I said to parallel the photo-voltaic driver, not the FET's.
you can parallel the fets providing you switch very fast - i.e. well within the SOA constraints.
;-)
I'm considering making up two modules to replace the relays in my amp... particularly after my posting #179 and reply at #181.
Looks like Farnell for the photovoltaic couplers. The FET's... well for my amp I already use a 0.22 ohms series output resistor so ultimate low on Rds probably isn't a major concern. I note that a lot of the FET's seem to be quoted at 10 volts Vgs and some of the super low Rds ones at 20 or in one case 30 volts. 4 or 5 volts is quoted as a typical figure.
I was looking at the IRF2907,
INTERNATIONAL RECTIFIER|IRF2907ZPBF|MOSFET, N, 75V, 170A, TO-220 | Farnell United Kingdom
or in a TO247,
INTERNATIONAL RECTIFIER|IRFP2907ZPBF|MOSFET, N, 75V, 170A, TO-247AC | Farnell United Kingdom
And here's a low Rds one but look at the test Vgs,
INTERNATIONAL RECTIFIER|IRFP3077PBF|MOSFET, N, TO-247AC | Farnell United Kingdom
Looks like Farnell for the photovoltaic couplers. The FET's... well for my amp I already use a 0.22 ohms series output resistor so ultimate low on Rds probably isn't a major concern. I note that a lot of the FET's seem to be quoted at 10 volts Vgs and some of the super low Rds ones at 20 or in one case 30 volts. 4 or 5 volts is quoted as a typical figure.
I was looking at the IRF2907,
INTERNATIONAL RECTIFIER|IRF2907ZPBF|MOSFET, N, 75V, 170A, TO-220 | Farnell United Kingdom
or in a TO247,
INTERNATIONAL RECTIFIER|IRFP2907ZPBF|MOSFET, N, 75V, 170A, TO-247AC | Farnell United Kingdom
And here's a low Rds one but look at the test Vgs,
INTERNATIONAL RECTIFIER|IRFP3077PBF|MOSFET, N, TO-247AC | Farnell United Kingdom
INTERNATIONAL RECTIFIER|IRFB3607PBF|MOSFET, N, TO-220AB | Farnell United Kingdom
Is a FET that I found when looking previously. Less maximum current capability, but far cheaper too! How powerful is the amplifier you're going to put the solid state relay in?
I don't think any of the FET drivers output any more then at most 10 volts. If you look at the datasheets though you will see how the MOSFETs current capabilities change with differing GS voltages.
Is a FET that I found when looking previously. Less maximum current capability, but far cheaper too! How powerful is the amplifier you're going to put the solid state relay in?
I don't think any of the FET drivers output any more then at most 10 volts. If you look at the datasheets though you will see how the MOSFETs current capabilities change with differing GS voltages.
Yeah farnells search engine threw back a load of errors when I first started seaching for FETs, I think I had far too many parameters selected that caused it to malfunction. I often end up browsing semi conductor manufactures websites to find parts, instead of relying on farnell/RS to do the job adequately.
Interesting. I'm having almost this exact problem on the right channel of my bass amplifier. It has DC protection by means of the Rod Elliot PCB and large, but unsealed, 30A relays. The left woofer cuts in and out, but will reliably come back in at high levels. I had assumed it was just deterioration of the contacts in the breadboard active crossover I'm still using
o) but I think I will try shorting out that relay this afternoon, see what happens.This is from Agilent/HP that has been around for decades:
Predicting relay life spans:
Relay manufacturers specify how long their relays will last, but the expected lifetime will vary depending on the loads they are subjected to. For resistive loads, manufacturers’ specifications are typically fairly accurate. On the other hand, if you are using capacitance or inductance, your relay life span will be shorter than the manufacturers specification. How much shorter depends on the type of loads you are switching. Derating gives you a realistic picture of how long your relay will last.
Relay derating factors for common load types:
Type of load Percent of rated value
Resistive 75
Inductive 40
Capacitive 75
Motor 20
Incandescent 10
Predicting relay life spans:
Relay manufacturers specify how long their relays will last, but the expected lifetime will vary depending on the loads they are subjected to. For resistive loads, manufacturers’ specifications are typically fairly accurate. On the other hand, if you are using capacitance or inductance, your relay life span will be shorter than the manufacturers specification. How much shorter depends on the type of loads you are switching. Derating gives you a realistic picture of how long your relay will last.
