First Post & Question

[T. B.]

Hi everyone.  I've discovered this forum and LOVE it.  Even for someone with an engineering degree and experience I find pouring through these forums have taught me many, many things I wouldn't have learned otherwise.

Now for my question.  In many schematic, particularly old Boss types where JFET switches are used, there is often a diode on the gate of the JFET.  Here is the link to an excellent article by R.G. Keen and a quote from it: http://www.geofex.com/Article_Folders/bosstech.pdf


"With the diode as shown in series with the JFET gate, Vswitch *can* go higher, as high as 9V, and the diode prevents this from changing the
switching operation of the JFET by not letting the higher voltage upset the gate-source voltage on
the JFET."

Can someone explain what the "changing of the switching operation of the JFET" means in laymen terms? 

Thanks,
T.B.

P.S.  I wanted to attach a picture but had no luck - sorry.

[amptramp]

The gate is left floating where leakage currents tend to pull the gate up to the signal potential and turn the device on.  If the diode was not there, the pullup of the collector driving it through R_slow would pull the gate positive to 9 volts and forward-bias the gate to channel junction.  The diode drop is still low enough that when you want to turn the device off, the gate goes sufficiently negative.

You want the channel to conduct when the gate is allowed to float up but you do not want to pull any more than leakage current through the gate to channel junction or the current will be added to the signal current.

[PRR]

Welcome.

Yes, it looks "wrong" because "we know" a backward diode does not conduct. So what is the voltage between two back-to-back diodes? On the usual assumption, this point is "disconnected from the universe" and we can't say what the voltage would be.

In fact, real diodes leak. Not much. But between a '914 and a JFET we can reliably expect the JFET to go to a "zero bias" point where it is "ON", which is what we want.

> "not letting the higher voltage upset the gate-source voltage"

That's the key. The driver circuit goes nearly to +9V, but the JFET's two ends are idling near 4.5V. If we pull the Gate to near +9V, the Gate diode conducts, pulls the two ends and the audio path to some voltage higher than +4.5V, POP.

With the added Gate diode, the Gate can't be forced positive of 4.5V with any more than the super-low leakage of the diode, which typically does not cause any problem.


___________________________

> I wanted to attach a picture but had no luck

The picture has to already posted on a public web server. This forum does not have a built-in Attachment feature. I have been using PostImg.org which is not crapped-up with ads and pop-overs.

[T. B.]

@PRR that makes a lot of sense.  Thank you for the explanation.

I take it if a MOSFET were used this wouldn't be necessary as there isn't the potential for an ohmic path from gate to source/drain.

Why are JFETs used instead of MOSFETs for switching?  Or is it just a design choice or is it the sensitivity of blowing the gate on a MOSFET with an overvoltage?

[PRR]

The main reason is that JFETs have been common longer than MOSFETs. And have often been cheaper. The dominance of MOSFETs in many fields is mostly the last few decades, while much pedal-work is stuck in the 1970s.

Audio switching with MOSFETs runs into the MOS's usual parasitic D-S diode. This tends to lead to needing two $1 MOS instead of one $0.50 JFET, math that any pedal wage-slave can figure. An alternate is CMOS chips like CD4066 which implement 2 MOS and a driver. But these switch devices are so small that (historically) the on-resistance variation was non-negligible in 0.0x% THD audio work.

And while there are pages of BIG size MOSFETs, we want quite small ones for low parasitic capacitance (but not as small as CD CMOS devices). This is more specialized field, and part numbers come and go a lot. Whereas there are not *that* many basic JFETs and they stay in production longer, and with more 2nd/3rd-sources to keep prices down.

Sure, MOSFET if you want. There's even some crazy-good parts now. You don't care if they stay in production "forever", you can lifetime-buy for a minimum order. (BOSS etc don't want to keep changing parts and designs or buy 30+years of parts up front.) But "to avoid a diode" is not a reason. This scheme is not just BOSS, it is a majority of the large mix-consoles over a couple good decades.