What causes "Motorboating"????

[vortex]

Motorboating: A steady drone that is present in the audio signal when the circuit isn't working properly. It sounds like a motorboat! Note: Not ground hum.

I hope my definition is enough to describe the problem. i have heard "motorboating" in other circuits and today encountered a situation where it happened again. I am thinking that there might be a common issue that causes motorboating and am hoping someone might have some insight.

I am helping a friend get started on pedal building and we both were building a Ruby amp today. We were building the same circuit with a vero layout I made. We both used identical parts etc. My circuit worked fine but his was "motorboating". I tried everything I could think of to debug his circuit, I swapped the chip and tranny with my working circuit, checked the solderwork/continuity/connections/wiring to no avail. I even tried biasing the JFet thinking that it might be the source of the problem.

Anyway suggestions?

[Paul Marossy]

I just had that problem with my Anderton Frequency Booster. Turns out I had connected V+ to the point where Vref was supposed to be originating from. Made for a nasty motorboating/oscillation sound! OTOH, could be useful for sound effects! 

[vanhansen]

That happened to me with the Ruby amp.  Can't remember exactly what it was but it had to do with the + DC.  It didn't take long to fix.  It's like a type of oscillation from overload.

[vortex]

That happened to me with the Ruby amp.  Can't remember exactly what it was but it had to do with the + DC.  It didn't take long to fix.  It's like a type of oscillation from overload.

Yes, it does sounds like some kind of overload plus the motorboating. After a chord is struck there seems to be a decent clean signal right at the very very end of the sustain.

Should I be upgrading the 100uF cap from the + to - rails?

[Paul Perry (Frostwave)]

Upgrading the cap is a good idea. The cause of "classic" motorboating, is impedance of the power supply being too high (cap too low can do this). The theory is, that as the audio signal swings, the power supply voltage swings as well. Now, most ICs are pretty resistant to power supply ripple.. but, it is still easy for power supply ripple to get back into the signal chain (eg through the 1/2 Vcc bias jumping up and down as well) and causing positive feedback.
Now, the filter caps usually mean the feedback is very low frequency indeed, hence the "motorboat" sound.
It was pretty common in the old tube radio days, because of the expense of making a good power supply.

[R.G.]

High frequency oscillation is caused by phase shifts and inadvertent feedback at high frequencies.

Motorboating is the equivalent at low frequencies. It's caused by series capacitors between stages and a lack of power supply isolation between stages.

The input to a capacitor coupled stage is an RC highpass network. That implies that there is a low frequency rolloff (and that's why you want it there!) and with that low frequency rolloff, there is a phase shift of 0 to not quite 90 degrees depending on the frequency. Stack two inverting stages together and you get a rolloff of up to - but not quite - 180 degrees of the low frequency signals that pass through them. Three stages an you're sure to get at least 180 degrees at some low frequency. That's enough to satisfy one of the demands of an oscillator, 180 degrees of phase shift. The feedback part is all that's lacking.

You get the feedback part by poor power supply decoupling. If your power supply is not very low impedance, then the power supply sags and rises with changes in signal current. Since classical tube stages - and non-differential transistor or FET stages - have substantially no power supply rejection, then any junk on the power supply  goes right into the signal chain.  If the junk on the signal chain is big enough to self sustain at the end of the amplification chain, then you get motorboating.

To prevent motorboating you have to change any of the oscillation criteria.
(1) phase shift of 180 degrees; oops, can't do that. Series RC coupling is going to do it. You *can* make them at vastly different frequencies. That helps.
(2) Loop gain of 1 or more at the phase shift frequency; either lower forward gain (ugh! bad!) or increase the attenuation in the feedback path  (good!). Since the feedback path is the power supply, all's you gots to do is either make the power supply so low impedance that it shows no sag/signal (easy to do nowadays with regulators) or make the circuits immune to power supply junk by increasing the power supply rejection ratio (Hey! THAT's why PSRR is on opamp data sheets!!) or differential circuits, or by preventing feedback over more than two stages. And this last is why you'll find only two triode gain stages per R-C filter in the power supply on almost all classical tube circuits. The R-C decoupler inserts a big R and a big C to attenuate power supply junk a lot. If each cap feeds only two stages, it pretty much can't motorboat. Solid state circuits have other ways, but with triodes it was simple.

Of course, this is all just theoretical, and as we all know, if you know the theory, it's always wrong because all that textbook learning is stifling, constipated, and shot full of errors by people who have no real world experience in how things REALLY work.

[vanhansen]

Hey, R.G. just joggled my mind, and taught me something in the process.  The problem I had with the Ruby motorboating when I built it was a low charged battery.  A fresh one made it stop, but since the 386 has a 40mA draw  , I use an adaptor anyway.

Thanks again, R.G.  Great explanation as always.   

[R.G.]

That makes sense - as the battery's internal resistance came up, the motorboating started. A big capacitor might help, but maybe not with 40ma of current coming out.

[Paul Perry (Frostwave)]

Anothher thing about instability (from any cause)... usually, it is only noticed when the device goes into full oscillation and howls or groans or whatever. But, suppose it is just on the edge of doing this.. it is going to have pretty severe tonal effects on the signal. And, I suspect, this is where a LOT of the arguments about what sounds GOOD or BAD come from, it might be the same circuit, but in two different builds by different people, the actual response might be very different.
Bercause, a circuit isn't just the stuff you bought from SmallBear, it's also the invisible capacitors of a few pf between each pair of components, the resistance in teh leads and wire, the impedance in teh power supply or battery.....!

[R.G.]

Anothher thing about instability (from any cause)... usually, it is only noticed when the device goes into full oscillation and howls or groans or whatever. But, suppose it is just on the edge of doing this.. it is going to have pretty severe tonal effects on the signal. And, I suspect, this is where a LOT of the arguments about what sounds GOOD or BAD come from,

It's all in the eye of the beholder. Marginal stability due to power supply impedance is one of the things making the Fuzz Factory do what it does. Some people like that.

[Paul Perry (Frostwave)]

Exactly my point, R.G.!

[vortex]

Well a massive thanks to all who responded to this thread.

I revisited the motorboating Ruby today and first thing I did was set it up with a fresh battery.  I had not realized that the Ruby required 40mA. Thanks to VanHansen for hitting the nail on the head.  Much to my surprise and embarassment the circuit was working fine. I guess the battery my buddy was using was not providing enough power. I hadn't even thought of the battery being the culprit during my initial debugging.

I am still absorbing RG's reply (as usual), much appreciated!

I built the Ruby with the Bassman mods and it is quite clean and warm sounding with my 1x12 cab. I am thinking I may tweak the gain resistor to see what's up with that. To my ear it doesn't sound even remotely close to a real Bassman but it's a decent little amp for practice etc. I have built several projects with the 386 chip and so far this is the best.

Cheers!

[vanhansen]

Hey vortex, glad to hear it was that simple for you too.  I decided to run my build off wallwart power only.  Once I looked at the LM386 datasheet and saw the current draw, I knew I'd be using a lot of batteries.  My build is stock, but I added a master volume on the output like the Little Gem.  It sounds great with the gain and volume cranked up, especially through a 4x12.  Try it out.