Jfet ccs for cathode loading

[5v333]

Hi.

Ive been playing with active cathode load in my NYD EQ tube PP gain stage.
First i used shared leds/diodes to set bias with great results.
but when i tried a J112 with a small degenerative resistor to set bias voltage (6V about 18ma) i knew that i found the lost treasure on the secret treasure island.
so i wanted to try the same with the first tubes (12ay7) which should be set to about 1.5V-3V and about 1.5ma together.
turned out a J113 in this case just burnt out. same with a 2sk170gr...

Why is this happening? too low VDS?

can i use a Jfet that suits the task or should i use BJT ccs or back to the leds/diodes?

 

[JohnRoberts]

You can cascode the JFET current source with a higher voltage bipolar, or just use a bipolar current source, if you need more breakdown voltage.

That's why they publish data sheets. 

JR 

 

[Winston OBoogie]

What's the voltage  of the rail you're using to bias the 12AY7 pair?
A K170GR is a pretty decent CCS for between 1mA to 5mA.  Max drain source voltage should be limited to 15 volts or a less though to avoid gate leakage.  IOW, don't go by the spec sheet value of 40 volts for best performance.
Easiest cascode to help with keeping voltage within  SOA (and a bit quieter than a BJT) would be another JFET with its gate tied to the lower 170GR's source.

 

[5v333]

I forgot to say that i struggle to understand transistors... tubes are more logical to me...

and here is a pic over how i want it to look like.

I like the simplicity a lot!

 

[Winston OBoogie]

Ah I see, you don't have a negative rail for biasing.  In that case, your problem is too "little" voltage rather than too much, I doubt there is enough voltage to get the K170 biased up and it's trying to push the cathodes up thus increasing valve currents.  Probably died fighting for its life and couldn't run to fight another day because it was boxed in. 
And it's already a low pinch off voltage part (circa .2V) so your other JFets won't help.
Rig up a -12V rail if you want to try this approach.

 

[5v333]

alright then.

thanks.

 

[Winston OBoogie]

alright then.

thanks.

Yep.  I think it's probably worth your while trying it out.  At least you'll have tried it out and will gain an opinion.
JFets are generally quite friendly little things so don't let the death of the last two put you off

 

[5v333]

unfortunatly i dont have any source for negative supply in the psu...

i was thinking about stealing a bit from the heater winding with 2 diodes and filtering but i dont know if that could work out...

 

[Winston OBoogie]

You could try it out with a 9V battery first, see if it's something you want to go ahead with.  Tie battery + to your circuit ground and use the - for the JFet. 
It's possible that tapping off with a couple of diodes might introduce noise into the AC heaters but without trying I don't know. 
If you aren't using cathode followers, running the heaters from negative DC voltage would be OK so, there's always that approach

 

[5v333]

Ok maybe i will try hooking up some negative supply from my lab generator and see how it works.

I also have a 5v supply for relays and power indicator lamp.
Maybe i could run thoose on -5v and use it for ccs aswell...

Good idea about negative dc heater supply but i have to think about it.

 

[Winston OBoogie]

Ok maybe i will try hooking up some negative supply from my lab generator and see how it works.

I also have a 5v supply for relays and power indicator lamp.
Maybe i could run thoose on -5v and use it for ccs aswell...

Good idea about negative dc heater supply but i have to think about it.

Yep using the lamp and relay supply would work.  Trying it with your lab supply first is a good idea. 
The IDSS of a 170 GR grade will be up to about 6mA so to dial it down to the (1.5mA in your case) design centre value with a degeneration resistor, I usually put a 1K resistor inline with the drain (although inline with the source will work too) and measure the voltage across while adjusting the current.  So, 1.5 volts across in your case and you're good to go