Guitar practice amp TIP31/32 w/ lm386

[Matthew Sanford]

Switched gears in my clean off breadboards/learn to solder better current state. For a bench amp (prior to iPhone mic input thingy) I had tried lm386, then S8050/S8550 push pull w/2N2222 and found that after lm386 increased volume and kept from oscillating...so I ordered better BJTs.

Now I used a similar schematic with TIP31/32, a 2n3904 in front, and lm386 in front of that. This is all to drive a classic 25th speaker from the walls of my house built in 1976. I figure I'll put some effect in later (or maybe send/return) but for now just want to accomplish good enough soldering on Vero to have it together. Was thinking to use the old PCB with 4 momentaries and the 100ohm pot (or was it 10 ohm?) but that can be later.

In current state having 1M pot at the front seems to bring it up tons, do get some hum after it warms up, and distortion after half (is better/louder than the 10k was).The old scheme had output feedback from the push pull back to the 2n2222 but this one didn't - should I add that? Looking for things obvious to others but not me to improve it before putting it together.

[PRR]

What does the discrete amp add that the LM386 won't do? Same supply voltage, same load impedance, same power pretty-much.

> distortion after half

"half" is meaningless. Adjust gain as needed, not just "MAX".

And it hums because it is not in a shielded box. (Or maybe crap power too, but an open build always has hum/buzz.)
=================

I did not know Scovill ever owned NuTone. Scovill is like one of the oldest manufacturers in the USA, first buttons in 1802 (my g-g-g-grandfather may have worked there) and other clothing hardware, but also a bunch of key camera developments in the brass/mahogany era.

[Clint Eastwood]

I guess you want to use that speaker because you have it on hand, but at 25 ohms your amplifier will only put out 100mW or so. With your schematic a 4 ohm speaker would do about a Watt. I have had good results  with similar amps using speakers you can find in old tube radios and tape recorders. They are typically 4 or 5 ohms, and often have alnico magnets

[PRR]

...at 25 ohms your amplifier will only put out 100mW or so. ....

Might be a big third-Watt. (9V/2.8= 3Vrms, 3Vrms^2/25 = 0.36W, minus losses)

But even the tenth-Watt may be fine in a lot of home practice situations.

Also depends on the speaker. That 50-cent job may not reach a mild shout. Tenth-Watt in a Full Stack can annoy neighbors. (But this NuTone wall panel is too too cute, and sooo mid-century.)

[Matthew Sanford]

What does the discrete amp add that the LM386 won't do? Same supply voltage, same load impedance, same power pretty-much.

> distortion after half

"half" is meaningless. Adjust gain as needed, not just "MAX".

And it hums because it is not in a shielded box. (Or maybe crap power too, but an open build always has hum/buzz.)
=================

I did not know Scovill ever owned NuTone. Scovill is like one of the oldest manufacturers in the USA, first buttons in 1802 (my g-g-g-grandfather may have worked there) and other clothing hardware, but also a bunch of key camera developments in the brass/mahogany era.

Hum was (mostly) caps on power. I'd joined the breadboards so once I moved it to the other side of the filter caps on the 386 board it went away.

With just the 386, just the 8050/8550, and 386 -> other it wouldn't drive the 25ohm speaker. With it this way it works well, and not really quieter than through my 4ohm 10W TEAC goodwill find. It's not super loud... I do make some MAX1044 chips, maybe I should step up voltage on the discrete amp?

I can't stand it though I love those things, couldn't toss them when remodeling the house! I still have the main one in the wall in the kitchen, was almost thinking to do a project that uses it all (3 like pictured and a speaker at the front door) but doing a practice amp for now.

[Matthew Sanford]

Of these recently acquired choices, is there a louder pair I might plug in?

[Rob Strand]

I haven't checked the details but what you might find is to get more power R3 needs to be reduce.  What often happens is the resistor on the base (R3 = 1k) cannot supply enough base current.

Bootstrapping can help a bit - add C5 and split your R3 into two resistors, here R8 and R9.
You still need R8, R9 to be low enough to supply supply the base current.



A Darlington output stage lets you use higher value of R3 but then you lose 0.7V swing, which can be significant, on a 9V supply.  However then you will need to add more diodes to bias the output stage.

