Agreed now consider like pass amps if we use 20 mosfets per rail what is the advantage? Reduced distorion is one as said but I always had this doubt regarding when Hexfets are used the bias required to get the Hexfet into linear operation is about 350ma per Mosfet for IRFP240 and 200ma for 9240 as per the Using HEXFETs in High Fidelity Audio
Now when you use 20 mosfets per rail. I believe Pass labs have about 70+ Hexfets in total so the quiescent current is definitely low per transistor so how far that helps to get the hexfets into linearity.
I agree the fact that they are true balanced designs but when you look at such a huge amount of mosfets how is linearity achieved? Or is it like when you draw little current through hexfets they behave linearliy?
You mix two different things. Rod is talking about the gate charging current, you mix it with the quiescent (drain) current of the output devices.
Now - bias is actually a voltage. Of course, there is certain drain current, associated with particular bias voltage (source-to-gate). As soon as you set some bias voltage at the gates, it is the same for each output transistor, connected in parallel, leading to roughly the same quiescent current at each drain.
Here are the results of practical tests, measurements and listening exercises:
1) CF-FET V2.0 amplifier - 5 pairs of IRFP240/9240, NJW0281/0302 driver stage. Quiescent current of each output pair = 90mA (450mA total). Quiescent current of the drivers = 18mA, at high output power, gate charging current peaks may reach 350mA. Very well-performing, powerful, high-speed OPS.
2) VHex+ amplifier - 2 pairs of IRFP240/9240, 3503/1381 driver stage. Quiescent current of each output pair = 80...100mA (160...200mA total). Quiescent current of the drivers = 12mA, at high output power, gate charging current peaks may reach 120mA. Excellent performance at this power rating (around 100W @ 8 ohm).
so how is gate charging current can be controlled?
What is the point in using 40 hexfets per channel? any specific advantage?
I believe high current delivery possible? Better grip on speakers?
Lower distortion just like in BJT?
Higher damping factor? ofcourse higher DF depends on design but in general what are the benefits?
I want to build a subwoofer amplifier at +/-75VDC for 200W into 8ohm and amost 400W in 4ohms so how many IRF pairs required in the above application?
Its also said that Hexfets doesnt have secondary breakdown so what if the load impedance dips to 1ohm and the demanding current into load will be much higher will the hexfet blows up?
Multiply your target output wattage by 5, the divide this by the wattage rating of your output devices to calculate how many devices you need in total. Hexfets have a lot of margin for error. They are hard to kill.
Gate charging current depends on overall gate capacitance (rather high for HexFETs). More output devices -> higher gate capacitance -> higher gate charging current peaks. One of the ways to keep it limited is split the output devices into groups and arrange a separate driver stage for each group. In this case, you have to be careful with overall driver stage input impedance, as it will be lower (single group impedance divided by number of groups).
Key advantages of using big number of output devices:
- high power capability;
- high current capability;
- lower output impedance (not that important in feedback amplifiers as the feedback reduces it dramatically);
- higher overall quiescent current, resulting in wider class A region.
You have to keep in mind, that your PSU power/current capability must be sufficient for feeding this kind of thing. Good soft-start/protection must be in place if you're looking for safe operation.
400W amp engineering is significantly more complicated, requiring more care and attention, than, say, 50-100W amp engineering. Good topology is 30% of success. Layout nuances, heat dissipation matters, overall build quality are critically important.
The question is - do you want to go through the whole process yourself (most likely involving a number of error / redesign iterations), or you'd like to have a much more predictable result, building some well-engineered and tested design?
For example, Ostripper's Slewmaster will fit your requirements perfectly. If you'd like to have some extra output current capability - build it with MT-200 Sanken devices. They are not HexFEts, but ... who cares, if they work perfectly?
So consider i need 400watts in 4ohms so the math goes like this
400 x 5 = 2000
2000/150W = 13.33 which is 14. So 14 per rail or 7 output pairs?
That would be 14 devices total, 7 output pairs. This will give you lots of headroom for impedance dips in the speakers frequency response.
So with that formula, MJL4281/4302 are 230W devices so you would need 9 devices so the arcwelder OPS should do it with ease.
Yes the Arc Welder OPS is very capable as is. Modifying it to run different devices doesn't really make a lot of sense. MJL4281/4302 or Sanken MT-200s will allow it to drive just about anything.
The calculations are in Watts. Voltage doesn't matter. Just use the formula as a rule of thumb for the intended load you expect to drive.
can anyone advice me on the following:
1. Does higher open loop gain at lower frequency will increase control in bass?
2. Does CFA with higher slew rate will have good transients across the entire spectrum?
3. In Kypton V2 i have tested both bjt and Hexfet and i found hexfet has phenomenal grip and more definition in bass even at 80ma quiescent per Hexfet.
4. Now if i want to build a subwoofer amplifier so which one is preferred?
single differential or double differential bjt input with hexfet output with current limiter?
1. Does higher open loop gain at lower frequency will increase control in bass?
2. Does CFA with higher slew rate will have good transients across the entire spectrum?
3. In Kypton V2 i have tested both bjt and Hexfet and i found hexfet has phenomenal grip and more definition in bass even at 80ma quiescent per Hexfet.
4. Now if i want to build a subwoofer amplifier so which one is preferred?
single differential or double differential bjt input with hexfet output with current limiter?
I was just simulating with 5 pairs of hexfets in OPS at +/-73V with any quiescent current in OPS in nominal range from 0ma to 100ma the distortion at 100Hz is about 0.17 even at 0ma or 100ma which i think is good enough for subwoofer. Now my main concern is the heat dissipation. Im thinking of using Fan and thermal shutoff but at what temperature i need to thermal shut down the amp? at 60 degrees 70 degrees?
Last edited:
What conditions are you talking about?
Load impedance, output voltage RMS, output power RMS.
What is the front-end that you're using with this OPS?
0.17% with zero bias is possible with enough feedback, however it will sound bad because of very nasty nature of crossover distortion. It will sound good at 100...150mA per output pair. If, setting 100mA per pair, you still see 0.17% THD - something is wrong with your front-end. It must be much lower with higher bias.