According to Borbely, zero tempco occurs at
Vgs = Vp + 0.63V
https://audioxpress.com/article/JFETs-The-New-Frontier-Part-1
Datasheet of JFE2140 specifies Vp as -0.9V ~ -1.5V.
Take the case of -0.9V. Vgs at zero tempco is -0.27V.
Fig. 6.1 of datasheet then implies zero tempco bias to be around 8mA.
Patrick
Vgs = Vp + 0.63V
https://audioxpress.com/article/JFETs-The-New-Frontier-Part-1
Datasheet of JFE2140 specifies Vp as -0.9V ~ -1.5V.
Take the case of -0.9V. Vgs at zero tempco is -0.27V.
Fig. 6.1 of datasheet then implies zero tempco bias to be around 8mA.
Patrick
Yep, 8mA even better. I just did a quick estimation from data sheet curves, 8mA = even less distortion.
These JFets need to run at decent currents, they will sound better and measure better. 🙂
TCD
These JFets need to run at decent currents, they will sound better and measure better. 🙂
TCD
I did a recent search for replacement N-channel JFETs used as switches in some vintage audio gear. The BEST I could find was a J109-D26Z with an Rds-on of 12Ω. In this application, N-channel enhancement MOSFETS (which require a + gate voltage to be on) will NOT work. Now I want to find a P-channel JFET switch with a similar low Rds-on. Any suggestions?
Depends. You want JFET or you want low on-resistance? It's easy to get below 10milliohms in switches.
At least in enhancement mode.
Jan
At least in enhancement mode.
Jan
Most of the Pro Audio products I've been across used J175 (equiv to P1086 Interfet) and I believe even the potted Otari switching modules used them. Nowhere near your 12 ohms.
Which vintage gear?
Are JFE150 and JFE2140 on the same process?
JFE150 has twice the Ciss, but its noise voltage advantage does not fully scale?
Cheers, Gerhard
JFE150 has twice the Ciss, but its noise voltage advantage does not fully scale?
Cheers, Gerhard
Yep. Exact same process. JFE150 has both JFETs in the 2140 connected in parallel. Not sure why there is a diminishing returns on the noise front, it could be a measurement issue getting far below 1nV/rtHz but that's just speculation on my part.
Connecting "N" identical BJTs in parallel, the equivalent noise voltage is reduced by a factor of N^x (N raised to the power x). For bipolars, it turns out that x=(1/2). For example when N=4 identical BJTs are connected in parallel, equivalent input noise voltage is reduced by a factor of ( 4 ^ (1/2) ) = 2. Four parallel bipolars cuts equivalent input noise voltage by a factor of two.
What is the corresponding value of the exponent "x" when connecting N identical JFETs in parallel, rather than bipolars?
What is the corresponding value of the exponent "x" when connecting N identical JFETs in parallel, rather than bipolars?
I'd ask Wayne, since he probably has practical experience with the Pearl RIAA design.
Edit - looks like the same scaling factor applies for jfets as bipolars - 4 devices cuts the noise in half...
Edit - looks like the same scaling factor applies for jfets as bipolars - 4 devices cuts the noise in half...
X should still be 1/2 theoretically. However, when the JFETs occupy the same piece of silicon and share the same metalization, you introduce thermal effects and impedances (from metal lines) that are not present when you are connecting multiple discrete JFETs in parallel on a PCB.
I used the JFE2140’s in this in place of the LSK389 in the MM amplifier below and got superb results. Great JFET 👍
https://hifisonix.com/projects/x-altra-phono-eq-preamp/
https://hifisonix.com/projects/x-altra-phono-eq-preamp/
Could you simply parallel to two fets in the 2140 and have the same part in effect?
This does not really belong into this thread, it's just that I need it for an 2140
amplifier: Is there somewhere a Spice model of the THS3091 CFB opamp ???
It seems to be a happy combination of noise, 1/f corner and speed.
If it is on TI.com, then it's carefully hidden, at least in ASCII form. I'd prefer
Spice 3 format, best would be LTspice compatible, sorry, or Altium Designer,
or Keysight ADS, even if ADS would cost me bonus points with a customer. ;-)
Cheers, Gerhard
amplifier: Is there somewhere a Spice model of the THS3091 CFB opamp ???
It seems to be a happy combination of noise, 1/f corner and speed.
If it is on TI.com, then it's carefully hidden, at least in ASCII form. I'd prefer
Spice 3 format, best would be LTspice compatible, sorry, or Altium Designer,
or Keysight ADS, even if ADS would cost me bonus points with a customer. ;-)
Cheers, Gerhard
It's on the THS3091 product page at ti.com. Go down to the 'Design and Development' section and then click on the 'Design tools and Simulation' tab.
Thanks, I did find a zip archive somewhere on ti.com that
had lots of binaries and in a folder "timing simulation", there
was a netlist that included the model inline.
I snipped it off and put it into /d/libs/spice/gerhard_opa.lib,
but did not yet try it .
I think a spice model should belong into the data sheet. It is
much more important than the picture of card board box with a
reel of 5000. And it would help to keep the data sheet and
the model in sync. We all know to handle ctl-c/ctl-v in a .pdf.
cheers, Gerhard
had lots of binaries and in a folder "timing simulation", there
was a netlist that included the model inline.
I snipped it off and put it into /d/libs/spice/gerhard_opa.lib,
but did not yet try it .
I think a spice model should belong into the data sheet. It is
much more important than the picture of card board box with a
reel of 5000. And it would help to keep the data sheet and
the model in sync. We all know to handle ctl-c/ctl-v in a .pdf.
cheers, Gerhard
Noise current is proportional to the root of the conductivity. Noise voltage is inversely proportional to the root of the conductance (See Titze, Schenk "Semiconductor Circuit Engineering", Chapter 3.3.4). Accordingly, when N transistors are included in parallel, the noise voltage will decrease by the root(N).