Relay derating factors for common load types:
Type of load Percent of rated value
Resistive 75
Inductive 40
Capacitive 75
Motor 20
Incandescent 10
"Yes it would be. My issue would only present itself at low drive levels. When driven harder it would allow a decent contact to form that would then persist when the level was reduced. If I turned the amplifier off, so the relays opened, when turned back on again the problem would reappear, only after turning it up did the problem 'go away'."
I would judge that relay to be defective.
When relay contacts make, they do so with a wiping action. If they just close without a proper wiping action, they behave like this.
I have tried cleaning contacts when they act like this, to no avail.
In the end I end up replacing them.
I would judge that relay to be defective.
When relay contacts make, they do so with a wiping action. If they just close without a proper wiping action, they behave like this.
I have tried cleaning contacts when they act like this, to no avail.
In the end I end up replacing them.
I did a little lash up yesterday. The only FET's I had were two different types, an IRF830and 630. Anyway, wired them up as per previous posts together with a PP3 to provide a floating Vgs supply for on and off.
I connected them to the output of an old Rotel RA820 amp and used music to test initially. Seemed fine. I then played a test disc (sine wave 6khz) and used a 5 ohm load resistor with dual scope on amp output and load.
Results I got are probably to be expected... significant voltage loss across the FET's but very surprisingly (or surprising in that it was so obvious on a scope) was the severe non linear distortortion. This has to be because the FET's were 1. totally different and 2. they had a significant Rds value of around 400 to 600 mohm. Each that is.
So that is no criticism of the technique or principle... just something to be aware of. I mentioned earlier that as my particular amp already uses 0.22 ohm series output resistors that a spectacularly low Rds probably wouldn't matter. I have revised my thinking on that... you need as low as possible and also for the devices to be matched and possibly in thermal equilibriam too. The parts I would use are around 4mohm, around a 100 times better than in my quick experiment and would be of the same type of course. For those that like distortion levels in the 0.00xx range then any imbalance of the FET's would dramatically alter that.
Strange thing is... I bet it's the "right" kind of distortion audibly.
I connected them to the output of an old Rotel RA820 amp and used music to test initially. Seemed fine. I then played a test disc (sine wave 6khz) and used a 5 ohm load resistor with dual scope on amp output and load.
Results I got are probably to be expected... significant voltage loss across the FET's but very surprisingly (or surprising in that it was so obvious on a scope) was the severe non linear distortortion. This has to be because the FET's were 1. totally different and 2. they had a significant Rds value of around 400 to 600 mohm. Each that is.
So that is no criticism of the technique or principle... just something to be aware of. I mentioned earlier that as my particular amp already uses 0.22 ohm series output resistors that a spectacularly low Rds probably wouldn't matter. I have revised my thinking on that... you need as low as possible and also for the devices to be matched and possibly in thermal equilibriam too. The parts I would use are around 4mohm, around a 100 times better than in my quick experiment and would be of the same type of course. For those that like distortion levels in the 0.00xx range then any imbalance of the FET's would dramatically alter that.
Strange thing is... I bet it's the "right" kind of distortion audibly.
Hi Mooly,
Yeah your right, it's pretty much mandatory to use the lowest Rds devices available. A MOSFET switch when turned on will exibit a small amount of resistance variation depending on how much current they are conducting and how much drive voltage is applied between gate and drain (the higher the better), lower Rds devices will have the least variation. I'm sure that's why you heard audible distortion with the high Rds ones you tried. So to use these effectively, use lowest Rds MOSFETs and drive the photovolaic isolators with the highest reasonable current for greatest output voltage. When I was trying these out early on, I listened to them with the dual photovoltaic outputs in series for double the drive voltage, and singly, but never heard any difference.
Mike
Yeah your right, it's pretty much mandatory to use the lowest Rds devices available. A MOSFET switch when turned on will exibit a small amount of resistance variation depending on how much current they are conducting and how much drive voltage is applied between gate and drain (the higher the better), lower Rds devices will have the least variation. I'm sure that's why you heard audible distortion with the high Rds ones you tried. So to use these effectively, use lowest Rds MOSFETs and drive the photovolaic isolators with the highest reasonable current for greatest output voltage. When I was trying these out early on, I listened to them with the dual photovoltaic outputs in series for double the drive voltage, and singly, but never heard any difference.
