I plan to set my bias current to zero on my class AB amp so that it could work in class B. The main reason is I will not have to deal with thermal issues on bias current. The other reason is, I don't see the need of it because I will use this amp for low frequency (sub Beyma 18G550 )so <500Hz and at these frequencies the sinwave looks clean without bias current. (Amp loaded by 8 ohm)
Any comment is welcome.
David
Any comment is welcome.
David
Well, if it works for you... 😉
If I had to make a choice, I would go on class D for a sub 😀
Pay attention that maybe under stress and heat the thermal compensation might want to reduce further the bias, you'll end up with a class C amp.
If I had to make a choice, I would go on class D for a sub 😀
Pay attention that maybe under stress and heat the thermal compensation might want to reduce further the bias, you'll end up with a class C amp.
nice advice thanks, I didn't think about it ... I'll check
I'll go class D but I'm looking for a simple schematic to start
David
I'll go class D but I'm looking for a simple schematic to start
David
It would be better to choose a very low but non-zero bias. Perhaps 0.4-0.5V Vbe. You will still get the thermal stability, but distortion will be a lot lower.
You have to be careful here. Crossover distortion may be heard as a "buzz" from the speaker. Depends a lot on the speaker and amount of distortion.
Mr. Evil has the best suggestion, with a bias circuit.
-Chris
Mr. Evil has the best suggestion, with a bias circuit.
-Chris
Why not just use a thermal compensation servo circuit like most class AB circuits do, and keep the bias in class AB but make the compensation factor larger so that when the outputs circuit get hotter the bias drops faster and into class B operation. Maybe use a higher gain device for the servo that has a larger temperature coefficient. This would reduce the "buzz" of crossover distortion, even to a low frequency. Of course if you are using a band pass speaker box, where the diaphram is not on the outside, who gives a hoot about a little crossover.
I think your bigger concern might be SOA and whether or not the circuit can handle the inductance, i. e. phase shift, of a heavy sub-woofer without letting the smoke out of the outputs. 😉
I think your bigger concern might be SOA and whether or not the circuit can handle the inductance, i. e. phase shift, of a heavy sub-woofer without letting the smoke out of the outputs. 😉
djdamix
Why? What's wrong with your AB amp that you need to set the bias to zero? Is it thermally compromised and you wan't to run it hard?
Why? What's wrong with your AB amp that you need to set the bias to zero? Is it thermally compromised and you wan't to run it hard?
cunningham =>can you explain further about phase sift and the SOA. Will I have to dissipate more power than with resistive load ??? How can I test it ? What inductance value ?
Thanks.
David
Thanks.
David
amplifierguru=> In fact the thermal compensation could be ok, (thanks to diyaudio members) but I think the amp will be less hot and it could be better and maybe more relyable : On my previous amp, when the amp turn on the bias curent rise to 600mA and then stabilize to 400mA. Maybe in some bad condition (low temperature or else, it could rise up to 600mA when turn on) maybe it is not a good thing. Furthermore I have to put a transistor for the VBE multiplier on one of the output mos and I need to make a mechanical piece to do that.
Thanks
David
Thanks
David
Lets see if I can get this right... In an inductive load, current lags the voltage, and a capacitive load, current leads the voltage.
At any rate, the current and voltage is out of phase. So when the voltage rises up to peak and then to zero on a resistive load, the current rises up to peak at the same time. So when the current is peaked, the voltage across the output transistor is at minimum. When the voltage is around zero, the full rail voltage is across the transistor but the current is zero.
On a reactive load, the voltage on the transistor may be half way to zero when the peak current flows, thus possibly exceeding the SOA of the device. Or the output voltage could be around zero when like half the peak current flows and all the rail voltage is across the transistor and may exceed the SOA. Anyway, you get the point?? 😉
This is where some people use paralleled outputs to drive heavy (inductive) sub-woofers with lots of power.
At any rate, the current and voltage is out of phase. So when the voltage rises up to peak and then to zero on a resistive load, the current rises up to peak at the same time. So when the current is peaked, the voltage across the output transistor is at minimum. When the voltage is around zero, the full rail voltage is across the transistor but the current is zero.
On a reactive load, the voltage on the transistor may be half way to zero when the peak current flows, thus possibly exceeding the SOA of the device. Or the output voltage could be around zero when like half the peak current flows and all the rail voltage is across the transistor and may exceed the SOA. Anyway, you get the point?? 😉
This is where some people use paralleled outputs to drive heavy (inductive) sub-woofers with lots of power.
thanks for the explanation ;-) I understand. Muy sub is a 18G550 (BEYMA) in bandpass box. Do you think it is an heavy inductive load ?
David
David
Do you need your butt kicked?
How about a 2.1KW class D amplifier for sub use; with a subsonic filter, 40hz~160hz crossover, all for the low price of $299?
What's it called?
Of course its the ButtKicker BKA1000
http://www.martinsoundpro.com/upload/item/7042.JPG
How about a 2.1KW class D amplifier for sub use; with a subsonic filter, 40hz~160hz crossover, all for the low price of $299?
What's it called?
Of course its the ButtKicker BKA1000
http://www.martinsoundpro.com/upload/item/7042.JPG
There is a 4.5 " diameter coil of wire inside this speaker having 1.4 mH of inductance...so Yes it has inductance. This speaker will require a heck of an amp to run to full power. With this monster inside a bandpass box, I would probably look into class D. Works a bit different but much more efficient. The problems with class D sould not be noticed from a bandpass low freq. sub box. Why amplify it when you can just switch it!...😀 😀
Building a class D amp is difficult. But if you have the pcb already designed and have good luck finding the right parts, it would be great. I still suggest buying the already-made class D modules at www.hypex.nl.
