Can't get Ruby Amp's buffer to work properly

[Borek]

Hey guys,
this one is really bugging me, so I finally decided upgrading from reading only to registering 
So here is my problem:
Whilst the sole amplifier stage of Ruby works well (on breadboard and on perfboard),
as soon as I add the buffer stage, the signal gets quite low with a lot of noise.
This noise is not related to any input, so that even when there is no cable connected, there is the
same hiss as before.
I'm really sure about connecting everything right (did this on perfboard and on breadboard - same results), following
the original schematic:
http://www.runoffgroove.com/ruby.html, except from swapping pins 2 and 3 to stop some heavy oscillations (but amp stage works well).
Also tried around 5 different FETs (all from the same package).
I measured the voltages of the Mpf102:

(Battery: 8.45),
Drain: 8.45
Gate: 0.035
Source: 0.539

So I'm wondering a bit about source being considerable over zero, also with no signal in. Seems like this is the annoying noise.
But as said before, this is not just for one transistor, but a few (whole package).
Also tried like three batteries, different brands.
Has anyone a clue about what to do next? 

[karbomusic]

http://www.runoffgroove.com/ruby.html, except from swapping pins 2 and 3 to stop some heavy oscillations (but amp stage works well).

I'm sure others can help you get it working but the above caught my eye (including the inverting to non-inverting swap). That schematic was one of the first projects I built. I had terrible oscillations and other issues with it - I remember literally trying to make it work by staying up all night and about 5AM, out of frustration, I pulled everything off the breadboard, grabbed datasheets for both the opamp and the buffer and just built essentially the same thing from the scratch but via datasheets, that worked.

I'm sure there is a secret to make it work fine as-is but with the stubborn non-ask-for-help attitude I have sometimes, I went to the datasheets. I would say build the buffer portion standalone and make sure it works first. I probably have a copy of my buffer section schematic lying around somewhere.

Edit: You'll likely want a zener diode on that MPF102 for protection, ask me how I found that out.

[Borek]

Yeah, sounds pretty similar to me last night. So I got the amp module running thanks to the Lm386 datasheet, but the buffer looks quite similar to those I found on the web.
Maybe I destroyed all of my transistors by not using a Zeener diode? 
It sure would be a gift if you found your buffer schematic. Then I would at least have something worth gettinng protected by a Zeener

[anotherjim]

Are you sure you have the right pins for the FET?

[karbomusic]

I think you just need a 9.1V Zener diode from Source to Gate. I'll see if I can dig up the schematic (I had drawn it on a whiteboard and took a picture of it). I think it's basically the same buffer as the ROG one which is essentially the same one in the AMZ article though (2nd example): http://www.muzique.com/lab/buffers.htm

[karbomusic]

Are you sure you have the right pins for the FET?

Great point!

[Kipper4]

 http://www.muzique.com/lab/buffers.htm


Kary beat me to it.

[Borek]

Ok, so the second buffer on this site is essentially the same as the RoG one (only a few other values).
But on this one, there is another cap in front of the whole thing - this should not make such a difference?

I double-checked the connector lugs for about 10 times - so, in words:
Gate is guitar input side,
Drain is supply voltage and Source leads the signal into the amplifier stage?

[anotherjim]

Thing with the ROG scheme is.... if you mis-wire and have the gate on +supply where the drain should be and the drain or source where the gate should be... and Fzzzzz, you fry the FET. There is no DC blocking cap on the input to prevent a low resistance path occurring via the guitar.

The arrow on the FET symbol denotes a diode between both drain and source - so it's very easy, with a DMM set on diode test or low resistance range, to prove which terminal is the gate - it don't matter which is the drain or source because they are interchangeable in the FET's we use - but you definitely have to have the gate where it should be. This is also a fair (but not complete) way to test if a J-FET might be good. Note that a J-FET is normally conductive between drain & source, but the gate, being the anode of a diode-like structure, will only read one way to either of the other 2 terminals.

Beware of bogus/dud J-FET's from cheap sources like e-bay.

This is not a MOSFET, so input zener protection isn't needed. A 10n input cap from the guitar and a 10k resistor directly to the gate pin would be more helpful protection. The 1.5M input bias resistor will discharge ordinary static charge before it can harm.

