Non complementary power MOSFETs generally disregarded by diyaudio community for their limited use as output devices. What if we are looking for good candidates for circlo or quasi like topologies that uses same type devices on both sides? In such cases, number of candidates becomes big but when we become more picky about specs, situation becomes different.
I think experienced members here can name some good but not well known N-MOSFETs regarding such conditions below:
Low (preferably very low) Input Capacitance
High Transconductance
Pd (Watts, higher is better for same or lower capacitance)
Specified Pd can be verified as DC SOA in datasheet
Low Thermal Resistance (especially for TO220 ones)
I came across some devices produced by Alpha & Omega Semiconductor which seems spot on most specs.
eg: AOT7N60, AOT8N50, AOT9N50, AOT9N40
Can you name some other ones?
I think experienced members here can name some good but not well known N-MOSFETs regarding such conditions below:
Low (preferably very low) Input Capacitance
High Transconductance
Pd (Watts, higher is better for same or lower capacitance)
Specified Pd can be verified as DC SOA in datasheet
Low Thermal Resistance (especially for TO220 ones)
I came across some devices produced by Alpha & Omega Semiconductor which seems spot on most specs.
eg: AOT7N60, AOT8N50, AOT9N50, AOT9N40
Can you name some other ones?
Last edited:
They have a lot of on-resistance, and high threshold voltages - eating into headroom...
The physics of MOS devices is working against you, low on-resistance and high transconductance needs higher capacitance to establish the larger channel region, as the charge on the gate mirrors the charge making up the channel.
The physics of MOS devices is working against you, low on-resistance and high transconductance needs higher capacitance to establish the larger channel region, as the charge on the gate mirrors the charge making up the channel.
I got quite a bit of insight when I slurped data from the DigiKey Parametric Search table, in .csv format. Then I opened it on my computer as an excel spreadsheet and added a new column "FOM" (Figure Of Merit).
Since there were a LOT more transistors with an entry in the "Gate Charge Qg max" column, than transistors with an entry in the "Input Capacitance Ciss" column, I decided to use Qg as a proxy for capacitance. So my figure of merit was (Qg * RdsOn). Lower is better!
I had Excel calculate my FOM in the new column, and sorted the entire table by increasing numerical value of FOM. Presto, the ones with the lowest FOM were the most promising devices. So I bought several candidates and measured them, and built them into prototype circuits. It all worked quire well.
ProTip: first tell DigiKey to display 500 items per page, THEN tell it to export the table data in csv format.
caveat: you may need to do a bit of manual editing to convert milli and micro and nano and kilo, into exponential notation numbers.
_
Since there were a LOT more transistors with an entry in the "Gate Charge Qg max" column, than transistors with an entry in the "Input Capacitance Ciss" column, I decided to use Qg as a proxy for capacitance. So my figure of merit was (Qg * RdsOn). Lower is better!
I had Excel calculate my FOM in the new column, and sorted the entire table by increasing numerical value of FOM. Presto, the ones with the lowest FOM were the most promising devices. So I bought several candidates and measured them, and built them into prototype circuits. It all worked quire well.
ProTip: first tell DigiKey to display 500 items per page, THEN tell it to export the table data in csv format.
caveat: you may need to do a bit of manual editing to convert milli and micro and nano and kilo, into exponential notation numbers.
_
Last edited:
I understand the problem regarding threshold voltage value but I think that it doesn't have much influence on dynamic behavior eg: quality reproduction of music signal.
Can you give some insight on significance of Rds-On on audio applications please? I was told that its significance is related with efficiency on switch mode power applications.
I use IXFH22N60P3 in my latest 20W A class amp. I chose them for small Crss and relatively linear parasitic capacitances. Sonically and objectively outperforms IRFP240.
Very good S. Good Ciss relative to Pd but no SOA for DC which makes Pd questionable in linear applications, doesn't it?
How does rds-on matter when using the FET linearly? I think Id rather have a lowish parasitic capacitance than most other parameters. Solid state capacitance is very non linear and should be minimized.
Nelson Pass offers a piece of very practical advice: look at the datasheet curves of Safe Operating Area. If the plot doesn't include a curve for DC (zero hertz) operation, stay away from that device. If there is a DC curve, apply a conservative margin of safety, and forget about thermal runaway.
In linear applications I guess it's worth much less than rated Pdmax=500W. Definitely no problem in small amps like mine with +-25V supply 20W idle dissipation per device.
rds_on is related to the transconductance. When it's very small, chances are that the MOSFET has Spirito instability, although the SOAR should show if it has (with the emphasis on should).
Last edited:
Mark, your DigiKey catalog export-csv-Excel-sort-qualify post is one of the greatest posts ever on this forum. Bravo. 😀
Thank you! I did that in 2013 for the Quasimodo project.
Another diyAudio member did a similar thing for his project: Winfield Hill (co author of the magnificent book "The Art Of Electronics") created his own Figure Of Merit to help him select bipolar transistors. You can read about it in post #8 of his thread
Winfield's 100W DC-10MHz 1000V/us amplifier
Another diyAudio member did a similar thing for his project: Winfield Hill (co author of the magnificent book "The Art Of Electronics") created his own Figure Of Merit to help him select bipolar transistors. You can read about it in post #8 of his thread
Winfield's 100W DC-10MHz 1000V/us amplifier
True. Physics is against us, small geometry for small capacitance gives us low transconductance... The figure of merit that is mentioned gives a good estimate of compromises necessary.