ByE!
ByE!
If you are using a MOS amp, SOA will be a far easyer thing to satisfy than for BJTs.
Regarding bias, I really don't see the need to use 600mA bias for a subwoofer application. Of course, unless you are talking many pairs of MOSFETs in parallel and the total bias current for all of them. Even so, there are a few things to consider:
1) A subwoofer normally has resonant peaks over it's ohmic resistance (usually this is lower than the rated impedance) unless you are doing really strange things like 1 and 1/2 way boxes etc. Because of this, even though it will behave very inductively and also very capacitively (depending on which side of the resonant peak(s) you are frequency wise), it will not necessairly be extremely difficult to drive. Depending on the Q factor it may be more, equally or less difficult than a pure resistive load, but you will not see something like double dissipation from the amp on this account. This is because even though voltages across the output transistors rise, impedance rises also so currents fall.
2) Very reactive loads will wreak havoc with an aplifier which is purely class B, especially if it is MOSFET based as in this case you have a relatively large 'dead zone' around zero. When the output stage crosses zero, it suddenly becomes very high impedance - and your inductive load is still trying to pass the same current. Net result is, in order for the same current to pass through the output of an amp which had just had an impedance on the rder of milliohms, and suddenly became kiloohms, voltage must rise by the same factor. You get all sorts of problems, from oscillations in the dead zone (as feedback is trying to get the amp to clamp the inductive kickback), to voltage spikes. Because of this, you need to at least operate the amp in very lean class AB - for a MOSFET this could be as little as 20mA idle current. Incidentally, in BJT output stages that are run like this, or have SOA protection, this is the very reason why reverse polarized diodes from output to supply rails MUST be used, or you will have dead otput transistors. Another consideration are output fuses if present - when they become open, arcing is possible due to the inductive component of the load. I even had a case where I accidentally removed a rail fuse out of an amp driving a speaker and got a REALLY bad electric shock from it - normally very unlikely from a 28V rail!
3. In a class AB amp driving a subwoofer, you want the amp to use the most of the power supply at it's disposal. Hence supplying the driver stages with higher rails, in order to get the output as close to rail to rail swing as possible, is in order. This will reduce power dissipation somewhat. In case you don't have extra transformer taps, the rescue comes from the seemingly forgotten art of voltage doublers 😉 which may be made deliberately underdimensioned, in order to get less than double the voltage, since you only need 5-10V more than the output power rails.
4. Driving bandpass and/or other ported boxes may not require as much power as one would think. You may run into the maximum excursion limit of the speaker long before you run into it's thermal power limit - this is something that needs serious consideration. Special attention needs to be given to subsonic filtering - absence of it is usually the reason for dead subwoofers, even those capable of withstanding many times the power of the connected amp.
Regarding bias, I really don't see the need to use 600mA bias for a subwoofer application. Of course, unless you are talking many pairs of MOSFETs in parallel and the total bias current for all of them. Even so, there are a few things to consider:
1) A subwoofer normally has resonant peaks over it's ohmic resistance (usually this is lower than the rated impedance) unless you are doing really strange things like 1 and 1/2 way boxes etc. Because of this, even though it will behave very inductively and also very capacitively (depending on which side of the resonant peak(s) you are frequency wise), it will not necessairly be extremely difficult to drive. Depending on the Q factor it may be more, equally or less difficult than a pure resistive load, but you will not see something like double dissipation from the amp on this account. This is because even though voltages across the output transistors rise, impedance rises also so currents fall.
2) Very reactive loads will wreak havoc with an aplifier which is purely class B, especially if it is MOSFET based as in this case you have a relatively large 'dead zone' around zero. When the output stage crosses zero, it suddenly becomes very high impedance - and your inductive load is still trying to pass the same current. Net result is, in order for the same current to pass through the output of an amp which had just had an impedance on the rder of milliohms, and suddenly became kiloohms, voltage must rise by the same factor. You get all sorts of problems, from oscillations in the dead zone (as feedback is trying to get the amp to clamp the inductive kickback), to voltage spikes. Because of this, you need to at least operate the amp in very lean class AB - for a MOSFET this could be as little as 20mA idle current. Incidentally, in BJT output stages that are run like this, or have SOA protection, this is the very reason why reverse polarized diodes from output to supply rails MUST be used, or you will have dead otput transistors. Another consideration are output fuses if present - when they become open, arcing is possible due to the inductive component of the load. I even had a case where I accidentally removed a rail fuse out of an amp driving a speaker and got a REALLY bad electric shock from it - normally very unlikely from a 28V rail!
3. In a class AB amp driving a subwoofer, you want the amp to use the most of the power supply at it's disposal. Hence supplying the driver stages with higher rails, in order to get the output as close to rail to rail swing as possible, is in order. This will reduce power dissipation somewhat. In case you don't have extra transformer taps, the rescue comes from the seemingly forgotten art of voltage doublers 😉 which may be made deliberately underdimensioned, in order to get less than double the voltage, since you only need 5-10V more than the output power rails.
4. Driving bandpass and/or other ported boxes may not require as much power as one would think. You may run into the maximum excursion limit of the speaker long before you run into it's thermal power limit - this is something that needs serious consideration. Special attention needs to be given to subsonic filtering - absence of it is usually the reason for dead subwoofers, even those capable of withstanding many times the power of the connected amp.
- Status
- Not open for further replies.
- Home
- Amplifiers
- Solid State
- Class B for low frequency ?