The simple answer is just set it for 100ma per device.
The bias current is not critical, between 100ma and 200ma per deivice will work OK for class AB operation. Lower runs cooler, which I prefer. Borbely recommends about 150ma for Hitachi lateral MOSFETs, while Pass recommends about 100 ma for IR vertical MOSFETs
The optimal way to set this might be the slick method detailed in the Nelson Pass article on the Citation MF12, where you set it based on minimizing crossover distortion using only a scope, oscillator and power resistor.
http://www.passdiy.com/pdf/citation.pdf
Even though that MF12 implementation used vertical MOSFETs, and you are using laterals, the procedure is identical. The bias voltage will of course be quite different, perhaps 1 volt versus about 7-8 volts for the vertical MOSFETs. Both result in the same bias current and minimal distortion.
The bias current is not critical, between 100ma and 200ma per deivice will work OK for class AB operation. Lower runs cooler, which I prefer. Borbely recommends about 150ma for Hitachi lateral MOSFETs, while Pass recommends about 100 ma for IR vertical MOSFETs
The optimal way to set this might be the slick method detailed in the Nelson Pass article on the Citation MF12, where you set it based on minimizing crossover distortion using only a scope, oscillator and power resistor.
http://www.passdiy.com/pdf/citation.pdf
Even though that MF12 implementation used vertical MOSFETs, and you are using laterals, the procedure is identical. The bias voltage will of course be quite different, perhaps 1 volt versus about 7-8 volts for the vertical MOSFETs. Both result in the same bias current and minimal distortion.
Here's a little more information:
1) 100 mA per device is a good starting point. Below that and you be in the area of a positive tempco, which isn't ideal.
2) There won't really be an optimum value that will minimize crossover distortion. There is an optimum bias for bipolars (besides full-on class-A), but not FETs.
3) Performance (sonic and measured) with FETs will basically increase without limit as the bias is increased. Unfortunately, so does the heat.
I would start at 100 mA, and go up from there. If you can rest your hand on the heatsink for a few seconds before it becomes uncomfortably hot, that is around 125 F. I wouldn't go much hotter than that or else you will risk danger of burns and also shortened component life.
1) 100 mA per device is a good starting point. Below that and you be in the area of a positive tempco, which isn't ideal.
2) There won't really be an optimum value that will minimize crossover distortion. There is an optimum bias for bipolars (besides full-on class-A), but not FETs.
3) Performance (sonic and measured) with FETs will basically increase without limit as the bias is increased. Unfortunately, so does the heat.
I would start at 100 mA, and go up from there. If you can rest your hand on the heatsink for a few seconds before it becomes uncomfortably hot, that is around 125 F. I wouldn't go much hotter than that or else you will risk danger of burns and also shortened component life.
According with slowhands and guymrob 100Ma in AB class is hr good way. In 5Ch Zenquito (SK1058/SJ162 push-pull) the front center CH and the rear Ch have 220mA/push-pull. The 2 front Ch have 1.12A/push-pull to have A Class up to 20W
Marc
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Marc
"However in one of Slone's books he does explicityly say that he thinks the optimal bias current for these devices is 40mA. There is no explanation given though."
He changed his figures between books. In the earlier one he wrote 50mA between the source and output rail per device. Is the 100mA figure quoted here for each device or is it for each pair?
In a later book he uses the figure of 40mA to multiply by the sum of both of the source resistors to get a voltage figure, voltages being a little more convenient to measure.
Having built only bipolar amps so far, but contemplating a MOSFET effort entually, I'm curious to know if this is just a question of how the figure is stated or if there is actually a 2:1 difference in recomended bias measurement.
He changed his figures between books. In the earlier one he wrote 50mA between the source and output rail per device. Is the 100mA figure quoted here for each device or is it for each pair?
In a later book he uses the figure of 40mA to multiply by the sum of both of the source resistors to get a voltage figure, voltages being a little more convenient to measure.
Having built only bipolar amps so far, but contemplating a MOSFET effort entually, I'm curious to know if this is just a question of how the figure is stated or if there is actually a 2:1 difference in recomended bias measurement.
As we speak about class AB (regarding values of quiescent current), the resulting transfer characteristics of the output stage is composed by addition of those of the N and the P MOSFET. Gm of the MOSFETs considerably dependends on Id. In the cross-over region, where the both devices conduct, the Id is small and so is the Gm. Sharper changes in the Gm will result in higher harmonics. I would suggest the use of simulation programm to find the optimal quiescent current for the MOSFETs used. And then you need temperature compensation circuit to maintain reasonable drift of the quiescent current.
Well, I have simulated with those devices. They are pretty linear and outpermorming commonly used IRF's. You will get reasonable results starting from quiescent current of 100mA. The higher the current, the better linearity and less distortion. 1A is quite ideal and sufficient even for non-feedback output stage. But this is a class A then. For class AB, I got the optimal quiescent current of some 370 mA. Very low distortion, predominantly 2nd and 3rd harmonics. Very quick decay of higher order harmonics.
waaay better in terms of resolution and 'imaging'.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.