Squealing ruby

[PRR]

> Does the source pin have to have some ground connection?

It does/should; through R2 3.9k.

(FWIW: the drawing had a "error", the internal source lead didn't go all the way to the chip. But that would not stop you from running the S pin to R2 to ground.)

[Locrian99]

Hey PPR thank you for the layout suggestion you were spot on.   So I've tried this a few different ways.  3.9k straight to ground as the layout/schematic read and I get sound it's very low and distorted (.  I tried following the suggestion of a voltage divider witch cap and running that straight to the source strip, and upping the resistor to ground to 8.2k (this got me to 2.69v)   No sound.   Tried a higher value resistor without the voltage divider back to low volume distortion.   Tried Antonius suggestion of 5.6k got me around 1.9.    Left just the 8.2k in got 1.98 still low volume distortion.   None of these seem to change the 1.3v at pin 1 of the lm386 though or the 4.59 at the output.  Not sure if that matters.   I'm suspecting this HEP802 is not a direct equivalent for the mpf102 at this point. 

[Rob Strand]

It's worth checking your JFET pinouts, see attached pic.

Based on your measurements I estimate the JFET has a VP of about 2.4V.
FWIW, the specs for either the MPF102 or HEP801 are all over the map.
Old school JFETs had very loose specs.   Modern day versions tended to be
tighter than the datasheets in practice.
(You can see some MPF102 stats at, https://runoffgroove.com/fetzervalve.html, near bottom of page.
You can see the specs are much wider, https://www.alldatasheet.com/datasheet-pdf/pdf/53879/FAIRCHILD/MPF102.html)

I've got doubts that you can get this to bias without a divider on the gate like this:


If you are going to do that you might as well leave the original 3.9k source resistor in.
You might need to lower the ground side gate resistor to tune the source voltage, maybe around 1M2.
Use 2M2 for the top resistor.

Here's all the info I have on the HEP801,

[Clint Eastwood]

You can also use an ordinary transistor in a bootstrapped emitter follower circuit.  Below is one with  Zin of about 1.3 Megaohm:

[antonis]

You can also use an ordinary transistor in a bootstrapped emitter follower circuit.  Below is one with  Zin of about 1.3 Megaohm:

Rc could be omitted (in fact, it MUST be omitted 'cause if you try to bias Emitter near around 4.5V you'll result into a saturated BJT..) unless the "buffer" above is in fact part a CE amp folowed by a buffer..

[Clint Eastwood]

You are right Antonis. I did a quick redesign of the circuit, I think it is better now:




Input impedance is about 1Megaohm, emitter voltage 3,3 volts.

[antonis]

It should be even better if you interchange R2 & R3 values and make R5 3k9 - 4k7 ..
(the lower the Emitter resistor value the higher the working current hence the bigger the load drive capability - translated into "linearity"..)

P.S.
See Figure 1. configuration (B) here: https://www.diystompboxes.com/smfforum/index.php?topic=129250.0

[Clint Eastwood]

Thank you for the link Antonis. I see you are an expert on the subject 

[antonis]

I see you are an expert on the subject 

.............................................................................
edit: Just realized how badly it sounds in English..
(sorry, no offence intention..)

[Locrian99]

Thank you guys for taking the time to throw out some ideas.  From your schem rob I can see I was doing the voltage divider thing wrong last night I was running the voltage output from it straight to the source pin rather than through the gate.   Probably explains why it killed my signal.   I'll try messing around with that.   Is 3v at the source pin the bias goal here.   Obviously I can use my ears, but voltage readings just seem so much more tangible.   The emitter follower circuit, is the transistor part crucial or should I just be able to throw something there.   Like a 2n2222?

[antonis]

The emitter follower circuit, is the transistor part crucial or should I just be able to throw something there.  Like a 2n2222?

Generally speaking, not so crucial..
But a high beta (h_FE) transistor could afford higher value bootstrapped bias resistor without too much voltage drop..
(Voltage divider Thevenin equivalent resistance could also be neglected so divider and Base voltages could be considered identical..)

[Rob Strand]

Thank you guys for taking the time to throw out some ideas.  From your schem rob I can see I was doing the voltage divider thing wrong last night I was running the voltage output from it straight to the source pin rather than through the gate.   Probably explains why it killed my signal.   I'll try messing around with that.   Is 3v at the source pin the bias goal here.   Obviously I can use my ears, but voltage readings just seem so much more tangible.   The emitter follower circuit, is the transistor part crucial or should I just be able to throw something there.   Like a 2n2222?

Biasing the source to anywhere in the 4.5V +/-1.0V range should be fine.

I'm not against the the BJT buffer version.   In many circuits JFET buffers don't add anything other than cost and headaches.  It's generally a good practice to use a high-gain low-noise transistor for a buffer.  If you use a low gain transistor it can still load the input and the emitter bias voltage can drop.

[Clint Eastwood]

edit: Just realized how badly it sounds in English..
(sorry, no offence intention..)

