Skip to content
Notifications
Clear all

[Sticky] USB-C Microphone (official topic)

1,384 Posts
203 Users
345 Reactions
382.6 K Views
marcdraco
(@marcdraco)
Posts: 1127
Moderator at Large and Cat's Butler
 

Diode mode is a bit of a tricky one - but the good news is - you can't fry a JFET with it. Esp a modern meter using the diode tester.

You can (potentially/certainly depending) kill a JFET by wiring the Gate where the Drain end [of the channel, more of that in a mo.] and the Drain AND put enough voltage on there AND put enough current behind that voltage... then you can kill a JFET. If you put loads of voltage beyond the rated amount on the channel (I can't claim to know what that is, but it's likely to be a good 150-200% rated) it will eventually go bang (assuming it's switched off anyway).

I hope me going a bit "The Dude" of you. The Dude is a character from a Cohen Brothers film you may not have scene. Jeff Bridges at his laid back best and... something I aspire to while leaning so far back I end up in a huge pile behind where the back of my chair used to be. Trust me, this stuff is weird and in order for me to explain it I've got to do something to get me into the mood. That means thinking about how elementary particles aren't actually particles at all and other stuff about causality that even makes clever people's brain's itch. 

See - LOADS of ways you might blow one up - most of them won't happen.

While I was researching what was supposed to be my next book (until a better idea presented itself) I did a lot of work in semiconductor physics and one thing kept cropping up.

Time.

Or specifically (and far beyond what most engineers would consider reasonable) in I think every instance I tested is the problem. 

Not the voltage - that's not real except as a useful measure of upstream power.

Not the current but also the current...

The what?

Current is a measure of the number of electrons (and those buggers are nothing like we were taught either) that cross a point in space in a second - there's some weird thing that I've done since college that we use to do with wires going off into infinity but smart people have come up with a better way now.

These little fellas come whistling through at a mighty speed 6.2 × 10¹⁸ of them per second. That's a lot. 

And this is also why resistors and (mostly) older small transistors but ultimately everything except (I think) superconductors create heat by the friction elementary charges cause and... man what a mess, right?

So time matters. In most cases we just think of current and the effect it causes like making ICs all burny, burny - and then complain to Amazon that they sent a faulty part... because Amazon does care and if it's eBay there's a good chance it didn't work to start with; some ICs are so commonly faked it became a running joke in the 1980s - yeah that far back! And the same fakes are still on sale to this day! (I picked a couple up about three years ago from different sources, checked against the spec. sheet and sure enough. Fake.

Current makes stuff hot. 

We say that 1 amp or mA or whatever is flowing. 

That's great but... so what?

Imagine having a bottle that has a ring of small holes around it about a half way up. Not many, not enough to break it, just enough to leak.

Now picture a dripping tap (I can't, but perhaps you can and this metaphor is simple).

As the water enters, the water collects and begins to fill the bottle and eventually the level rises to the point where water starts to leak away - we'll assume there's a limitless supply of water (the amount of water, the water pressure, is the voltage if you're interested).

You probably realised that the water leaks out and the level never increases beyond that point provided the drip rate remains at or below the loss rate.

If you get that just have or are just about to  wrapped your head around one of the weirdest things in engineering, the principal of entropy. You already understand integration (the the basis of integral calculus) which is how (if you stick with this) low-pass filters work! Neat eh?

Our lossy bottle is a little bit away from a better explanation.

So let's replace the bottle with a really tough piece of filter paper in the shape of a container. 

The loss is much clearer now (water appears to fall from the lowest point if you do the experiment because it slips around the surfaces under the direction of gravity and even interia,

What matters is the loss is coming from everywhere.

That's radiation and any system the amount of energy that comes in is the same amount that comes out. 

TL;DR

Now this is where it matters.

In a resistor, we deliberately make the transit of electrons harder to limit current (charge over time) but crucially we consider time like that dripping tap or the coffee filter that never overflows because the amount of loss is always greater than or equal to the amount of energy passing in a ... ready for this... our timescale.

You can imagine a second or heartbeat, maybe you can just about see the 1/10ths of second flashing past on an Olympic event... but we're approaching our limit.

Blink of an eye? An eternity,

The flash of from a non-nuclear explosion... billions of years.

You'll allow me some poetic licence here, I'd usually check these numbers I just don't have the time.

It amounts to this. You can put hundreds, thousands or even millions of volts across non-vacuum and if the pulse lasts for a sufficiently short time the component won't be harmed (at least, not by the effects of heating).