Also make sure the DC bias point at the emitters is mid-supply to maximize swing.

The biggest problem with the amp is no emitter resistors on the output transistor emitters.  If the output stage heats up it could get thermal run away.   Thermally connecting (touching) the diodes onto the power transistors helps.

At the end of the day.  It's a small amp but you need to *design* the amp, and that's not a simple task for a newbie.


This perhaps summarized a lot of the points.  The feedback bias with the 470k helps DC biasing of the output stage.  You design is hard to maintain the DC voltage at the emitters.  You might need a resistor in series with the input to tame the gain.  Change the value to suite the lower supply voltage.


[FWIW, you don't need such large driver transistors for the Darlingtons.]

[Matthew Sanford]

Thank you Rob, I've saved those schematics for study. The upper left 3904...well, I'll study it a bit and come back with questions.

For full disclosure, these were the schematics I used. First with the 8050:8550 transistors



And this for the TIP31/32

[Rob Strand]

For full disclosure, these were the schematics I used. First with the 8050:8550 transistors

The first one has the emitter resistors and a low value collector resistor (330R) and the 33k resistor from the output back to the base of the first transistor.  While simple it tries to do the right things.

The second one is likely to be problematic.

(The second circuit I gave uses two NPN transistors for both output stage polarities but the driver stage is NPN for top and PNP for the bottom.  This is called a quasi-complementary output stage.  The upper combination of transistors is a Darlington but the bottom combination of transistor is called a Sziklai pair.  The quasi-complementary output stage was popular in the old days and an musical instrument amps.   

These days it's more common to see the use of a NPN power transistor and a PNP power transistor:
- the upper transistor combination would be an NPN Darlington and the lower a PNP Darlington.
-  or, you can flip the whole idea and use a Sziklai pair (NPN driving PNP) for the top and a Szikair pair
   pair (PNP driving NPN) at the bottom.

You can't just chuck any only thing in there you have to design the amp to suit whatever type of output
stage you choose.

I only mention these things so you can read up on output stage options.

Amplifier design can get fairly involved.)

[PRR]

I haven't checked the details but what you might find is to get more power R3 needs to be reduce.  What often happens is the resistor on the base (R3 = 1k) cannot supply enough base current.

That's often true. But for these transistors (hFE ~~150 at 0.2A) and 25r load, it won't be too very starved.

.....you need to *design* the amp,...

That's always true. That's why I was poking at "what's wrong with the '386?" It really is a fine design, and is underloaded here. It may be confusion between gain, distortion, and power output (three very different things that melt together in the mind).

[Matthew Sanford]

Thank you for all of that, much to learn on. Time to study up on basic transistor circuits too, do a study guide. Might try your second circuit with the 122/127...and adjust accordingly. Maybe try adjusting the bias, the resistors, and adding a feedback resistor (and emitter resistors) on this one.

[antonis]

The upper left 3904...well, I'll study it a bit and come back with questions.

Just a humble V_BE multiplier (rubber diode)..
It serves for voltage reference precise setting contrary to series diodes "stepping" one..

https://en.wikipedia.org/wiki/Rubber_diode

[Matthew Sanford]

Thank you all so much again, my learning has been so sporadic and random in 10 minute blocks, the fancy of the day, and you're pointing me to how to fill in the gaps. This will take a minute, but I'll actually work to design it properly with your suggestions and the math.

[Rob Strand]

That's often true. But for these transistors (hFE ~~150 at 0.2A) and 25r load, it won't be too very starved.

For nominal devices the 1k is likely to be losing some swing even with a 25 ohm load.  If you only want to lose 0.5V swing then the voltage across the collector resistor needs to be < 0.5V at full output current, so maybe 470 ohm.   The  lower limits on the o/p transistor gain are quite low on these devices - that really screws things up.

That's why I was poking at "what's wrong with the '386?" It really is a fine design, and is underloaded here. It may be confusion between gain, distortion, and power output (three very different things that melt together in the mind).

Agreed, it's very hard to beat those integrated amps with a simple amp design.   All the hard work is done, and it will work.  If you want to do a HiFi geek solution and throw a lot of transistors at a complicated design, that's better than the LM386, then that's possible too.   For simple designs integrated generally wins.


