The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Semiconductor Devices >> JFET vs. MOSFET vs. BJT
https://designers-guide.org/forum/YaBB.pl?num=1162476437

Message started by vivkr on Nov 2^nd, 2006, 6:07am

Title: JFET vs. MOSFET vs. BJT
Post by vivkr on Nov 2^nd, 2006, 6:07am

Hi,

I was wondering why the JFET which found some usage in ICs until the 70s (used in low-noise opamp front-ends) has all but disappeared, and was trying to chalk up the pros and cons of the 3 principal devices that can be fabricated and I have the following chart:

Property BJT MOS JFET

Gm/I Best Worst Moderate
Conduction Bulk Surface Bulk
Speed High Medium Low
Noise Moderate Worst Best
Good switch No Yes Yes
High-Z gate No Yes Yes
Scaling Not much Best Not much
ESD less sensitive Most Less

From the above, I see that one could potentially use a JFET for sampled-data systems which are
the backbone of modern analog. The JFET does not scale well like the BJT and so is at a commercial
disadvantage, which is one of the only advantages that MOS has along with High-Z gate (also in JFET)
and offset-free switching. Also, since JFETs are bulk conduction devices, they ought to suffer less from
1/f noise due to surface states.

Are there any other factors that are relevant? I am thinking in terms of an analog process and not a digital
one. I think fabrication of JFETs is not particularly harder than that of MOSFETs, but I may be wrong as I
do not really know if the simplest CMOS circuits (inverters) are realizable with an equal ease with JFETs
and if they provide similar degree of performance (speed/average power).

Regards
Vivek

Title: Re: JFET vs. MOSFET vs. BJT
Post by Geoffrey_Coram on Nov 2^nd, 2006, 12:38pm

My company (Analog Devices) makes a number of low-noise, low input-current precision amplifiers with JFET input stages...

Title: Re: JFET vs. MOSFET vs. BJT
Post by vivkr on Nov 2^nd, 2006, 11:08pm

Hi Geoffrey,

I am aware that ADI does make some JFET-input amps. Also, some other companies make low-noise opamps with JFET input stages. But why do we (almost) never hear of the humble JFET in the technical literature. It is only BiCMOS or CMOS? Perhaps you or someone else could tell me more. From what I see, it seems that the JFET might outperform CMOS for some analog applications and be useful for sampled-data systems.

I am looking at ways to compare analog processes from different foundries, and I am wondering if adding a JFET to an analog process is not worth it, and if so, then what are the limitations that one must bear in mind.

Thanks for your feedback.

Regards
Vivek

Title: Re: JFET vs. MOSFET vs. BJT
Post by Geoffrey_Coram on Nov 3^rd, 2006, 3:26am

Our JFETs are (all?) on internal bipolar processes, not on bulk/fine-line cmos. But I don't to process development, so I don't know how easy or hard it would be to add a JFET to a CMOS process. Maybe you need extra masks because you can't just use the existing implants to make a good jfet. I wouldn't think there'd be any thermal cycling or such issues, where something you need for a jfet would degrade the mosfets.

Title: Re: JFET vs. MOSFET vs. BJT
Post by Croaker on Nov 21^st, 2006, 8:18pm

I'd say MOS is used as often as possible because it's cheap and refined thanks to digital IC volumes. Also, analog circuits often have to co-exist with the digital ones.

Title: Re: JFET vs. MOSFET vs. BJT
Post by loose-electron on Nov 27^th, 2006, 12:47pm

I got curious on this one and had to go ask a friend --- His reply --

I "think" it is because JFETs don't scale very well. The parasitic capacitance of the gate junction is prohibitive at small geometries. So, below about one micron, their performance is really poor. And then chopper stabilization techniques meant you could have good 1/f noise without them using MOSFETs

I think in a few years no one will know what a JFET is anymore.

Title: Re: JFET vs. MOSFET vs. BJT
Post by Croaker on Nov 27^th, 2006, 2:36pm

loose-electron wrote on Nov 27^th, 2006, 12:47pm:

I have the 3rd ed. of Gray & Meyer and JFETs are covered as much or more than MOSFETs. The 4th ed. doesn't mention them (at all, I think).

I won't shed any tears over JFETs. The emphasis on bipolar is down as well.

Weren't JFETs losing popularity before 1u? Were they ever used for digital circuits?

Title: Re: JFET vs. MOSFET vs. BJT
Post by carlgrace on Jan 3^rd, 2007, 12:28pm

The output stage of a CCD imager is functionally equivalent to a JFET, so I don't see them vanishing too soon.

Title: Re: JFET vs. MOSFET vs. BJT
Post by didac on Jan 4^th, 2007, 12:58am

Hi to all,

Years ago I asked the same question to my advisor here in the research group at University(we only use CMOS processes and from time to time BJT). He's answer was that technology follow market needs. Customer want low-cost IC's for consumer electronics(most of them digital) so the foundries refines their digital CMOS processes and after they got a good process they supply analog processes based in that tech. Also it's cheaper to try to do all the cicuits in the same process so you can easily integrate the full system in the same Die. for example we are currently working in UWB at 8GHz and we are using CMOS although it will be easier to design using a SiGe process but we need to integrate the digital part after the RF-Front end.

Title: Re: JFET vs. MOSFET vs. BJT
Post by Geoffrey_Coram on Jan 4^th, 2007, 5:35am

vivkr wrote on Nov 2^nd, 2006, 11:08pm:

I am aware that ADI does make some JFET-input amps. Also, some other companies make low-noise opamps with JFET input stages.

I think that low-noise is key. I would guess that JFETs have lower noise because they conduct through bulk silicon and don't have the Si/SiO2 interface that creates traps that affect the conduction through the mosfet's channel.

Chopper stabilization techniques (and/or correlated double-sampling) may add too much complexity or affect the op-amp's transfer function in the band of interest. There are still a lot of applications where you really just want a simple op-amp.

Title: Re: JFET vs. MOSFET vs. BJT
Post by carlgrace on Jan 4^th, 2007, 6:26am

Quote:

I would guess that JFETs have lower noise because they conduct through bulk silicon and don't have the Si/SiO2 interface that creates traps that affect the conduction through the mosfet's channel.

This is correct Geoffrey. In fact, JFETs have such low 1/f noise it can be difficult to measure without specialized equipment. The dominant noise source in JFETs is charge fluctuation in the R-G centers in the JFET gate junction. This noise level is much lower than a typical MOSFET and about on par with a BJT. Another key difference between the JFET and the MOSFET is the JFET has a larger g_m, which reduces the noise impact of the op amp's second stage (and perhaps first stage active load). So if you want a low noise op amp to be used in an application with a high source impedance, the JFET is the device of choice.