Hi,
I know it's been a few months since the last post, but it sounds like there are some experts here who are familiar with TI JFETs, so I'm asking for your advice.
Does anyone have experience with crosstalk in the JFE2140? I'm happy with the performance and I'm using it in the input stage of an older amplifier? Now I'm starting a new design and need well matched JFETs but for different channels. The pinout is great to use one side per channel, but I'm a little worried about crosstalk. Unfortunately, nothing is specified in the datasheet. Does anybody knows more?
Thanks for any input or estimation.
I know it's been a few months since the last post, but it sounds like there are some experts here who are familiar with TI JFETs, so I'm asking for your advice.
Does anyone have experience with crosstalk in the JFE2140? I'm happy with the performance and I'm using it in the input stage of an older amplifier? Now I'm starting a new design and need well matched JFETs but for different channels. The pinout is great to use one side per channel, but I'm a little worried about crosstalk. Unfortunately, nothing is specified in the datasheet. Does anybody knows more?
Thanks for any input or estimation.
The datasheet doesn't say much about it but my gut feeling says that you should not use this device for two channels. It seems that the main application is diff input stage or a cascode stage.
Thanks @peranders, you are right. Using different devices is safer, but good matching is a big advantage when playing around with stereo amplifiers without feedback. I'll think about it and do some tests. Has anyone done practical tests?
I had designed a phono stage with 2140 since 150s was a disaster. I strongly recommend to use one 2140 per channel.
If you need 30 dB or around it in channel separation then I think you can use this device for that. I mean the cartridge has 20-30 channel separation to begin with.
150 are too small, so i had problems soldering and problems with heat. 2140s are much more stable and really bigger. So I decided to use one 2140 per channel and everything stabilized fine.
The outcome of the phono was good but not top (maybe it was my design, it was based on Salas' circuit). It had hard competition against ECC88s however
Why not use one per channel? They are not expensive, unless you have terrible problem with space
I used one device per channel (so one complete package in each channel) here
https://hifisonix.com/projects/x-altra-phono-eq-preamp/
The JFE2140 is a superb JFET.
https://hifisonix.com/projects/x-altra-phono-eq-preamp/
The JFE2140 is a superb JFET.
It's an input stage with high gain and included RIAA LP filter (1st stage). I like to work this time without feedback and am afraid that with separated JFETs without matching, the gain of both channels has a huge difference. That's the only reason. They are working in a cascode so I will save no JFET if I use a separated design. But matching may be not fitting.
Last edited:
Just thinking out loud. There are dual OPAs (like NE5532) in the same package that are specified with 110dB of crosstalk attenuation. On some older pictures it looked like both OPs were placed on the same piece of silicon.
JFE2140 is a matched pair and without trimming options (???) they must be placed on a separate piece of silicon.
This leads me to believe that crosstalk can be negligibly small even when components are placed in the same package. How does that answer my question... actually it doesn't? I'm waiting for feedback from TI and/or need to test it, but there's a chance it's not that bad, right?
JFE2140 is a matched pair and without trimming options (???) they must be placed on a separate piece of silicon.
This leads me to believe that crosstalk can be negligibly small even when components are placed in the same package. How does that answer my question... actually it doesn't? I'm waiting for feedback from TI and/or need to test it, but there's a chance it's not that bad, right?
I would think unless your gate impedances aren't brutally large there won't be any capacitive coupling to worry about from fast-moving pin potentials on the other FET, assuming the clamping diodes are connected to low impedance points.
Thermal coupling, though, is unavoidable. If you have varying dissipation in one FET, it will have an influence on its own threshold voltage and transconductance but also will heat up the other FET and change the parameters, which is normally what you want in a LTP or other differential arrangement.
I am not sure about this, so am open to correction.
The JFETs in the JFE2140 share the same substrate so you have to be careful about running each device in completely different circuits. In other words, imagine two of these dude by side, but the drains are at very different voltages and so are the sources.
For my money, if your circuit only requires 1 JFET, you should use a single, or put the two devices on parallel (you get a 1/rt2 noise reduction like this).
The offsets on these devices and the LSK389/489 are surprisingly good. I used them in a line stage design (https://hifisonix.com/projects/x-altra-mini-ii-discrete-line-stage/) and the output offset of the amp with offset pot set to mid point was never more than about 10 mV, so they are well matched.
Fantastic parts!
The JFETs in the JFE2140 share the same substrate so you have to be careful about running each device in completely different circuits. In other words, imagine two of these dude by side, but the drains are at very different voltages and so are the sources.
For my money, if your circuit only requires 1 JFET, you should use a single, or put the two devices on parallel (you get a 1/rt2 noise reduction like this).
The offsets on these devices and the LSK389/489 are surprisingly good. I used them in a line stage design (https://hifisonix.com/projects/x-altra-mini-ii-discrete-line-stage/) and the output offset of the amp with offset pot set to mid point was never more than about 10 mV, so they are well matched.
Fantastic parts!
Thanks for you feedback.
Good point. In my last design I don't use the clamping diodes because I was afraid that they generate additional noise. As it's a phono amplifier and the levels are pretty small, there is no critical voltage and ESD seems to be no issue with FETs. I understand your point, that it’s important to connect die diodes to the supply voltage because of cross-coupling? I assume you mean that an impedance at the clamp will generate a voltage drop in case of a current and that couples to the other FETs gate capacitively? Or is there another reason reducing cross-talk by using the clamping diodes?
Thanks for your thoughts and if so, you are absolutely right. But how is matching done if they are on the same substrate?
I think TI disclosed the 150 is the exact same die as the 2140, just pre-bonded for devices in parallel. The catch diode pins are a giveaway for that, too.
Interesting note from Interfet about crosstalk: https://www.interfet.com/jfet-application-notes/thermal-soic-8-package-jfets-at-interfet/
There actually could be some interference. Let's wait what TI has to say.
Interesting note from Interfet about crosstalk: https://www.interfet.com/jfet-application-notes/thermal-soic-8-package-jfets-at-interfet/
There actually could be some interference. Let's wait what TI has to say.