Little Gem MKII not working on the BB

[Ben Lyman]

Good questions Rich, as far as I know this should work with 1/4watt resistors.
I have tried it with a single 12" 8ohm, a single 4" 8ohm, 2x12" 8ohm series & parallel, 2x4" 8ohm series & parallel, all with the exact same results.

These chips can still do a Ruby or Smokey just fine but not the LG MKII, perhaps as they try to do this "different kinda task" they simply lose their minds.

How about a possibly bad Q1 doing tricky stuff to the chips?

Def IC2 that gets red hot, I can't even read the print on it anymore because it's been heated up so many times, but still drives a Ruby/Smokey circuit with the Mesa 12" cab.

Here's todays readings:

Q1
D 9.3
S 3.4
G .03

IC1
1 0.21
2 0.01
3 0.29
4 0.01
5 5.2
6 8.5
7 4.5
8 1.0

IC2
1 1.0
2 0.41
3 0.01
4 0.44
5 2.5
6 9.3
7 4.8
8 2.0

[jonnyeye]

Sounds like you may have ultrasonic oscillation; instead of the 10 ohm plus 0.1uf damping/Zobel network across the two outputs (pin 5), try a 10 ohm plus 0.1uf from each output (pin 5) to ground, using the shortest wiring you can.

Also, the LM386 is only really spec'd for 4 ohm operation and thus the minimum load it can drive when bridged like this is 8 ohms; even then the output stage is current starved. You'll get most power out with a 16 ohm load (your 2x8 ohm speaker cabinets with the speakers in series will work, once the amp does).

[Ben Lyman]

Thanks jonnyeye! I'll do that for sure ASAP.

I thought LM386 took an 8ohm speaker?
Anyway, the ROG LG MKII said use 4ohms or more for this circuit.
Oh well, I'll see if I can get it going. Anybody else wanna BB this? I'd really like to know if I'm working on something that only exists on paper 

Edit: no go on the 2x 10r plus 2x .1 pins 5 to ground. bummer

I pull IC2 out and IC1 starts working like a champ (more like a Ruby). replace IC2 (with new chip) and volume splutters, chip heats up.

[slacker]

I pull IC2 out and IC1 starts working

Something's wrong here, if you just pull IC2 IC1 on its own can't make any sound because only one side of the speaker will be connected to anything. Sounds like you're connecting the speaker wrong, one side goes to IC1 pin 5 the other side goes to IC2 pin5 and no where else. You're not connecting one side of the speaker to ground are you?

[Ben Lyman]

No, the speakers are hooked up just as you say, pins 5 and 5 only. Not sure why it worked without IC2. Is it possible for the speaker to work if (-) goes to (+) through the 10r and .05uF? Because that was still there. I think something is wrong with my 1/4" in/out jacks w/DPDT ground wires or my breadboard when it comes to this circuit, anything else works fine though so I dunno.

btw, I just went back to the Ruby to test these chips and they are both still working fine. They are LM386 N1 which sounds like maybe lowest of the low for heat tolerance. Maybe they heat up and pass all kinds of signals through them, then cool down and work again for the Ruby/Smokey.

[Phoenix]

Is it possible for the speaker to work if (-) goes to (+) through the 10r and .05uF?

No, if you pull IC2, only one end of the 10R and 50nF are connected to anything (except the speaker in parallel of course), there is no return loop.

Do you have your input and output jacks mounted to a metal plate or something so that the sleeves of both jacks are shorted together?
The output jack MUST be isolated from ground for this bridge mode operation to work.

Also, if you leave IC2 and pull IC1, does IC2 still get hot, or does it work in the same way as the reverse?

[Ben Lyman]

I'm not using the output jack so I don't think it should matter but maybe I'm missing something here. I will bb it again later and go from guitar straight to bb. Now I wonder if unwanted connections are being made within my bb but I've made and remade this thing so many times in different positions and on 5 different breadboards, doesn't seem likely I'd get the same issue everytime by coincidence.

[anotherjim]

Ben, are those voltages with or without speaker connected? Without speaker, you should get typical 386 voltages, but if something is amiss in that case, particularly output pin 5 volts different, the low resistance of the speaker is trying to pull the outputs together against thier wishes. Unless both outputs of a bridge amp are proven to have the same DC (within a few 10's of millivolts), don't even try to connect the speaker. Do that with a proper meaty power amp, and you fry the voice coil and other choice bits.

In case oscillation is a problem, try 100nF ceramic disc from pin 6 of each amp to ground. I think always fit one with IC's, even if it makes no apparent difference!

Try 2 Zobel networks. One for each output to ground rather than 1 across the speaker. I've just searched "bridge connected IC amp schematic" and every one uses 2 networks.

[duck_arse]

how come both pin 6's are connected to the same supply, but show different voltages? maybe power off and test continuity, to prove your links are linking like you're thinking.

also both pin 4's. should be 0V, or whatever your meter reads when you short the probes together (my chitbox wanders quite a bit with its zero reading).

[Phoenix]

I'm not using the output jack so I don't think it should matter but maybe I'm missing something here.

So how are you connecting to your speaker? Crocodile clips or something?

I'm quite convinced that you're shorting the output of IC2 (pin 5) to ground - what's labelled as the "negative" terminal - entirely meaningless and should have been omitted from the ROG schematic, as well as the "Pay attention to the speaker polarity" which doesn't matter at all, and they missed out the most important detail which is that the "negative" speaker terminal must be isolated from ground.
This would explain why IC2 is heating up when on the breadboard, and why the circuit works (at half power) when it is removed.

