Aside from lateral MOSFET types specifically marketed towards audio are there any complementary MOSFETs that are reasonably well matched? I've been looking around at various products and data sheets and it seems that there aren't many complementary power MOSFETs out there that aren't laterals.
Can you guys recommend pairs known to have reasonably complementary parameters, even if they span different product families or even manufacturers? Preferably readily available current production parts, if possible.
I'd like to order for my stock some different parts to experiment with but since I've been a BJT kind of guy for so long I'm not familiar with what may be available.
Can you guys recommend pairs known to have reasonably complementary parameters, even if they span different product families or even manufacturers? Preferably readily available current production parts, if possible.
I'd like to order for my stock some different parts to experiment with but since I've been a BJT kind of guy for so long I'm not familiar with what may be available.
Since all vertical mosfets i know of are made for switching, there are no true complementary ones, just ones that are very similar, but not close enough to be true complementary.
For example, the P channel device always have atleast 0.1 ohm on resistance while the N channel one is a few tens of milliohms.
For example, the P channel device always have atleast 0.1 ohm on resistance while the N channel one is a few tens of milliohms.
In MOSFET selection for linear amplifiers Rds(on) is one characteristic that I totally ignore.
I would hope I am not listening to music that is hitting the rail supply. 🙂
Gate capacitance, etc is far more important to a complimentary MOSFET matching.
Max V, max I, SOA, Pdiss, don't need to match, just make sure they exceed requirements.
Transconductance should be as close as possible.
IMHO 🙂
I would hope I am not listening to music that is hitting the rail supply. 🙂
Gate capacitance, etc is far more important to a complimentary MOSFET matching.
Max V, max I, SOA, Pdiss, don't need to match, just make sure they exceed requirements.
Transconductance should be as close as possible.
IMHO 🙂
OK, perhaps I'm asking the wrong questions. I'm still interested to get opinions on what devices are 'close enough' to use where typically complementary devices would be used, be specific.
As far as selection goes, let's see if I'm on the right track.
1) Based on application choose a P channel device that meets or exceeds the requirements since there are fewer of those from which to choose.
2) Based on the P channel device characteristics then begin to seek a 'best fit' complement from the plethora of N channel devices.
So, based on my initial selection process what are the parameters to try and match? What order to consider them in? As DUG suggested certain parameters can take a back seat while others are more important. I'd like to set it straight in my mind what is important and what can be safely ignored.
As far as selection goes, let's see if I'm on the right track.
1) Based on application choose a P channel device that meets or exceeds the requirements since there are fewer of those from which to choose.
2) Based on the P channel device characteristics then begin to seek a 'best fit' complement from the plethora of N channel devices.
So, based on my initial selection process what are the parameters to try and match? What order to consider them in? As DUG suggested certain parameters can take a back seat while others are more important. I'd like to set it straight in my mind what is important and what can be safely ignored.
I use FQP50N06 & FQP47P06 in my EC mosfet stereo amp. They are planer stripe devices, very rugged, very cheap.😉 I understand the fabrication process is a bit cheaper than for hex type and similar 'cellular' die structures. Stear clear of certain architectures such as Trench fets, U-fets, and similar types. These are not suitable for analog operation.
Because the P-ch has a different input capacitance and gate charge requirements vs Gm than the N-ch, I like to use a totem pole driving stage for each gate, something similar to a diamond buffer, that provides a seperate path for gate charge from the N-ch and P-ch devices to the driver stage.
Because the P-ch has a different input capacitance and gate charge requirements vs Gm than the N-ch, I like to use a totem pole driving stage for each gate, something similar to a diamond buffer, that provides a seperate path for gate charge from the N-ch and P-ch devices to the driver stage.
No one else has any suggestions for V-FET 'pairs'? Based on some searching of datasheets I think these may be candidates..
http://www.fairchildsemi.com/ds/FQ/FQA36P15.pdf
http://www.fairchildsemi.com/ds/FQ/FQA46N15.pdf
http://www.fairchildsemi.com/ds/FQ/FQA36P15.pdf
http://www.fairchildsemi.com/ds/FQ/FQA46N15.pdf
Jason (post 6),
http://www.fairchildsemi.com/ds/FQ/FQA36P15.pdf
http://www.fairchildsemi.com/ds/FQ/FQA46N15.pdf
Can't seem to be found.
http://www.fairchildsemi.com/ds/FQ/FQA36P15.pdf
http://www.fairchildsemi.com/ds/FQ/FQA46N15.pdf
Can't seem to be found.
I have used FQA36P15 and FQA40N25.
These are not well suitable to audio, but their 300W Pd rates and their cheap make them attractive.
Both have high gm, around 30S.
It is mandatory that 47R and 220pF series snubber between gate and drain to prevent cross conduction.
