It's necessary to get these parts from ARC, they are all hand selected for the exact circuit purpose.
They always keep parts in stock.
They always keep parts in stock.
The problem with FET is they have variable pinch off voltages. So the 6525 will have to be matched. By the time you got 20 or so j112/j113 and matched them you might as well have got them from the supplier.
I would get a matched quad of the LSK170’s they sell out fast, load resistors may need some adjustment, you may find some authentic Toshiba 2SK170’s. These are some great JFETS, have used them a number of times!
Another problem, is when you use a large dropping resistor on a low voltage device running off a high B+, the load resistance is crucial, I would recommend using a smaller dropping resistor, and a high wattage Zener and decoupling cap to arrive at a more reasonable load voltage for those JFET,s! It’s like playing Russian Roulette, the load resistor has to be perfect and right on, for the FET’s to operate properly.
Last edited:
And this is why those need replacement. It is truly weird they didn't even introduce at least an RC decoupling network for the input. Were this a diy project it would have received lots of criticism for this.
Arc is known for some really unorthodox practices! It’s almost like they welcome
Something to go wrong. I’ve been close to buying ARC gear in the past, but I’m glad I did the research and examined many of their schematics.
The PS already has several mods, including outboard Lambda tube regulator (with mods to lower noise) and LC filters for each input stage which then feed RCR filters (load R split in half with matched R and Muse/Ploypro cap). Really didn't want to add zener noise. This was a major improvement over the ARC PS.
Will 2SK170 handle the current needed for 4 fets in the difamp? I was thinking maybe 2SK369? I do have some DN2540 I can use for the tube stages.
Thanks for the recommendations.
Will 2SK170 handle the current needed for 4 fets in the difamp? I was thinking maybe 2SK369? I do have some DN2540 I can use for the tube stages.
Thanks for the recommendations.
According to the indicated voltages this current should be around 14mA. So, yes, a selected violet device will do.
Fortunately there's quite a lot of help on the schematic. So there is 200v across 50k drain so that is 4ma so 2ma in each FET. So you want a device with 2ma at VGS of approx -.19v so the 2SK170 looks OK. You will still have to match all four FETS for VGS though.
So providing RV1 is adjusted for 8ma through Q6 the VGS will find its own level.
You may only need two FETs if you can find suitable devices.
I think I would look at it in simulation to get the best fets, but there are plenty of other ways.
So providing RV1 is adjusted for 8ma through Q6 the VGS will find its own level.
You may only need two FETs if you can find suitable devices.
I think I would look at it in simulation to get the best fets, but there are plenty of other ways.
Last edited:
OK its not quite that simple. The FET must not end up having a voltage on the source of less than -300mV or the gate source will conduct and the circuit goes non linear.
Tried with 2SK170 in LT spice gives +300mV on source and works fine.
Hope this helps.
Tried with 2SK170 in LT spice gives +300mV on source and works fine.
Hope this helps.
Attachments
Last edited:
Thank you very much for the LT spice analysis. Seems I'll go with 2SK170 for the cc.
Anyone have a clue what Q5 is...or could be?
Anyone have a clue what Q5 is...or could be?
For some reason the sim does not provide separate source resistors and this is the reason the bias voltage differs from the 190mV shown in the circuit.
No. There is also current passing through the 10k which reduces the drain voltage.
Sorry copying error - does not affect the voltage on source as the current through each leg is still the same but does increase the gain.
Yep sorry 5.5ma through tail gives 12v on drain.
Q5 not critical as long as vgs is low so output does not clip at low voltage and vgs does not forward conduct opposite of 2SK170 may do.
Last edited:
I'll give the 2N5460 a try. I don't have any sim software, but from the data sheets, looks like the 2SJ74 (compliment of 2SK170) could work? A lot quieter.
All the help has been awesome!
All the help has been awesome!
Yep should do - there's plenty of gain at the front so the noise figure is not so important for the P-FET.
The 2SK170 will always work as the Drain current from spec:
IDSS(Note)VDS= 10 V, VGS= 0 2.6 ⎯20mA
Min value of 2.6ma > 5.5ma tail current / 4 = 1.375ma. Which means VGS will be negative. You will need 4 accurately matched ones.
The 2SK170 will always work as the Drain current from spec:
IDSS(Note)VDS= 10 V, VGS= 0 2.6 ⎯20mA
Min value of 2.6ma > 5.5ma tail current / 4 = 1.375ma. Which means VGS will be negative. You will need 4 accurately matched ones.
The biasing is still wrong. The CCS should be connected to a negative voltage instead of ground.
Does not matter in simulation as the CCS can generate energy. In practise you are correct and if replaced with MOSFET that would necessary in simulation.