Please follow my above advice and remove IC1 from the breadboard, I'm certain that IC2 will still heat up, and will produce no sound.

[anotherjim]

I found this...

Note cross connection of the unused inputs  - not grounded. Balance control between pin 7's? Hmmm....

[Phoenix]

I found this...

Note cross connection of the unused inputs  - not grounded. Balance control between pin 7's? Hmmm....

Pin 7 is the Vbias output, normally used to attach a bypass cap for better PSRR, but connected in that manner it can also be used to help balance out DC offset between the two chips at the expense of a small amount of headroom.
It's not relevant to the problem that Ben is having though.

[Ben Lyman]

Thanks everyone, I'm trying as fast as I can to catch up to all your suggestions.
I had it working a minute ago.
I eliminated all extra jacks and connections, went straight from guitar cord to BB, straight from pins 5&5 to speaker (yes, gator clips everywhere) but even doing this was no fix for the problem.
I am 99.99% certain I have everything exactly correct to the ROG schematic.

Then...
As I lifted IC2 up and almost out, I heard a noise. So by resting it gently on top of it's BB holes and adjusting IC1 in the same manner, I was able to get a very loud and "real" sounding guitar amp.
Then both chips heated up so I shut it all down again.
It must be my BB, I am very careful, slow, patient, etc. I usually don't have any BB connection problems.

Do you think, when I push the IC down into the holes, the fat part of the IC pins could be spreading the BB terminals and making them all short together?

It almost seems like that is happening but I have made tons of IC projects and never had that happen before.

So, it was sounding great (for what it is) and seems promising, except that the chips still heated up way too hot.

[anotherjim]

All pins are different V. If adjacent rows were shorting, the pins in them would be shorting & have same voltage.
Pin5's are very different which is cooking the chips.
IC1 = 5.2V. IC2 = 2.5V (unless you read the DMM backward there?   2.7v across speaker. DC R of speaker approx 3R?
Then current = 900mA. The chips no like, even if speaker R=4 it won't like. Maybe 386 outputs have different self limit sensitivity to source or sink current fault - One shuts down & stays cool, the other gets hot.

Even when volts are ok, I think a 4R speaker is too much load. Bridge operation is a wonderful trick - it doubles the voltage swing across the speaker. That doubles the current. Power is proportional to Volts x Amps  (it is AC and load is inductive, so the formula isn't exact, but it's near enough). You then have about 4x the power for a given supply voltage compared to a normal single amplifier. A car audio amp rated 22 to 28W RMS per channel is very likely to be a bridge amp. 6 to 7W RMS rated ones are single.

[Ben Lyman]

OK, chips are cool, volts are stable, got some accurate readings although I don't really know what they mean.
Problem is the volume is super low, clean, useless.
Pull IC1 pin3 and volume cranks, overdrives, sounds great, then IC1 overheats 

One Spot 9.4v

Q1
D 9.37
S 3.36
G 0

IC1
1 1.19
2 0
3 0
4 0
5 4.95
6 9.23
7 4.72
8 1.14

IC2
1 1.20
2 .01
3 0
4 0
5 4.52
6 9.43
7 4.76
8 1.28

I call BS on this circuit 
Your challenge, should you choose to accept it:
Prove me wrong using a breadboard 

Thanks for all your help everybody, it really sounded great in that brief moment but I want to build something I know will work so probably going to start a noisy cricket or ruby instead.
That's not to say I'm giving up on this one, if you guys have anymore ideas I will keep working on it because it's definitely louder than a single LM386

[Ben Lyman]

Update: got it working perfectly (almost) by trying a bunch of 386's, maybe they needed to match up better or something, maybe I fried one or two previously, I dunno. Strange that they still work on their own for a ruby. 

Components and leads are solid, I have not changed a thing from the original schematic, sounds great driving a 12" 8ohm, loud enough to jam out on, probably even gig with a mic on it, depending what kinda music you play.

Both chips get too hot, very hot. I have to stand by my claim for now, the challenge stands... any takers? 

[Ben Lyman]

Nobody wants to try this? Too bad, I just had a breakthrough moment and discovered the missing component to the ROG schematic!
This thing cranks a 12" speaker, I like it best at lowest gain, it's super clean and loud like a PA or something.
If you wanna know the trick, PM me and I will send you my paypal address 

Just kidding, I was just running through all your suggestions one at a time when I decided to put a volume pot on it the way the data sheet shows in the "common uses" schematics.
A 10k pot solved all the problems, lost the teeny tiniest wee bit of output volume but well worth the trade off. Lug CCW to ground, lugCW to .22uF, wiper to input pins, I guess a voltage divider would work too.

I plan on fussing with it some more tomorrow, I don't want to build it until I'm sure it will remain stable.
I still wish someone else would tackle it and see what happens

[anotherjim]

Good going. I was going to suggest the inputs could be interacting when just tied together like ROG scheme shows, but needed to read up the data sheet. I thought separate input caps might do it, because the data says they don't have to be ground referenced by the input. It looks as though common ground reference can work. I have a worry though that in some cases, the offsets between the 2 chips could still be bad.

[duck_arse]

so, if you did fit a pair of 220nF from the fet instead of the single connection, what DC voltages do/would you have at the two inputs?

and if a pot cures it, maybe a series resistor to each input from the single cap would fixxe?

[Ben Lyman]

I like those ideas, thanks guys.
I seem to have another problem, I noticed my forearm burning as it rested on my guitar strings, feels like more than 9v and it goes up and down as I turn the volume knob.

My guitar cord goes straight from guitar to the bread board, tip to Q1gate, sleeve to the ground rail