Powerful gate drive is required, but they perform well.
Cheers,
Hugh
These are not well suitable to audio, but their 300W Pd rates and their cheap make them attractive.
Both have high gm, around 30S.
It is mandatory that 47R and 220pF series snubber between gate and drain to prevent cross conduction.
Powerful gate drive is required, but they perform well.
Cheers,
Hugh
Junm: "You mean Crss, Coss and Ciss..."
Yes, I grouped them together as I usually work with these parameters in a switching mode where Ciss and Crss interact (as Miller capacitance) to create a simpler model of "Gate Charge".
For linear operation Ciss and Crss still interact but not quite in the same way.
I would match Ciss as a first priority. Crss second and I don't use too much calculating time on Coss.
You will find that if you get P and N channel closely matched for the C's then the max current can be as much as 60% different.
That is why I think that:
"Max V, max I, SOA, Pdiss, don't need to match, just make sure they exceed requirements."
There will be variations in the gain curve of each but I feel that the feedback circuit should take care of that.
IMHO
Yes, I grouped them together as I usually work with these parameters in a switching mode where Ciss and Crss interact (as Miller capacitance) to create a simpler model of "Gate Charge".
For linear operation Ciss and Crss still interact but not quite in the same way.
I would match Ciss as a first priority. Crss second and I don't use too much calculating time on Coss.
You will find that if you get P and N channel closely matched for the C's then the max current can be as much as 60% different.
That is why I think that:
"Max V, max I, SOA, Pdiss, don't need to match, just make sure they exceed requirements."
There will be variations in the gain curve of each but I feel that the feedback circuit should take care of that.
IMHO
Yes, it seems the only choice from Fairchild for the P-channel device is the FQA36P15. Other N-channel that look like they may play well with the noted P-channel device are the FQA28N15 and FQA32N20C.
Ok, so I agree that voltage, current, dissipation are not important parameters to match so long as the weakest link is strong enough so to speak. I have been looking at input capacitance, transconductance and threshold voltage as my primary parameters to match as close as possible. I figure the closer the devices match in the most critical parameters, the more the NFB goes to keeping the amplifier linear and less to making up for grossly different devices in the first place.
So, the IRFP240/IRFP9240 appear to be in common use. What else are folks using? I'm looking to see what is being used in the DIY community for MOSFETs that are relatively inexpensive and readily available. If you have a favourite set I'd like to hear from you!
Ok, so I agree that voltage, current, dissipation are not important parameters to match so long as the weakest link is strong enough so to speak. I have been looking at input capacitance, transconductance and threshold voltage as my primary parameters to match as close as possible. I figure the closer the devices match in the most critical parameters, the more the NFB goes to keeping the amplifier linear and less to making up for grossly different devices in the first place.
So, the IRFP240/IRFP9240 appear to be in common use. What else are folks using? I'm looking to see what is being used in the DIY community for MOSFETs that are relatively inexpensive and readily available. If you have a favourite set I'd like to hear from you!
So, an equivalent p-channel mosfet will always have a larger die than an equivalent n-channel device. Maybe try matching an N-channel part with a p-channel one die size larger. 9540 vs. 530, for example.
It depends on the topology as to which capacitance is more effective to the circuit. If you intend to use them as complementary source follower, Cgs is for the most part bootstrapped. You have the change in Vgs vs Id that will be effected by Cgs directly. If the load is driven by the drain then Cgs is directly driven by the previous stage. Cgd is really the problem when dealing with complementary follower. When Vds becomes small at clipping, Cgd increases exponentially and can reak havoc. (This is where a lot of the distortion mosfets create come from.) For this reason a strong driver stage is needed to reduce distorton, prevent rail sticking, cross conduction leading to failed components. Also Cgd, being a variable dependent on Vds, can cause the mosfet to resonate at a frequency dependent on that and the lead and PCB trace inductances. This usually happens under load when the signal peaks and Vds is reduced. There are simple ways to remedy this.😉
Due to this fact, the P-ch of equal Gm tends to be more robust. Also the input capactitance is higher. This is good because the N-channel is usually much cheaper so if you design the circuit so that a burnt ouput device results in deactivation of the amplifier output stage(s) then the fault of overdriving the amp would be on the cheaper N-ch device. This way if you overdrive and blow the OPS, you simply replace a transistor that costs maybe a dollar.😛
Due to this fact, the P-ch of equal Gm tends to be more robust. Also the input capactitance is higher. This is good because the N-channel is usually much cheaper so if you design the circuit so that a burnt ouput device results in deactivation of the amplifier output stage(s) then the fault of overdriving the amp would be on the cheaper N-ch device. This way if you overdrive and blow the OPS, you simply replace a transistor that costs maybe a dollar.😛
Last edited:
- Status
- Not open for further replies.