The simplistic Salas low voltage shunt regulator

Status
Not open for further replies.
is something like this a good idea?

mosfet instead of bjt...
 

Attachments

  • circuit121.jpg
    circuit121.jpg
    33.8 KB · Views: 596
Last edited:
To put it a nice way...

Would someone please point me to the most current
SALAS PSU and Phono Stage BOM, Board GB, Kits, Etc...
Much appreciated..to bad this platform still runs on 20 year old software and it is impossible to find anything, unless you got 897455 hours to spend on reading threads with 10304808 posts....Very educational and all, but after spending 65 hours entertaining myself I would like to have some satisfaction and build some gear....Please have mercy!

Latest linear low noise PSU:
Latest CNC, schematics:
Latest Salas PSU:
Latest Salas Phono:

Bom, Schematics, PCB purchase options.

Thanks a million! (Am I even on the right forum?_
 
BTW have you measured BSP129 IDSS ?. I have 20pcs or so and while it starts at ~ 60mA very soon it drifts up fast while still warmish not hot. Talking 3 times up and going.

isnt it typical of mosfets to change current if Vgs remains the same...?

i always used a few ohms to degenerate and it was not a problem if i remember...

you want to try a b1 style buffer if i remember correctly?
 
yes...one would have to watch for a nice Vds without overheating...

still i like the idea very much and will probably be my next regulator CCS build

wich rises the question again....where are those to-92 heatsinks hiding on this planet?

that is why i make my own minimalist ones...
 

yes i know these exist.

though i have not tried them, i m wondering how effcient they really are. the part that joins the ''helmet'' to the actual heasink looks like it has a smallish cross-section, which would limit their range of effectiveness. Maybe the to-92 are not made for high level of dissipation, but with the zetex e-line to-92 style mosfet, i succesfully put around 1.250 watts. with the face against a ''big'' heatsink(530002B02500 Aavid Thermalloy | Mouser) the device casing remained at almost room temperature. but when i put one device on each side facing with 2.5w total, the thickness of the aluminium,the portion between the two devices, proved to be to thin to acommodate such power, and as a result the device casing heated up considerably compared to room temperatue.
 
i thought the following might inrest some.

today, i tested the fairchild j111 package temperature for different wattage, using a simple but effective method.the goal is to see what kind of wattage is ''safe'' for my minimalist heatsink.

the face of the transistor (part that diffuses the most heat) was place against a 1mm or so thick aluminium piece and with my thumb against the backside of the jfet. paste was used in between.

with an idss of around 100mA, the jfet was tested with different Vds and it seems that 1 watt(10Vds) is about the limit where i can feel my thumb having a marked rise in sensation of heat. in other words, i could actually feel my thumb sinking heat.

so take the result as seriously as you want, but i personally would not put more than 1 W on this jfet with the minimalist heatsink i built (jfet squeezed between 2 aluminium bars, or plate, depending on how you shape it).

for more wattage, you would have to pot the whole thing, or use a helmet style fitting and even then i,m not sure if the nodes inside the casing would nt overheat anyway, regardless of the heatsink...
 
Last edited:
Status
Not open for further replies.