Now there are some other things that people like me (who are also pedants with nothing better to and maybe Claude..) will be wagging their fingers saying "Ah, but..."

And they'd be right. But this isn't about edge cases it's about understanding what's safe, what's not and when to say: well, that's buggered it.

And more importantly: WHEN NOT TO.

If you can't be sure, buy another but sling the other aside if you haven't chopped all its legs off and then expected it to fly (I mean what sort of person does that...)

No really. Assuming you haven't physically busted the part (chopping the wrong lead off at the "root") etc. you'd be surprised at how tough even modern ICs are when pushed way beyond their design specs. They will usually degrade that's true, but I'd rather put a degraded part in a suspect circuit than a brand new one. So they have their uses.

STL;SDR

Ummm... had to skip this far... Still Too Long; Still Didn't Read.

OK. First of alll the diode in your JFET - it really IS a diode. Not something that works like a diode unless you tie a couple of leads together but that's a bit of an expensive way to do it. MOSFETs are not - write this in the blood of your enemies. J/MOSFETs are not the same.

Yes they use the "field" effect but the way they generate the field is structurally very different. MOSFETs have a tiny capacitor made from (in the chips I worked with, Silicon dioxide) -whatever it its, it's molecules thick on the really fast ones. 

And it's also a very large but exceptionally low-powered resistor. That's why beefier "protection diodes" are present on many chips where these are exposed but transistors generally excluded then. Again there is some weird stuff showing up at the highest frequencies but you need a degree in Weird with a 2nd in Strange or Peculiar before you even get outside most radio reception.

The thing that makes it strange in JFET is it's got one input and two outputs and if that doesn't weird you out, it's wired back to front too!

In "normal" transistor (the sort of thing those little hFe meters are testing) there are two diode junctions with three terminals and the diodes are wired (at least for a NPN transistor as if the two "inputs" are wired at the base.

Looking at at from the base - you can see (from diode action) that current cannot (normally) flow from the Collector out through the Base but it CAN move though the Diode formed from Base to Emitter.

An N-type JFET (the 2N4416 here) is the same polarity (it's not strictly, but work with me) as the NPN BJT. 

There's one things that matter here more than almost any other for design - not how much current IS flowing but in the first few picoseconds of time, how much of that field established and causing energy to ripple through the material; and or more consequentially: did the amount of impeding field cause sufficientt loss to friction cause the material to be unable to dissapate heat to a point where the matrix starts to break apart.

That's a fancy way of saying. You applied to much force.

Voltage is the force; current is the result and heating is the effect we're dealing with.

REALLY REALLY TOO LONG, GET TO THE POINT DRACO

Grab some diodes, they're easy to source and cheap if you don't have some. Watch how your meter reacts in diode mode and see how the numbers change as you switch the ends. I'll keep the discussion relatively simple but please do ask and I'll do my best to explain, time allowing.

A modern meter can probably light a small LED because modern LEDs are rather more efficient than the junk we had to put 20 mA through to get a feint glow. Or maybe they were just duds. They were cheap now I think on...

See how that reading changes.

For a typical, el-cheapo diode you're expecting about 0.6 to 0.8 - it might show in mV, it might show in V it might not say but that's the sort of number you want.

That's the Vf - the amount of push the diode needs (this is to do with fields and stuff) to start conducting. There's a curve you can see called the Diode knee or some such but this is mostly stuff you need for design.

One way - nothing, the other way - a fraction of a volt up to a couple of volts for a small RED LED, other colours are different for the reasons I've just outlined. It's dead 'ard.

That Vf matters because that's the number you'll see (something like it) with most simple silicon junctions like this. Not all semiconductors are silicon and each one has its own voltage. Said it was 'ard, didn't I. 😉

Remember that diode in your JFET? Same thing.

You test it the same was the diode in your hand and prove it in a rather shocking way... walk this way if you will.

Find a warm surface. Not hot or failing that get some ice or a bag of peas handy.

Do your diode test on a diode - and see how the voltage changes as you heat or cool it.

Now the same on a BJT (if you have one) - doesn't matter what type or polarity.

And finally the same test on your JFET.

The important point to recall is in the JFET what would be the Emiiter diode is missing.

Right there you can see how similar these two devices are - with a simple slice of semiconductor material making a whole world of difference.

And for anyone who has been paying attention - you might have noticed that ALL transistors use the field effect as part of the control mechanism, the difference is HOW we control it.

