F-5 with Toshiba outputs, R3 and R4 values

[Russellc]

I have noticed on this forum and a few others that F-5 builders using the Toshiba mosfet output devices SOMETIMES need to alter R3 and R4 from 2.2K
to something less around 1.5K or so for the bias to have sufficient "adjustibility", while others seem to do fine with the standard 2.2K R3 and R4. Is there something in the specs of the j-fets or mosfets that causes this? I'm asking because I have matched J-fets from high 6 to low 9
(BL grade) and also have a few different sets of the outputs and was wondering if certain specs in the parts caused this, or just build to build variations? I ask as I am getting ready to do an F-5 with the Toshiba outputs, am using Caddock resistors and have already stuffer the board (peter Daniels) with the 2.2K and dont relish trying to redo

Russellc

[6L6]

If I remember my discussions with Blues, the lower value of R3 R4 is an attempt to keep the pots in the middle of their range.

Zen Mod said that if you run out of adjustment on the pots it can be partly the input jfets being on one side of their tolerance, or the output mosfets being (again) on one side of their tolerances, or (more probably) a mix of both.

Regardless, changing the series resistor is not that hard, and although I understand that you want to keep all your resistors Caddocks, I would bet that the primo resistors, in that position, don't make much difference... Remember, you are mostly listening to the pot, the bypass probably doesn't make that much change.

[flg]

The reason is the Toshiba deivces have a lower Vth G-S. The gate threshold voltage on an IRF or similar part is about 4V or so to bias the output on. A Toshiba device G-S threshold is more like 2 volts.
When the JFET current passes throught the 2.2k || pot, the resultant voltage sets up the output bias of 4V for an IRF output stage. This voltage must be about 2V for toshiba parts. 1/2 the resistance The pot probably goes down low enough but I would not want all my current going thru a pot.
Using a smaller drain resister for the JFETs also reduces the first stage gain somewhat...

[Russellc]

Originally Posted by 6L6

If I remember my discussions with Blues, the lower value of R3 R4 is an attempt to keep the pots in the middle of their range.

Zen Mod said that if you run out of adjustment on the pots it can be partly the input jfets being on one side of their tolerance, or the output mosfets being (again) on one side of their tolerances, or (more probably) a mix of both.

Regardless, changing the series resistor is not that hard, and although I understand that you want to keep all your resistors Caddocks, I would bet that the primo resistors, in that position, don't make much difference... Remember, you are mostly listening to the pot, the bypass probably doesn't make that much change.

I used the resistors because they fit well, and I understand Peter designed the boards with them in mind. The change would be simple, its just that the boards dont take kindly to it many times...I do have extra boards.

If I can find out what characteristics of the fets and mosfets are that cause this, I may have enough with different numbers that it could be avoided.

My mosfets as well as J-fets are all matched.

Russellc

[flg]

Because they fit well? Like 501's? Or maybe Latex?
The IR datasheets call the parameter Vgs(th) but it's a useless number from the datasheet (2-5V). I don't know what toshiba calls it but Toshiba actually has 2 grades. Off the top of my head B-Y? 0.8Vth and 1.8Vth? Naturally you want the higher threshold part...
As far as pots. I have a 620 ohm in the place of the 2.2k. Hopping to get very little current and audible effects from the 5k pot. I would measure the finished combination and use your favorite botique resister there.

[Blues]

Originally Posted by 6L6

If I remember my discussions with Blues, ...

Attached is an illustration I shared with a member of how the values can be derived for practical purposes...with these values, power on P2 (and P1) is very low at around 12mW and shouldn't be a concern. The 2.2k resistor will dissipate around 4mW.

Hopefully this helps...

[flg]

My desire to avoid the pot (did I say that) is not a power related function. I have 2 basic reasons:
1) When a pot goes bad, what type of sound does it make? It also almost always causes a rediculous noise but, it goes open. Causing the bias to increase big time. A smaller pot does have less current/power in it but, any trouble it causes is going to be less if it is a larger size.
2)Go read George's description of why he created the lightspeed attenuator. if you were to use a boutique resistor in this spot. You would be wasting your time with it next to a pot. You can buy metal foil pots from Vishay...
I just don't like'em. I avoid them if possible. I try to put them in a position where if they go bad they do not cause catastrophy.

[prooptiki]

I've seen also (with Toshiba mosfet) changing the source R12 resistor from 0.47R to 0.22R, the reason is?
Thanks

[Zen Mod]

lesser xconductance

[flg]

Source Rs are a little touchy subject. The idea is, you are adding a Source resister similar to the source resistance inside the device. The res in the device actually changes with differing devices conditions causing the non-linearity you despise. The one you are adding "swamps out", we say, the varyability of the internal source res. Basically you are adding to the total source R with a fixed resistor to reduce the varyability generated inside. Although I describe it as source resistance, I beleive designer types think in terms of a voltage across the external source R. In that case when you have higher source current you use a smaller source resister to generate the same voltage. It is likely that the actuall internal source R of the Toshiba's is lower than the IR types. It is also likely that the Xconductance, as ZM said, is lower and a smaller Rs corrects for that somewhat. Also the bias current in some designs is higher than others. N.P. seems to stick with a suitable and safe value like .47.
Well, I tried