No I am just trying to understand how and why do one select a certain Voltage, why headroom? I do know that in my usage, 99.9% of the time do we not get over maybe 70 dB and do not need much power.
The easy part would bges unit in amplifier
The easy part would bges unit in amplifier
The Safe Operating Area of the output devices is often the limiting factor. Rod Elliot has a good article on this:
Semiconductor Safe Operating Area
Semiconductor Safe Operating Area
Thanks a lot, I have now read it about 5-6 times, it's a new angle I had no idea about, but it make a lot of sense. What I get out of it is that the heat is the limiting factor, so short burst of extra power (read headroom) is no problem, but a constant max wattage would kill the fet's.
If we take the Mooly amp, just to have a concrete example, is it specified to +45 0 -45, 50W 8 Ohm and I hope to mount a set of 4 Ohm speakers.
(45V^2 / 8 Ohm = 253.125W)
(45V^2 / 4 Ohm = 506.25W)
(sqrt(50W * 8 Ohm) = 20V
(sqrt(50W * 4 Ohm) = 20V
8 Ohm (50W / 20V = 2.5A)
4 Ohm (100W / 20V = 5A)
It do, to me look like 45V 6A to be sure? but headroom?
Have to point out that I am only trying to learn the theory and logic behind this, since Mooly already have given me a schematic for a simple power supply.
If we take the Mooly amp, just to have a concrete example, is it specified to +45 0 -45, 50W 8 Ohm and I hope to mount a set of 4 Ohm speakers.
(45V^2 / 8 Ohm = 253.125W)
(45V^2 / 4 Ohm = 506.25W)
(sqrt(50W * 8 Ohm) = 20V
(sqrt(50W * 4 Ohm) = 20V
8 Ohm (50W / 20V = 2.5A)
4 Ohm (100W / 20V = 5A)
It do, to me look like 45V 6A to be sure? but headroom?
Have to point out that I am only trying to learn the theory and logic behind this, since Mooly already have given me a schematic for a simple power supply.
Headroom is going to depend what on how much power you actually need. It’s probably lower than you think. Check out this thread:
A Test. How much Voltage (power) do your speakers need?
Based on this poll, must people (70%) need 25W at most.
Coincidentally, most First Watt amplifiers deliver up to 25 watts.
A Test. How much Voltage (power) do your speakers need?
Based on this poll, must people (70%) need 25W at most.
Coincidentally, most First Watt amplifiers deliver up to 25 watts.
Oh yest that site have I seen before, as my system is right now, do I not have the possibility to use that test.
But headroom, how do you add that to your amp?
I.e. Mooly's 50W?
By the way, it's funny how my calculation shows 500W 4 Ohm, because, as I understand it, by changing it to rms and so on, it do look more like 70W 🙂
But headroom, how do you add that to your amp?
I.e. Mooly's 50W?
By the way, it's funny how my calculation shows 500W 4 Ohm, because, as I understand it, by changing it to rms and so on, it do look more like 70W 🙂
Let's say that the Mooly amp is about 50% efficient. 500W output power means 500W dissipated power in the amp. You would need a 1000W power supply and a really big heatsink.
Why do you think you need more power from this amplifier? You can't really make an educated decision on this unless you know how much power you actually need
Regarding Mooly's amp, your math isn't right. The rails are ±45V DC. That doesn't mean it can push a 45Vrms signal. If the output devices where 100% efficient, it could support 31.8V RMS (45V peaks / square root of 2). Output transistors aren't 100% efficient. And MOSFETs are less efficient that BJTs. And the nature of lateral MOSFETs (Mooly's amp) means they'll support less current that vertical MOSFETs
In LTSPice simulations, Mooly's amp can put out 23.6Vrms into 4 ohm loads before clipping. That comes out to around 140W. At that level, the output MOSFETS are dissipating around 46W. That's going to take a big heat sink to keep the junction temperature reasonable. You'll need a heatsink of 0.40 °C/W or less per channel to keep the junction temperature below 100C.
