It is the current as well. Take a look at the specs of any relay.
The higher the voltage that is being interrupted, the less the current capability. And the same applies if you try to switch a DC load because you get arcing (due to ionization) as the contact breaks. And it is the arcing that causes contact wearout and damage. Sure, inductive loads make it worse, but it is not the only mechanism causing arcing and contact wearout.
FETS: You will need clamp diodes to each of the rails on the output side of the SSR. If you stay within the FET specs, thats all you need. I'll happily bet a decently spec'd mosfet will switch a high current inductive load all day long.
Last edited:
You have repeatedly stated that the source impedance due to the relay must be 1/1000 part of the load impedance.
In my opinion this is not a requirement.
Suppose we have an 8milliohm impedance in series with the 8ohm load and all the other source impedances that cannot be reduced to near zero ohms.
Let's further suppose that the Rds on or the relay contact impedance of 8milliohms is made up of a resistive element and a reactive element.
Is the resistive element linear with Vdrop across it?
Is the reactive element linear with Vdrop across it?
Is some of the Vdrop is due to a non linear resistance or a non linear reactance then that part of the element will create an added distortion fed along with the wanted signal.
Let's guess that the distortion attributable to the non-linear elements of the resistive and reactive elements of the contacts/Rdson are 10% of the Vdrop.
If the non linear elements make up 50% of the total impedance and the distortion from those elements is 10%, then the total contribution of the contacts/Rdson to the distortion is 0.5 * 0.1 * 0.001 = 0.00005 (or about 0.005%) of the total signal or about -86dB.
I suspect the non linear parts of the relay contact impedance make up for a distortion contribution that is very much less than the -86dB for the example 8milliohms of contact impedance.
I agree, Andrew.
Unless I have missed something (I haven't read the thread in detail) the claim seems to be that the contact resistance contributes 100% to non-linearity and therefore distortion.
In fact only the delta-R makes such a contribution. If R is stable it makes no contribution other than that of a slightly longer speaker wire.
Unless I have missed something (I haven't read the thread in detail) the claim seems to be that the contact resistance contributes 100% to non-linearity and therefore distortion.
In fact only the delta-R makes such a contribution. If R is stable it makes no contribution other than that of a slightly longer speaker wire.
Last edited:
I can only recall my actual experience of "relay distortion" and leave you to draw your own conclusions.
On listening to music I was totally unaware of any problems whatsoever. On sinewave test into the speakers it sounded awful. How bad ? Well imagine a true Class B amp with zero bias at a few 10's of milliwatts. That's bad quality wise. Now scale that sound up to around 1 watt. However increase the volume more and the current "cleaned" the contacts and the problem vanished until the next time. On music though I was blissfully unaware.
That for me was the decider on doing away with mechanical relays. I'd already gone the SS route for all other switching.
Any "distortion" caused by a FET's Rds seems pretty harmless after that. Can a variation of Rds with current cause "harmonic distortion". I would imagine it's just a form of amplitude distortion rather than creating other components in the signal. The mechanical relay is a different matter altogether.
On listening to music I was totally unaware of any problems whatsoever. On sinewave test into the speakers it sounded awful. How bad ? Well imagine a true Class B amp with zero bias at a few 10's of milliwatts. That's bad quality wise. Now scale that sound up to around 1 watt. However increase the volume more and the current "cleaned" the contacts and the problem vanished until the next time. On music though I was blissfully unaware.
That for me was the decider on doing away with mechanical relays. I'd already gone the SS route for all other switching.
Any "distortion" caused by a FET's Rds seems pretty harmless after that. Can a variation of Rds with current cause "harmonic distortion". I would imagine it's just a form of amplitude distortion rather than creating other components in the signal. The mechanical relay is a different matter altogether.
"Take a look at the specs of any relay. "
I have, it's the inrush current and the inductive flyback, not the current per se that causes the problem.
Motorola had an app note for a solid state 100A contactor that used a 5A relay to short out the thyristors after they handled the inrush current, and then the relay dropped out on turn-off before the thyristors were turned off.
I have, it's the inrush current and the inductive flyback, not the current per se that causes the problem.
Motorola had an app note for a solid state 100A contactor that used a 5A relay to short out the thyristors after they handled the inrush current, and then the relay dropped out on turn-off before the thyristors were turned off.
You are right, and this is also exactly my intention. By having the contribution of the switch at -60 dB the total (maximum) distortion contribution goes below -90 dB (I think) and that again guarantees that the amplifiers design goals stay as intended.
I have put my thoughts on SSR's on my website here Audio Amplifier Design and Circuits | hifisonix.com - its the first .pdf at the top of the page.
My thanks to Michael Bean for triggering this discussion in the first place.
My thanks to Michael Bean for triggering this discussion in the first place.
You need reference any switch distortion to the amplifier output level. My simms show about 1ppm.
I should be 1/1000.000.000'th but that's not real life 🙂 Besides that we should strive for the best possible, and honor our design goal's.
"Any "distortion" caused by a FET's Rds seems pretty harmless after that. Can a variation of Rds with current cause "harmonic distortion". I would imagine it's just a form of amplitude distortion rather than creating other components in the signal. The mechanical relay is a different matter altogether. "
For normal speaker load currents, when the mosfets are fully ON, the Rdson is as near as damn it to a piece of wire with some low resistance. I would therefore expect the distortion to be absolutely minimal.
For normal speaker load currents, when the mosfets are fully ON, the Rdson is as near as damn it to a piece of wire with some low resistance. I would therefore expect the distortion to be absolutely minimal.
"Now switch that same current at the voltage appearing across the thyristors and see what happens. "
Exactly.
Exactly.
I'm sorry you don't understand the Motorola app note.
The 5A relay handles 100A currents, it just can't handle inrush currents and flyback voltages.
The 5A relay handles 100A currents, it just can't handle inrush currents and flyback voltages.
AFAIK, solid state relays do affect the quality of the output signal. That has been reported by many users. SCR protection also does.
As output protection is a necessary evil, it looks like a regular relay might be the lesser one. Isn't it?
As output protection is a necessary evil, it looks like a regular relay might be the lesser one. Isn't it?
Hi,
The best choice would be a crowbar relay, mechanical combined with a "fast" mechanical mute at the VAS output. If not operational it is as close as anything to "not there", with both relays open.
If triggered in accident the small signal mute relay at the output stage input should engage quicker than the output crowbar and hence there is no risk of damage to the Output Stage and if the output stage is blown up, then it is blown up and the crowbar will short the DC ground.
On the other hand, regular relays at times will arc over when trying to break large DC currents/voltages which will promptly arc-weld the relay contacts and then burn out the speaker anyway, protection non-withstanding.
The FET based output relays - we would have to test their ability to correctly break the output current loop, without burning out themselves to see if they are really better than mechanical ones for safety.
Ciao T
The best choice would be a crowbar relay, mechanical combined with a "fast" mechanical mute at the VAS output. If not operational it is as close as anything to "not there", with both relays open.
If triggered in accident the small signal mute relay at the output stage input should engage quicker than the output crowbar and hence there is no risk of damage to the Output Stage and if the output stage is blown up, then it is blown up and the crowbar will short the DC ground.
On the other hand, regular relays at times will arc over when trying to break large DC currents/voltages which will promptly arc-weld the relay contacts and then burn out the speaker anyway, protection non-withstanding.
The FET based output relays - we would have to test their ability to correctly break the output current loop, without burning out themselves to see if they are really better than mechanical ones for safety.
Ciao T