No offence taken! I thought it maybe is typical Greek humor 

[bluebunny]

I think such wisdom needs to go on a T-shirt.   

[Locrian99]

Did up the bjt buffer last night got it all set up voltage going into the output cap of it was right at the 3.6.   Hooked it all up and the squealing was back no signal.   Took it out of the signal path, and still the squeal.   I've got some different jfets coming and an386-3.  I'll keep tinkering on this and play with the voltage divider and buffer will let you know if I get anywhere.   Thanks again for sharing your knowledge. 

[Rob Strand]

Did up the bjt buffer last night got it all set up voltage going into the output cap of it was right at the 3.6.   Hooked it all up and the squealing was back no signal.   Took it out of the signal path, and still the squeal.   I've got some different jfets coming and an386-3.  I'll keep tinkering on this and play with the voltage divider and buffer will let you know if I get anywhere.   Thanks again for sharing your knowledge.

Well, if it's not a LM386 oscillation issue it could very well be a layout or bypassing issue, for example:

The wiring of the volume pot is too close to the speaker output wire.  Best is to move the wires away from each other.  Beyond that is adding a 10k resistor between the volume wiper and the IC, located at the IC end of the wire.  If that doesn't work also add say a 220pF cap from the IC input to ground.

You can also get oscillation problem when the speaker ground currents flow through the same ground tracks as the smaller audio signals.   You could try connect the 0V from the 0V near pin 4 of the IC and the ground wire of the speaker back to the sam point.   If you have a 6.5mm jack for the speaker, or you are switching the 9V power on and off using the input jack, like an effects pedal does, it can promote bad grounding when you have high current circuits such as an amp driving a speaker.

Yet another possibility is the oscillation is caused by power supply fluctuations feeding back into the buffer.   In this case you need to filter the power rail going to the bias divider on the buffer; or filter the supply to the buffer.   At first you could just try a 100uF cap across the power rail.


Here's an example of a filtered bias set-up.  As used on Ibanez pedals.



The unequal resistors are a tweak to raise the bias voltage on the emitter.
You can increase the filter cap as designed.
The actual part values aren't so important it's the idea behind it.

[Locrian99]

Yea i feel like it's something being too close but it really seems mostly tied to the volume pot, not sure if it's the nature of the circuit.   So I will try the resistor, my bench is pretty crowded so it could be plenty of stuff.  I actually took a picture of the layout without the squeak I'll check that and see if it has any clues.  Also I wondered if somehow my test box was causing something like you mentioned.   I can take that out of the equation easily enough with a 9V and an input jack just wired to it.  Unfortunately the only speaker I have to hook it up to at the moment is a home theater speaker I had laying around. 

I've got a half built deep blue delay I should finish I may do that while I wait for the other Jfet to arrive.  Along with my 3.5 mm jack etc

[Rob Strand]

Yea i feel like it's something being too close but it really seems mostly tied to the volume pot, not sure if it's the nature of the circuit.

You have to draw a line between the cause and the fact the volume control has an effect.     Oscillation occurs when the output signal feeds back to the input and that the fed back signal is strong enough to sustain oscillation.   The volume control is just a dial that sets how much signal goes back around, backing it off has to stop the oscillation at some point.   However, the thing that is coming back can come from any part of the circuit before the volume control upto and including the volume control itself.

The added resistor (and added cap) reduces the amount of signal that gets around the unintentional feedback loop.    The unintentional feedback signal, especially that due to wiring, is usually stronger at high frequencies than low frequencies.   The added resistor and cap knocks down the level at high frequencies only,  frequencies outside audio band.   That's where this solution can work.

In severe cases the unintentional feedback signal is both strong and within the audio band.   This can happen for example when the signal gets in through the power supply or ground.     You can't filter the audio path to prevent this without affecting the audio.   It's best to filter the problem signal at the source, ie. on the power rail itself, using the filter bias.

When the buffer wasn't working this automatically blocked or reduced the audio around the unintentional feedback path.  It's only now when the buffer is working that a deeper problem has been exposed.

[Clint Eastwood]

My gut feeling is that you have a ground layout problem. Here is a link to a video I found very instructive:

https://www.youtube.com/watch?v=lBGE5lwbruE&ab_channel=JohnAudioTech

[anotherjim]

The Beavis breadboard layout assumed you had linked power rails between the 2 power strips - I can't clearly see any links to do it. Also, some cheap breadboards may actually have power strips split in the middle of the run, even though there are red & blue stripes printed the whole length.

It isn't actually illegal to have the JFET buffer with the simple bias in the original scheme, but at 9v supply, it's too sensitive to the variance in the JFET characteristics. If you had 12v or higher, the success rate would improve.

I think it was said that the amp chip here is an LM386-1. The datasheet spec only gives a power rating for a 6v supply (Why? It can run up to 12v)  with an 8ohm speaker. I don't think I saw what speaker this build has.
Did someone praise the JRC/NJM386? This one is, I believe, only made to the LM386-4 specs.