Mike
Seems like I wasn't alone in having issues with relays! It's nice too that there's an easy and pleasing way to solve this. It makes sense that a high RDSon FET would create non-linear distortion. If the FETs effective resistance is going to change as the sinusoidal voltage changes, then that would obviously give rise to mechanism that would create distortion. It's good to see that using low RDSon FETs completely renders this insignificant.
These were the cheapest OMRON relays I could find that fit the application. As they were sealed and contained the same contact material I figured they'd be pretty much the same as any other similar spec'd relay, little did I know that the switching action can be quite different from one to the other.
I popped the lid off one of the offending relays and indeed no sweeping action to be found. What surprised me was the state that the relay contacts were in, black with a layer of something that quite obviously doesn't conduct well! In comparison the contacts that should only pass current during a failure were clean as a whistle.
Have a relay.
Here I've bent the relay contacts out of shape but the way it works is quite simple. The signal enters the large arm on the left and then the relay action moves the arm up and down between the two contacts on the right.
Have a close-up of the contacts.
As you can see the upper contact is clean, whereas the bottom one... it's not a surprise that the relay was acting up, with the amplifier giving gross amounts of distortion.
What surprises me is how quickly this happened. I tended not to turn the amplifier on with a signal going through it and I also don't listen at particularly loud levels. The other part is that the system is entirely active so any crazy reactive loads, as created by a crossover wouldn't be present either. The most reactive part of a un-cross-overed loudspeaker is what happens at resonance too and I've always actively crossed over way before that, so perhaps the degradation could have happened a lot faster given different circumstances. I don't think I'll bother with mechanical relays again.
These were the cheapest OMRON relays I could find that fit the application. As they were sealed and contained the same contact material I figured they'd be pretty much the same as any other similar spec'd relay, little did I know that the switching action can be quite different from one to the other.
I popped the lid off one of the offending relays and indeed no sweeping action to be found. What surprised me was the state that the relay contacts were in, black with a layer of something that quite obviously doesn't conduct well! In comparison the contacts that should only pass current during a failure were clean as a whistle.
Have a relay.
Here I've bent the relay contacts out of shape but the way it works is quite simple. The signal enters the large arm on the left and then the relay action moves the arm up and down between the two contacts on the right.
Have a close-up of the contacts.
As you can see the upper contact is clean, whereas the bottom one... it's not a surprise that the relay was acting up, with the amplifier giving gross amounts of distortion.
What surprises me is how quickly this happened. I tended not to turn the amplifier on with a signal going through it and I also don't listen at particularly loud levels. The other part is that the system is entirely active so any crazy reactive loads, as created by a crossover wouldn't be present either. The most reactive part of a un-cross-overed loudspeaker is what happens at resonance too and I've always actively crossed over way before that, so perhaps the degradation could have happened a lot faster given different circumstances. I don't think I'll bother with mechanical relays again.
Attachments
Thanks Michael.
I'd wondered too about using the two photovoltaic couplers in series (two in one package) with a super low Rds FET that can handle the Vgs that the series combination would provide.
I would certainly want to repeat my test (5ohm load) using better FETs and again scopeing and comparing input and output plotted against Vgs.
You can't beat a bit of practical experimentation
5th element...
I'm no real expert on properties of relays but seem to remember reading about this blackening of contacts being caused by atmospheric polution reacting with the contact plating and causing oxidation. In the same way that silver tarnishes. But yours is only one of the contacts... maybe as the film appears some "electrolytic" action occurs as a minute voltage is developed across the closed contacts. Just guessing there.
Again like you, I never turn my amp on with music playing and never turn off either with it playing.
Yours looks a chunky enough relay, those contacts look like those on a 3.4kw storage heater
I'm no real expert on properties of relays but seem to remember reading about this blackening of contacts being caused by atmospheric polution reacting with the contact plating and causing oxidation. In the same way that silver tarnishes. But yours is only one of the contacts... maybe as the film appears some "electrolytic" action occurs as a minute voltage is developed across the closed contacts. Just guessing there.
