just wondering if power mosfets from infineon are any good or
not , or just stick with visay part irfp140 / irfp9140
Thanks
not , or just stick with visay part irfp140 / irfp9140
Thanks
There is variation between manufacturing process variations from the same brand. Generally more modern processes are less suitable for linear operation.
Maybe you can enlighten us why that is.
Or you have measurements to demonstrate that NOS MOSFETs made decades ago show better linearity than the same device from today's production in the same circuit ?
Modern semiconductor equipment and processes are far more accurate and consistent than those 20 years ago.
That much I know for sure.
But maybe silicone is like wine, needs time to mature ?
🤓
Patrick
Or you have measurements to demonstrate that NOS MOSFETs made decades ago show better linearity than the same device from today's production in the same circuit ?
Modern semiconductor equipment and processes are far more accurate and consistent than those 20 years ago.
That much I know for sure.
But maybe silicone is like wine, needs time to mature ?
🤓
Patrick
Let me please put my two cents on modern MOSFETs.
Yes they looks great - tremendous current capability and power dissipation, but we must be clear 99% of MOSFET application nowadays is switching, and manufacturers optimize for fast switching with minimal losses. The penalty is linear mode vulnerability.
This app. note from Infineon explains in details on page 10.
Then what may stop the manufacturer from making new batches of old parts using new technologies on smaller die?
Yes they looks great - tremendous current capability and power dissipation, but we must be clear 99% of MOSFET application nowadays is switching, and manufacturers optimize for fast switching with minimal losses. The penalty is linear mode vulnerability.
This app. note from Infineon explains in details on page 10.
Then what may stop the manufacturer from making new batches of old parts using new technologies on smaller die?
Last edited:
You can easily determine die size from the Rthjc value.
Patrick
Patrick
The link by @Maserito above is part of the story. When vertical mosfets first appeared ~40 years ago, much was published about the lack of secondary breakdown. Originally vertical mosfets were the Siliconix V-fet type. Since then the designers have switched to a cell structure, similar to IC manufacture. This had the effect of much higher gm for a give area of silicon, but also created a region where current increased with temperature. As processes shrank, new generation mosfets had more and more of their operating range in this region. This did not matter for switching and the reduction in gate capacitance for any Rds was good.
What happened was that linear operation started to get SOA problems, just like BJTs.
Many modern parts are actually worse than BJTs for linear operation - one reason why so few vertical mosfet class AB amplifiers are on the market these days. Decent BJTs are now expensive, so VFets would be used if it was so easy.
Spirito was the author of a paper that highlighted the issue
https://linearaudio.net/sites/linea...mulation-of-power-MOSFETs_12-Jan-2004_PDF.pdf
Are you suggesting that e.g. an IRFP240 from today's production has a different structure and uses a different reticle set than one from 19 or 29 years ago ?
Or are we comparing say a modern IXYS switching MOSFET with the IRFP ?
Post #3, according to my poor understanding, refers to the former case, as it refers to processes and not transistor design, not ?
Patrick
Or are we comparing say a modern IXYS switching MOSFET with the IRFP ?
Post #3, according to my poor understanding, refers to the former case, as it refers to processes and not transistor design, not ?
Patrick
The tell is in the SoA graph. More modern designs are likely optimized for switching and their SoAs are much, much smaller than the older designs. As an example, compare and contrast the original IRF540 against the newer IRF540N.
You have to look carefully at the datasheet. Common generic parts like IRFP240 have changed. Nobody uses 6" or even 4" wafers these days and are running newer generation processes with smaller cells to get the same voltage and current switching from a smaller die - faster and cheaper, which is what 99.9% of their customers want for smps and class D.
Comparing a 1999 Intersil sheet with the 2021 Vishay - the same brand after a long line of takeovers.
Single pule SOA: 40V 10ms was 10A, fallen to 40V 6A
DC line was given in 1999, removed in the 2021 version
The DC curve is the one to use for audio - many speakers go to minimum impedance and wild phase in the reflex port tuning range, so the worst case for SOA.
International Rectifier (IRFx...) and Vishay / Infineon parts are vertical enhancement MOSFETs designed to operate in the ohmic region and not in the pinchoff. These parts are easily identified by their high gm figures. The goal is to get the device to enhancement as soon as possible, with the smallest change in Vgs.
Also, MOSFETs are positive temperature coefficient (PTC) devices only with sufficient drain current and not immediately above the threshold, as we see in the picture below for IRF540N (example).
However, IR used to advertise their MOSFETs as being suitable for linear amplifiers in their own application note AN-948 (attached).
Also, MOSFETs are positive temperature coefficient (PTC) devices only with sufficient drain current and not immediately above the threshold, as we see in the picture below for IRF540N (example).
However, IR used to advertise their MOSFETs as being suitable for linear amplifiers in their own application note AN-948 (attached).
This AN-948 appears to have been created in 1996, back before this SOA issue was discussed. I remember Creek selling a range of vertical mosfet amplifiers back then, that suddenly stated getting a reputation for blowing up. This was likely due to the mosfets becoming vulnerable to SOA.
Yes, AN-948 was discontinued, as IR possibly learnt their lesson. Some of their parts like IRF610/9610 etc. are still used however.