Having said that, that is from a simulation driving a sine wave into 4 ohms just before clipping. Actual music won't push an amplifier this hard. I believe Rod Elliot's article discusses this.
If you really do need more power, then adding parallel devices to the output stage allows it to handle more current. I believe Prasi has a PCB layout for Mooly's amp that has paralleled lateral MOSFETs.
Regarding Mooly's amp, your math isn't right. The rails are ±45V DC. That doesn't mean it can push a 45Vrms signal. If the output devices where 100% efficient, it could support 31.8V RMS (45V peaks / square root of 2). Output transistors aren't 100% efficient. And MOSFETs are less efficient that BJTs. And the nature of lateral MOSFETs (Mooly's amp) means they'll support less current that vertical MOSFETs
In LTSPice simulations, Mooly's amp can put out 23.6Vrms into 4 ohm loads before clipping. That comes out to around 140W. At that level, the output MOSFETS are dissipating around 46W. That's going to take a big heat sink to keep the junction temperature reasonable. You'll need a heatsink of 0.40 °C/W or less per channel to keep the junction temperature below 100C.
Having said that, that is from a simulation driving a sine wave into 4 ohms just before clipping. Actual music won't push an amplifier this hard. I believe Rod Elliot's article discusses this.
If you really do need more power, then adding parallel devices to the output stage allows it to handle more current. I believe Prasi has a PCB layout for Mooly's amp that has paralleled lateral MOSFETs.
Thanks a lot🙂 I do not think I need more power then Mooly have specified or change anything in his design.
What I am trying to do, is to understand what is going on, how it works and why. 🙂
I know what Mooly have specified is prefect for his amp and I could just read from his great instructions, copy it over ans be done with it.
But that won't teach me anything, so therefore do I ask about it all.
About the parallel fet's, I have by Mooly's great help, chosen some fet's that have two in parallel, build in.
100C sounds way to high for me, I would rather hit 50C instead. But since I plan on building the cabinet myself, is there nothing stopping me in using huge heat sinks.
But what does it take to keep one of the ECW20N20 or the ECW20P20 cooled? I have one fet on each side.
What I am trying to do, is to understand what is going on, how it works and why. 🙂
I know what Mooly have specified is prefect for his amp and I could just read from his great instructions, copy it over ans be done with it.
But that won't teach me anything, so therefore do I ask about it all.
About the parallel fet's, I have by Mooly's great help, chosen some fet's that have two in parallel, build in.
100C sounds way to high for me, I would rather hit 50C instead. But since I plan on building the cabinet myself, is there nothing stopping me in using huge heat sinks.
But what does it take to keep one of the ECW20N20 or the ECW20P20 cooled? I have one fet on each side.
The 100C i mentioned was the junction temp. Not the heatsink temp. The heatsink temp would be lower in that hypothetical situation.
It’s also worth noting that in class AB amplifiers, max dissipation occurs between 70% to 80%, not at max power. So in the scenario I gave, the junction temp would be even higher at 75% power.
It’s also worth noting that in class AB amplifiers, max dissipation occurs between 70% to 80%, not at max power. So in the scenario I gave, the junction temp would be even higher at 75% power.
Last edited:
Yes that did you make quite clear🙂 But it's shortening the component's life a lot if it gets up to about 100C.
A heat sink should be able to lead the heat away from the component, before it get to 100C, as long it follow your advice about 0.40C/W or better.
A heat sink should be able to lead the heat away from the component, before it get to 100C, as long it follow your advice about 0.40C/W or better.
100C or less junction temp is a good target to aim for. But FETs are rated for much higher. As you stated the 100c target is to prolong life.
Also just to clarify, I’m not advocating a 0.40 C/W heatsink. That was under worst case scenario of a sine wave at max power into 4 ohm resistive load. Unless you running a PA, you’re not going to actually experience this
Also just to clarify, I’m not advocating a 0.40 C/W heatsink. That was under worst case scenario of a sine wave at max power into 4 ohm resistive load. Unless you running a PA, you’re not going to actually experience this