Speculations about Pass JFET Beast R2 disclosed in another thread.
It looks like only 1 resistor per JFET cell. That causes me to think that one JFET is the current source and the other JFET is the buffer. It is a single-ended buffer.
My guess is that all JFETs are J113.
Wild guess is 7mA idle current per buffer cell.
If I counted correctly, there are 560 JFETs on each PCB.
My guess is around 22 watts per channel just before clipping of the current source.
I am thinking of similar sized build for a project but the buffer cell will be the J113 Cirlcotron from EUVL. I constructed one such cell and it works great and has 2x current capability of idle. The downside is many more resistors than a single ended cell and 2 extra power supplies per channel. I like to build something that is not a clone.
It looks like only 1 resistor per JFET cell. That causes me to think that one JFET is the current source and the other JFET is the buffer. It is a single-ended buffer.
My guess is that all JFETs are J113.
Wild guess is 7mA idle current per buffer cell.
If I counted correctly, there are 560 JFETs on each PCB.
My guess is around 22 watts per channel just before clipping of the current source.
I am thinking of similar sized build for a project but the buffer cell will be the J113 Cirlcotron from EUVL. I constructed one such cell and it works great and has 2x current capability of idle. The downside is many more resistors than a single ended cell and 2 extra power supplies per channel. I like to build something that is not a clone.
The late diyAudio member Scott Wurcer is quoted that you can parallel as many as 4 jfets with one gate resistor before parasitic oscillation becomes an issue. Similarly with device matching you can parallel with one Source resistor.
Oh. Wow. I did not know that. Thank you for increasing my knowledge. I will give this a try on a proto board.
I neglected to consider the gate resistors and only considered the source resistors.
I did notice that there are two colors of resistors. Now I see 1/2 the resistors are source resistors an 1/2 the resistors are gate resistors.
If the buffer cells are single ended, my guess remains at around 22 watts / channel assuming 7mA idle current per buffer cell.
Thank you for pointing out something new consider. This could decrease the size of some of the PCBs that I am working on.
I neglected to consider the gate resistors and only considered the source resistors.
I did notice that there are two colors of resistors. Now I see 1/2 the resistors are source resistors an 1/2 the resistors are gate resistors.
If the buffer cells are single ended, my guess remains at around 22 watts / channel assuming 7mA idle current per buffer cell.
Thank you for pointing out something new consider. This could decrease the size of some of the PCBs that I am working on.
After re-visiting the photo, it looks like 448 JFETs / PCB.
Revised watts/channel estimate now 17.8 watts/channel assuming 7mA / JFET pair.
Revised watts/channel estimate now 17.8 watts/channel assuming 7mA / JFET pair.
For the "reduced resistor trick" you have to match quads or octets of JFETs. How are you going to do that with 1000 surface mount parts? For SMT, I would have a PCB house populate the JFETs and resistors. The devices are on reels or in tubes.
For my BEAST endeavors, I am characterizing and labeling every JFET of 1000+ pcs of J113 and 1000+ pcs of J176. It is months of part time work. All of them are through-hole parts. I have 2 test stands. One for J113 and one for J176.
The hope is to be ready to populate PCBs this year.
I get a different numerical result
- 560/2 = 280 cells
- Each cell is a source follower with a constant current source load
- 0.007 amperes DC per constant current source (assumed!)
- 1.96 amperes constant current for the array of 280 parallel cells
- If peak output current is 1.96 amps, sine wave RMS output current is at most (1.96 / sqrt(2)) = 1.386 amperes
- RMS power into 8 ohm load is (1.386 * 1.386 * 8) = 15.37 watts
I recounted the votes. Its 448 JFETs per pcb. There are 2 PCBs per channel. Somebody double check my JFET count from the picture.
I also just found that I entered power formula incorrectly.
I also just found that I entered power formula incorrectly.
I revised my calculations.
Assuming + / - 13.8v rails
224 buffer cells per pcb
At 8 mA idle per buffer cell, total pcb idle current is 1.79 amps
2 pcbs in parallel yields 3.58 amps idle
13.8 v / 4 ohms yields 3.45 amps pk
A pair of these Beast R2 pcbs in parallel will just clip at 27.6v p-p which is 9.79 v rms.
9.79^2 / 4 ohms =~ 24 watts RMS per channel
12 watts RMS into 8 ohms
This assumes zero output impedance. Realistic wattage will be some lower.
Higher idle current will allow full voltage into lower than 4 ohms. If you raise the idle current high enough, you can operate the pcbs in balanced/bridged mode and drive 8 ohm loads. With enough idle current, in balanced/bridged mode, output power would be ~ 48 watts/ch
Assuming + / - 13.8v rails
224 buffer cells per pcb
At 8 mA idle per buffer cell, total pcb idle current is 1.79 amps
2 pcbs in parallel yields 3.58 amps idle
13.8 v / 4 ohms yields 3.45 amps pk
A pair of these Beast R2 pcbs in parallel will just clip at 27.6v p-p which is 9.79 v rms.
9.79^2 / 4 ohms =~ 24 watts RMS per channel
12 watts RMS into 8 ohms
This assumes zero output impedance. Realistic wattage will be some lower.
Higher idle current will allow full voltage into lower than 4 ohms. If you raise the idle current high enough, you can operate the pcbs in balanced/bridged mode and drive 8 ohm loads. With enough idle current, in balanced/bridged mode, output power would be ~ 48 watts/ch
I like the idea of a circlotron Beast with J113 because you can achieve ~2x idle current into a speaker load without requiring matching N and P JFETs. J113 JFETs are ~ USD $0.10 each in large quantity
The circlotron has more resistors and an extra power supply to set idle current. That is the downside
The circlotron has more resistors and an extra power supply to set idle current. That is the downside
My guess is that this is the basic cell. It is a B1 where the JFETs are parallel doubled. There are 4 of these cells per row of 16 JFETs. 4 gate resistors per row of 16. 4 source resistors per row of 16.
It is a clever design to cut the number of resistors by 50%
I will give this a try with the J113 circlotron.
The value of the source resistor depends on how much Idle current is desired.
It is a clever design to cut the number of resistors by 50%
I will give this a try with the J113 circlotron.
The value of the source resistor depends on how much Idle current is desired.