Reverb pedal squealing

[jrdworak]

So, built a reverb pedal from GGG. Tested it and it seemed to be working fine. First time I used it for any length of time it was fine for about 20 minutes then it started squealing. There was no hint of the guitar coming through. The volume of the squeal lowered when you turned the reverb down. Bypass worked fine.
Took it home and opened it up. Looked like the jack was maybe shorting on the case. Turned the jack a bit and lined the case cover with electrical tape. Tested fine with and without the case closed up. Figured that was it.
Fast forward a month and the next time I use the pedal and the same thing happens. Fine for 20 minutes or so then squealing. Just tried it again at home on a small battery powered amp and same thing. Fine for a bit then squealing.
No idea what it could be. Any suggestions?

[niektb]

sounds like something might be heating up internally causing something else (or itself) to go out of spec. But more details would be helpful (like a schematic)

[bluebunny]

Welcome!  Pictures of your build would be useful too.

[amptramp]

Is this their digital or spring reverb?  You will get different answers depending on which one it is.

[Ripthorn]

My best guess is that you used a 78L05 regulator instead of an LM7805. You are pulling enough current that the 78L05 can't dissipate all the heat and it begins to shutdown. This has happened to me on higher current draw digital pedals, which a Belton-based reverb is.

[Kipper4]

Is it this  schematic ?

http://generalguitargadgets.com/pdf/ggg_verb_sc.pdf

Looks like a belton brick scheme.

[anotherjim]

^ We don't like the way the unused opamp +input is grounded, do we?

[Kevin Mitchell]

^ We don't like the way the unused opamp +input is grounded, do we?

*raises hand*
In a single supply design? No, we don't like it at all. According to TI it's better practice to put a virtual ground (or rather, a reference voltage of half the supply) on the input.
In a dual supply circuit, grounding the input is OK and is considered proper use of an unused opamp channel.

[antonis]

[Rob Strand]

According to TI it's better practice to put a virtual ground (or rather, a reference voltage of half the supply) on the input.

It's generally better and IIRC the TI docs going into the why's and wherefore's.  However imagine a real circuit that operates the chip according to the specs.   A valid set of inputs can be grounded inputs.  So it is valid to connect to ground, but, as indicated *some* opamps might pull more supply current.  Floating inputs are definitely out, you are just asking for all sorts of difficult to debug problems there.   If routing tracks to Vref is inconvenient then the buffer with a grounded input in antonis's pic is very desirable over grounding both inputs.

[Rob Strand]

Is it this  schematic ?

http://generalguitargadgets.com/pdf/ggg_verb_sc.pdf

Looks like a belton brick scheme.

No output cap on the regulator (IC3) is asking for trouble.  Needs at least 100nF right across the regulator output pins.  Caps on the remote board not good enough.

[FiveseveN]

A valid set of inputs can be grounded inputs.

TI says not for TL07* (but V+ is fine):

[Rob Strand]

TI says not for TL07* (but V+ is fine):

That's for linear/controlled operation.   

While you don't want to push absolute ratings, if the inputs are within the supply rails things will still be OK.  It won't work correctly as an amplifier but you don't care.   IIRC when you connect the +input to ground, ie. outside of the common mode range, the output goes high due to phase reversal but nothing bad happens.  Also I don't believe it will oscillate.

When you connect the +input to +Vcc the output saturates.  Still not linear, but perhaps less dependent on the inner workings.

One of TI's documents talks about getting more noise when unused op amps as left in a saturated state.   In most cases it's not like that "creates" noise.  What they mean is when the opamp output saturates it draws a little more current, that warms up the IC.  Noise increases with temperature, so noise will go up because the IC is warmer.   A big round trip cause and effect that you don't want in critical circuits.

At the end of the day if you connect the unused opamps as a buffer and bias to Vcc/2 you don't have to get your hands dirty with all the geeky details (which is easy to get wrong, maybe I'm even wrong  ).