circuit I designed uses a "Form C" contact which has a normally open and a normally closed contact. I used the highest current ratings I could find to meet the design goals. The normally closed contact shorts out the speaker. The normally open contact is in series with the speaker. Normal is defined as shelf state meaning the relay is de-energized.
When developing the circuit I had a sacrificial speaker and used a knife switch across the output stage of a P/L 400 to simulate a failure. This dumped 80 VDC into the circuit under test and the speaker. I ran the test about 30 times and captured scope photos of the output to optimize the circuit. The test speaker still works today. I have not experiencedd one relay failure.
At my full time job I've worked with industrial inverters for nearly 40 years. The early inverters used SCR's as switches. Todays inverters use IGBT's (insulated gate bipolar transistors). SCR's and diodes can be protected from short circuits with high speed fuses. IGBT's can not be protected with fuses and still make use of their max current. They require sophisticated monitoring circuits that don't always work. I don't trust a transistor to open under short circuit conditions. My only experience with mosfets was Hafler amps which were reliable.
Although it appears that there may be some merit in using mosfets to protect speakers from amplifier failures, I still prefer the advantage of the form C relay contact. The UPC1237 is a reliable detector and is still plentiful and I'm sticking with it. Anyone using mosfets should look at the short circuit capability of a mosfet and compare it with the time/current specs of fuse that is protecting it. The fuse has to open before the transistor fails/shorts. The detection circuit must be fast but not create nuisance trips at low frequencies.
It is my opinion all of the off the shelf relay boards on EBAY are misapplied and will result in disappointment when called upon to serve it's purpose. The relay contacts will weld. They do not use a form C contact.
I have no tried or even looked at the new mosfet driver IC that Rod mentions - I will have to investigate it.
The original push to use mosfets on my side came after a 240 Watt amp of mine designed 2004/2005 (you can read about it here http://hifisonix.com/ovation-250/ blew in 2012 (I had done loads of mods/upgrades to it so it wasn't the most reliable amp), taking out the bass speakers in one of the B&W 703 speakers (there's a whole thread on this somewhere on the forum). The relay contacts were welded short so the speaker took the full 75V rail. Michael Bean showed a solid state relay and that's what I've used ever since along with a lot of other folks here on the forum. I don't agree with Rod that SSR's will be bigger than a decent relay. Assuming the standard relay is being controlled by circuits that are monitoring for all the right things (muting, DC offset etc), the actual switch part can be made to fit straight into the PCB footprint of a Tyco RT34 relay as shown here http://hifisonix.com/solid-state-loudspeaker-relay/ . The additional circuitry around the mosfets all operates at low currents and I doubt suffers from issues that would make it less reliable in any way than a mostfet solution - semiconductor switches will always trump electro-mechanical for reliability which is why the auto industry has largely moved away fro relays and gone over to mosfets the last 25 years. In addition, as Rod points out, the mosfets will switch reliably (provided they are driven correctly) under fault conditions ad-in-finitum - they designed for switching heavy loads. EMR's can't do that.
Its often not the amplifier that's the problem - well designed amps should be reliable. But, the user is often the issue eg speaker connections shorted out on the rear panel (I've done this a few times). If you sell an amp and it goes faulty for some reason, there's the expense of getting it back to the workshop for repair and you will always have an unhappy customer.
I all of this, it should be remember that the main reason for protection is for the speaker - the two bass units in the B&W cost over $400. Imagine blowing a $10k speaker or worse?
The original push to use mosfets on my side came after a 240 Watt amp of mine designed 2004/2005 (you can read about it here http://hifisonix.com/ovation-250/ blew in 2012 (I had done loads of mods/upgrades to it so it wasn't the most reliable amp), taking out the bass speakers in one of the B&W 703 speakers (there's a whole thread on this somewhere on the forum). The relay contacts were welded short so the speaker took the full 75V rail. Michael Bean showed a solid state relay and that's what I've used ever since along with a lot of other folks here on the forum. I don't agree with Rod that SSR's will be bigger than a decent relay. Assuming the standard relay is being controlled by circuits that are monitoring for all the right things (muting, DC offset etc), the actual switch part can be made to fit straight into the PCB footprint of a Tyco RT34 relay as shown here http://hifisonix.com/solid-state-loudspeaker-relay/ . The additional circuitry around the mosfets all operates at low currents and I doubt suffers from issues that would make it less reliable in any way than a mostfet solution - semiconductor switches will always trump electro-mechanical for reliability which is why the auto industry has largely moved away fro relays and gone over to mosfets the last 25 years. In addition, as Rod points out, the mosfets will switch reliably (provided they are driven correctly) under fault conditions ad-in-finitum - they designed for switching heavy loads. EMR's can't do that.