Just for fun I compared a number of simple designs.   I wouldn't call them working designs.  They are barely enough to compare the *output swing* of the different amp structures.  I used an 8 ohm load.  For a 25 ohm load we would scale-up many of the resistors and get similar results.

So clearly the boot-strapped output stage helps get more swing.  For a 9V design at least, the extra voltage drop of the Darlington out weights the gains.   The Sziklai does look promising.

[schematic]


[clipped waveforms]




In order to keep the simplest structure amplifier I didn't investigate adding transistors to the voltage gain sections.  That would also be a way to get higher input impedance and no doubt less distortion (which I have not considered).

Notice this 2.5W design from TI uses a very low collector resistor on gain stage:
https://www.petervis.com/Amplifiers/tip29-and-tip30-amplifier-design/2-5-watt-amplifier.html

https://www.petervis.com/Amplifiers/tip29-and-tip30-amplifier-design/amplifier-design.html

I should mention a couple more things:
- the power transistors in my simulation have a gain of about 80
- the designs with low collector resistors on the gain stage will draw
  extra supply current with no signal, perhaps +20mA.

[Matthew Sanford]

I was just thinking I had gotten some lm1875T chips and it could be simpler to use those for now, for this, in bridge or to drive a power transistor push-pull. But all this makes me want to push on, so maybe a ic for now then go for the learning after...I'll let you know

[Rob Strand]

I was just thinking I had gotten some lm1875T chips and it could be simpler to use those for now, for this, in bridge or to drive a power transistor push-pull. But all this makes me want to push on, so maybe a ic for now then go for the learning after...I'll let you know

The LM1875's are great but they are only spec'd for 16V single supply (+/-8V dual supply).

The 9V supply is always going to cause limitations because the majority of amps won't swing to the rails.   As far as easy to get chips go the LM380 can be used in a bridge (LM386's bigger brother).   There's zillions of amp chips which were targeted for 12V to 16V and operated down 8V to 9V and a large proportion could operate in bridge mode.  Many have gone obsolete and sourcing them is risky due to fakes.

[Matthew Sanford]

Thank you as always for the advice and info. I may do it anyway for now with those and multiple 9v batteries...maybe an option for an old laptop or other plug that offers higher voltage... if I were to run it from different voltage levels, would the resistor values need to change?

[Rob Strand]

if I were to run it from different voltage levels, would the resistor values need to change?

Most IC power amplifiers don't require any changes when changing the supply voltage.

If you change from a single (say +18V) supply to a dual (say +/- 9V) supply then you will need to make some changes to the connections.  Best to check the datasheet for tips.   

IC's like the LM380 and LM386 are designed for single supply so you need to keep ground pins wired to the 0V.   So in a +/-9V circuit with an LM1875 you need to keep your existing LM386 running off +9V and 0V.

[Matthew Sanford]

Ok, sounds good. I think I may ditch the 386, it (I believe, no extra mark) is the lowest, so 1/4 watt. May just do an non booster first, or the first (better) of the first schemes I shared with S8050/S8550 instead... though I did see a schematic with a TIP31/32 output after the 1875.

Still, might start simple 2n2222 buffer to the 1875 then play from there. Will definitely pour over the data sheets when this goes (and likely come back with questions!)

[Matthew Sanford]

So I set up the data sheet LM1875 single supply circuit with an 18v (repurposed laptop supply) source and it works out pretty well on the 25ohm house intercom speaker. There is a bit of...whine I'll call it when not playing, so I'm not sure why, not too many flying wires on it.




I was thinking it might want smoothing caps directly by the power supply coming in though recommendations seem to say closest to the chip. I had tried a TL072 buffer first with Rf 10k and Rin 1.5k (I'm out of 1k) and the other op amp out to -in and + to ground but no sound to the power amp, so ditched (after trying 10k from 1875 out to 072 -in) it for now. I'm thinking once it's off the breadboard the slight noise should reduce, but are there other things I could do to filter the sound better? One other thing, it is only when the input is connected; with the 072 it was silent, and also until the guitar wire is put to input. Volume all the way down is only slightly louder noise. I'm reviewing my ground situation now so will report back if I find issues there...

I found the best size drill bit so I will cut Vero traces properly! Still, any suggestions are most welcome!