From ???@0x00001401 Tue Sep 18 13:17:14 2001 Path: pitt.edu!newsflash.concordia.ca!sunqbc.risq.qc.ca!newsfeed.direct.ca!look.ca!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail From: Walt Jung <_walter.jung@analog.com> Newsgroups: rec.audio.tech,sci.electronics.design Subject: Re: Importance of impedance matching for Jfet op-amps? Message-ID: References: <9o4bvt$ov6$1@usenet.Stanford.EDU> <9o5tlr$k35$04$1@news.t-online.com> <3BA68832.928C18E3@rowland.org> X-Newsreader: Forte Agent 1.8/32.548 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 90 Date: Tue, 18 Sep 2001 14:45:24 GMT NNTP-Posting-Host: 24.23.50.141 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.md.home.com 1000824324 24.23.50.141 (Tue, 18 Sep 2001 07:45:24 PDT) NNTP-Posting-Date: Tue, 18 Sep 2001 07:45:24 PDT Organization: Excite@Home - The Leader in Broadband http://home.com/faster Xref: pitt.edu rec.audio.tech:151558 sci.electronics.design:258297 Status: N On Mon, 17 Sep 2001 19:33:06 -0400, Winfield Hill wrote: >> "alan" wrote: >> >>> I am building a headphone amp based on the Burr-Brown OPA134 >>> op-amp. In the data sheet, they say that you should match >>> the input source resistances, but don't talk about what >>> magnitude of errors you will create if you don't. I read >>> some more about this effect in this article by Walt Jung: >>> >>> http://www.planetee.com/planetee/servlet/DisplayDocument?ArticleID=1518 > >Helmut Sennewald wrote: >> >> ... the idea from Walt is to build a lowest distortion >> amplifier. Lowest distortion means -100dB distortion. >> >> The theory behind from Walt: >> ---------------------------- >> The gate-drain capacitance of the JFET in the OP input stage >> is voltage dependant. That can be proved in any JFET databook. >> The source resistance and this nonlinear gate drain capacitance >> forms a lowpass and of course leads to distortion ... > > Circuit solutions can be easily implemented that maintain > a constant Vdg for the JFET, thereby completely eliminating > the problem. > It could be that doing a cascode and/or bootstrap of a conventional IC JFET diff-pair input could completely *miss* this distortion mechanism. The offending culprit is the JFET gate's junction capacitance *to the substrate*. See further comment on this point, below. > We discussed one such solution in detail in AoE, 1st edition, > page 250. The circuit, a bootstrapped high-impedance JFET > amplifier, survived into the 2nd edition as a "good circuit" > without any description, page 172, fig. 3.77 - A. > > In the case of opamps, there are many JFET designs, such > as the Burr-Brown OPA111 or the Analog Devices AD549, which > employ internal cascode stages serving to keep Vdg constant. > The newly-popular OPA627 has similar circuitry. Using one of > these opamps would surely be a better solution than matching > impedances, as Walt suggested in his article. Given that his > employer, Analog Devices, makes some of the desired higher- > performance JFET opamps, I'm surprised he didn't suggest it! > Thanks for the thought Win. But, an audio designer type likely wouldn't use the OPA111 and/or AD549 (or related parts like the AD7975) because of speed and of course, the relatively high price. To comment further on bootstrapping, I did write about it, in ADI AN232, "Bootstrapped IC Substrate Lowers Distortion in JFET Op Amps". This note shows a method of dynamically driving the -Vs rail of an AD744 op amp, to negate the effects of the delta-C/V (Fig. 1 of this note). A summary quote from this note: ".. It should be noted that the general principles of the non-linear C/V input characteristic distortion mechanism applies to virtually all JFET input IC op amps available today, independent of their source. Bipolar input op amps built on junction isolated processes can also be subject to the phenomenon, and may also benefit from bootstrapping. However they are not as likely to be used with large source impedances where the distortion magnitudes become a serious problem, as is true in the case of JFET op amps." Helmet Sennewald said: >The problem: >------------ >If you have to connect different sources you can never match >the source impedance. >In this case you could use circuit tricks to keep the >gate-drain voltage consThe problem: ------------ A middle-ground approach (when working from a volume control, say) is to match the feedback network impedance to the normal- listening-level impedance of the control. If the control is low in Z (<50k), the Zsource can still be low, ~2-3k. But, speaking more generally, yes, I agree that some sort of bootstrapping is the only way to preserve low distortion. Thanks to those making these helpful comments. Walt Jung