Its often not the amplifier that's the problem - well designed amps should be reliable. But, the user is often the issue eg speaker connections shorted out on the rear panel (I've done this a few times). If you sell an amp and it goes faulty for some reason, there's the expense of getting it back to the workshop for repair and you will always have an unhappy customer.
I all of this, it should be remember that the main reason for protection is for the speaker - the two bass units in the B&W cost over $400. Imagine blowing a $10k speaker or worse?
Last edited:
My English is not the best, so I must ask for clarification - is this the type of relay you use?
https://eu.mouser.com/ProductDetail...ices/JW2SN-B-DC5V?qs=bpFJJ1fyfoCVi702uUo8JA==
Unisonic are still making the UPC1237 but because they are s small outfit, don’t get to list products at places like Mouser and DK unfortunately. NEC pulled out of this business 15 or 20yrs ago so be very wary of sellers who claim they have NEC parts. I see even the devices from St. Quentin are UPC!
After one day of reading various sources:
It seems to work perfectly, in ltspice at least, since I don't have opto parts at hand to test it (it will take approx one month for them to arrive here). The circuit needs external power to run, everything from 5 to 15V (Vcc range of NE555). Resistors R5, R6, R7 and R8 shall be adjusted in case of different supply voltage and different diodes.
D2 is status LED, it is on when the speaker is on, but if one want to make it show the opposite, simply move it together with R7 to OUT pin of NE555.
I used C4 and C5 to remove switching spikes on LED, but they are probably not necessary.
C6 is paralleled with MOSFETs, as shown on ESP page. The simulation runs very slow without it, so it is probably needed in real life, too, although I didn't see it on other similar designs. D1 is also from that page.
Hifisonix also has 12k resistors from speaker leads to ground for some reason.
So, if anyone is interested, you are free to comment or try the circuit. I will build it once the parts arrive. Please find all the ltspice files in an attachment, I hope I didn't forget to include all needed for this to run.
Thanks to everyone for sharing their thoughts and knowledge!
- Douglas Self - Audio Power Amplifier Design
- ESP - MOSFET SSR
- Speaker Protection and Muting with an Optical Coupler
- Hifisonix - Solid State Loudspeaker Relay
- and various pages on this site
It seems to work perfectly, in ltspice at least, since I don't have opto parts at hand to test it (it will take approx one month for them to arrive here). The circuit needs external power to run, everything from 5 to 15V (Vcc range of NE555). Resistors R5, R6, R7 and R8 shall be adjusted in case of different supply voltage and different diodes.
D2 is status LED, it is on when the speaker is on, but if one want to make it show the opposite, simply move it together with R7 to OUT pin of NE555.
I used C4 and C5 to remove switching spikes on LED, but they are probably not necessary.
C6 is paralleled with MOSFETs, as shown on ESP page. The simulation runs very slow without it, so it is probably needed in real life, too, although I didn't see it on other similar designs. D1 is also from that page.
Hifisonix also has 12k resistors from speaker leads to ground for some reason.
So, if anyone is interested, you are free to comment or try the circuit. I will build it once the parts arrive. Please find all the ltspice files in an attachment, I hope I didn't forget to include all needed for this to run.
Thanks to everyone for sharing their thoughts and knowledge!
After noticing that there is a big voltage spike between MOSFETs while switching off, I added R13 and R14 and the spike is gone. Probably additionally protecting VOM1271 as well.
However, I see that, whatever I do, this kind of switch produce crossover distortion. No idea why and how to remove it...
However, I see that, whatever I do, this kind of switch produce crossover distortion. No idea why and how to remove it...
It seems that crossover distortion is gone if I use different MOSFET for switching, and it works with both PMOS and NMOS (just have to connect gate to the opposite). IRF540 works really fine in simulation.