Again like you, I never turn my amp on with music playing and never turn off either with it playing.
Yours looks a chunky enough relay, those contacts look like those on a 3.4kw storage heater
I've just temporarily (I suspect it'll stay like this for a while ) bypassed the relay on my right channel and the speaker seems to be reliable even at low levels now, so it seems as though that was indeed the problem.
The relays in this amp are very much like the one you posted 5th element, heavy duty things, 30A I think. Clearly they are prone to tarnishing, perhaps due to being totally unsealed.
Seems as though MOSFETs are the superior choice here, I need to get hold of some low Rds On ones to get the amp back up and running with some protection. I can use them on the ground return of the speaker so I can drive the MOSFET gate without a seperate power supply/optoisolator? I have the output of a Rod Elliot P33 PCB to utilise which currently operates the relays.
EDIT: I actually have a few RFP50N06 MOSFETs, 0.022Ohm Rds On, may be suitable?:
http://www.datasheetcatalog.org/datasheets/90/346714_DS.pdf
Wonder just how simply this can be implemented?
) bypassed the relay on my right channel and the speaker seems to be reliable even at low levels now, so it seems as though that was indeed the problem. The relays in this amp are very much like the one you posted 5th element, heavy duty things, 30A I think. Clearly they are prone to tarnishing, perhaps due to being totally unsealed.
Seems as though MOSFETs are the superior choice here, I need to get hold of some low Rds On ones to get the amp back up and running with some protection. I can use them on the ground return of the speaker so I can drive the MOSFET gate without a seperate power supply/optoisolator? I have the output of a Rod Elliot P33 PCB to utilise which currently operates the relays.
EDIT: I actually have a few RFP50N06 MOSFETs, 0.022Ohm Rds On, may be suitable?:
http://www.datasheetcatalog.org/datasheets/90/346714_DS.pdf
Wonder just how simply this can be implemented?
Last edited:
Are those 16A relays PCB mounted, fully sealed types? This 30A industrial one has failed in this application after maybe 2-3 years:
30A SPDT Power Relay : Power Relays : Maplin
no idea why the left channel is seemingly fine though, they will have been used identically! I think this type of relay really likes to switch AC load regularly, so the arc cleans the contact.
30A SPDT Power Relay : Power Relays : Maplin
no idea why the left channel is seemingly fine though, they will have been used identically! I think this type of relay really likes to switch AC load regularly, so the arc cleans the contact.
I don't know a huge amount about FETs, but if the circuit was used in the ground return then the drain of one of the FET pair would connect to circuit common, with the other drain of the FET pair connected to the signal return (black terminal) on the amplifier. In this case if the amplifier was turned on but passing no signal then the output of the amplifier and what would be present to the sources would be close to ground potential. If a voltage was applied, say 10 volts, to the gates, then the FETs would turn on. But what happens when a signal passes? If the signal going through the FETs was 30 volts negative then wouldn't this increase the Vgs to 40 volts and cause the FET to expire? Also if a positive signal of 30 volts was passed then Vgs would decrease to -20 causing the FET to turn off.
From the looks of things, the voltage applied to the FETs would at least have to be done with respect to the output rail, say a 10 volt voltage reference connected between +ve and the amplifier output. If you were using the Rod Elliot circuit then you'd have to opto couple the output of the Elliot PCB and then drive the gates from the other side of the opto from a voltage derived between a power rail and the amplifier output. Would that work? Of course a normal opto coupler can be very inexpensive and if you've already got a voltage bias from a zenner or similar in the amplifier I guess you could use this to drive the gates.
From the looks of things, the voltage applied to the FETs would at least have to be done with respect to the output rail, say a 10 volt voltage reference connected between +ve and the amplifier output. If you were using the Rod Elliot circuit then you'd have to opto couple the output of the Elliot PCB and then drive the gates from the other side of the opto from a voltage derived between a power rail and the amplifier output. Would that work? Of course a normal opto coupler can be very inexpensive and if you've already got a voltage bias from a zenner or similar in the amplifier I guess you could use this to drive the gates.
Mine were sold as fully sealed and would switch a nominal 10A AC load. As can be seen the unused contacts were clean so I don't think standard corrosion was an issue.