[karbomusic]

This is not a MOSFET, so input zener protection isn't needed. A 10n input cap from the guitar and a 10k resistor directly to the gate pin would be more helpful protection. The 1.5M input bias resistor will discharge ordinary static charge before it can harm.

Interesting, I toasted no less than 10 MPF102s until I added a zener (based on asking here) - I don't think I tried the 10k but the cap is there out of habit, so I can use the 10k instead?

[anotherjim]

Yes, that is interesting. If you look at all the J-FET inputs schemes, you won't find many with a protection zener.
In many commercial designs, especially amplifiers, you do find a pair of ordinary diodes from device input to + supply and 0v as well as a series current limit resistor before there. That is essentially what there is built into the inputs of CMOS gates like the CD4049 etc.
A J-FET, is no more delicate than an ordinary BJT transistor.

If you were soldering the FET's in, a leaky/un-grounded soldering iron tip can apply harmful voltage - much more dangerous than static IMHO.

[karbomusic]

If you were soldering the FET's in, a leaky/un-grounded soldering iron tip can apply harmful voltage - much more dangerous than static IMHO.

Thanks, I can describe the behavior as I remember it. When I first built it, I'd use it as my tiny test rig for pedals. I should mention I added an FX loop to the main design which is a send/receive that lives between the buffer and LM386 amp. I'd often test pedals by going in the main buffer input and/or using the FX loop. When connecting/disconnecting these, is always when the MPF102 would fry and go belly up. So it wasn't from soldering et al because it only occurred post build and more importantly because that particular build uses a transistor socket for easy swapping. It never happened again after adding the zener.

[Borek]

Ok, so reversing the transistors (swapping source and drain) kills the noise. But still the amp is much lower in output with buffer than without.
Actually, my multimeter says that there is no conductance between source and gain.
Looks like bad JFETs? Got them from Ebay, but not too cheap, I think.

[antonis]

Actually, my multimeter says that there is no conductance between source and gain.
Looks like bad JFETs?

Despite DRAIN or GATE, for a n-channel FET you must have a very HIGH resistance between Gate & Drain and Gate & Source if you place Black lead of multimeter to Gate and a very LOW resistance for the opposite arrangement..

(modified stolen pic)


Checking Drain - Source channel is another matter and it demands some kind of "neutralizing" possible build up voltage across the gate-junction PN channel - like inserting FET pins into antistatic foam..
The resistance across Drain - Source should be relatively LOW (a few hundred ohms in most cases) which resistance should be equal for both "direction" measurements..

A final test could be to apply a reverse-bias voltage between Gate and Source (pinch-off the channel) and read an increased resistance on your multimeter..

P.S.
Physically revercing FETs DOESN'T swap Drain & Source 'cause almost all FETs don't have a Gate middle pin-out..

[Borek]

Oh, of course, physically inverting the transistor would kill it immediately.
I will try the method you described later. Also some local store has BF545a there, so I'll try these. Although I have to deal with them beeing surface-mount. Maybe just going to solder some short jumper to it 

[Borek]

Ok, so the Bf545 works nicely.
But since it's surface-mount, I'd really want to get this thing running also with the Mpf102.
So ich measured some resistances with the diode test function of my multimeter using the scheme posted by antonis.
I tried this with a few of these chips and with every one its gives some strange results:
According to this test the gate is actually middle pin, because then the resistances are either OL or around 0.63, when connected to the other (outer) pins.

I actually tried wiring one of those Mpf102 considering middle pin as gate, no more hum/noise, but quite a low output.
So this does not seem right either.
Can there be some discrepancys in the pinout arrangement also for the same chip?

[bluebunny]

I actually tried wiring one of those Mpf102 considering middle pin as gate, no more hum/noise, but quite a low output.
So this does not seem right either.

Indeed it doesn't.    Every pinout picture I can find the for MPF102 shows D-S-G.  The gate isn't in the middle.

[antonis]

@Borek: Did you meaure RESISTANCE (not diode/continuity/beep test) between Drain & Source..??
(after proper Gate-junction PN channel discharge..)

[Borek]

@antonis: I did not yet, because I was not even sure anymore where gate is, because of the findings of the diode test between alleged gate and drain/source.
So I will do this as soon as I find out how to do a channel discharge

[antonis]

Your fingers are the most handy tool...

For Drain - Source channel you MUST measure resistance..