In school and in many electronics course, we're taught that the CURRENT flowing into the control terminal of a BJT controls the current from Collector to Emitter and that's partly - true but that current can never even begin to flow without a field appearing at the base. And a field is a voltage...

I could make some Planet of the Apes reveal gag there but you're likely to young, sorry. But yeah, the're all FETs.

Oh and the reason at JFETs are so hard to kill?

The control terminal Gate doesn't have to deal with the channel current as it would in a BJT (Collector -> Emitter).

The cost of that robust feature with other really desirably functions? They're an absolute bugger to work with in all but the simplest of circuits so we avoid them until we're in a pinch and this (as you're starting to realise) is one of them!  



Take everything I say with a pinch of salt, I might be wrong and it's a very *expensive* way to learn!

 
Posted : 16/11/2026 2:25 pm
marcdraco
(@marcdraco)
Posts: 1127
Moderator at Large and Cat's Butler
 

Fantastic. A hour explaining how to fix that and the server crashes with a HTTP: 500  

AGAIN,

@DIYPerks - mate you really need to give your ISP or server providers a good kicking. Three times I've lost stuff this morning. None of us can give or get help (not even you) when the database keeps spitting its dummy out.

OOO... my bad. It just feel over the page boundary, 500 be damned!

The backup was recovered.

See PM mate, that's likely easier.


This post was modified 9 hours ago by marcdraco


Take everything I say with a pinch of salt, I might be wrong and it's a very *expensive* way to learn!

 
Posted : 16/11/2026 2:30 pm
(@klemmie)
Posts: 8
Active Member
 

Jeeze, that's a very lengthy explanation (tirade?). Thanks again, all of this is really helping me get my head around these different components. I should probably learn from this to not look at a "How To" video and think "Pfft, how hard can it be?", because this is definitely harder than I originally thought it would be.

At this point I'm still unsure as to what I could be looking into to understand where my issue is coming from. The preamp seems to be in working order considering I can use a cheap electret with reasonable success. The connection to the JFET and JLI capsule are also measuring correctly. There seems to be no physical (or mental) failure in the capsule. The JFET however seems to not want to play along.

While I (roughly) understand the physics around how the JFET is supposed to work, I am having a difficult time nailing this specific one down. I cannot tell if the pins are wrong, or how can I even tell how to tell the pins apart. I followed Matt's video and expected that pin 3 is Gate, but my (admittedly crappy) multimeter is saying something different. I have yet to test it in each permutation, but I have a feeling that the JFET took too much to the chin (Gate) and is now slammed permanently open.

The new one from ebay is taking its sweet time to get here, in the meantime I doubt there's anything I can do since I don't have anything remotely closely rated to the 2n4416 in my parts bin. 

I guess for now I should just wait patiently for the replacement JFET and once it's here I'll try to test it before soldering it in place. I'm hoping through the measurements that I can determine which pin is which before I fry another one. I'll also get the seller to provide me with a datasheet this time since the original listing and subsequent communications are lacking that.

Thanks for the time taken to explain this in crayon eating terms so that I can follow along and not only fix my problem but also understand why I even have this problem to begin with. Education is key here!


 
Posted : 16/11/2026 3:16 pm
marcdraco reacted
marcdraco
(@marcdraco)
Posts: 1127
Moderator at Large and Cat's Butler
 

It's easy (sitting from the position of someone who knows enough to know how little they know) to decide that everyone who doesn't is a dumb-ass.

That's a fact isn't it?

I've been that dumb-ass. Hell, who am I kidding... but I digress.

I'm one of those people, like yourself, who's like a dog with a bone. My friend here the @imp-inventor is similarly differently able. It's wonderful that we learn stuff with all the voraciousness of ... actually, that joke almost went a bit dark.

But yeah, we're then utterly buggered when no one (who knows enough to give you a hand) is prepared to give you a hand over a fence.

Worse - the SOB's are laughing cheerfully about your loss while they carry on feeling very clever.

That's a passive form of bullying of a form I won't abide and I encourage everyone with a skill to always try and help when someone less knowledgeable is struggling. Even when it means getting the crayons out.

But here's the best bit - I get a fair bit of payback myself because these discussions give me a chance to challenge my own memory to see how much I really did absorb from the people who really do understand this stuff.

You honestly can't buy that!



Take everything I say with a pinch of salt, I might be wrong and it's a very *expensive* way to learn!

 
Posted : 16/11/2026 4:25 pm
Page 47 / 47