Trace Elliot Commando Boxer 65W - need expert help is it oscillating ?

[chazpope]

actually I was wrong, the wire is .6mm, I wound it up to 27V AC, new current will be higher so no idea yet what DC voltage I will have under full load, hope it's not much over 30V. I made an additional other coil in case I need to install a 12Vfan, I will try to use the existing heatsink if possible by just bolting the chip there and soldering a minimum of components on it, haven't assembled the transformer yet.

will you elaborate on the 'unity gain buiffer' from above - do you mean leave the LEDs but take R20 out ? RV4s the master pot, what about R18 ..

[J M Fahey]

1) 27+27VAC is too much for LM3886, it means +/-38V. Do not count on the amp loading that down.
Rewind for 22+22VAC.
2) Current won't be higher but lower.
Stick to your 8 ohm speaker, do not load it with 4 ohms.
You want to rebuild it as a trusty usable amp, not the ticking time bomb it was before.
Since you can't get rid of 1/2 TL072 anyway, I'm suggesting you a mod yo use the 2nd half as a unity gain buffer, which is useful.
Until now itīs used as the input stage of a now non longer needed power amp.
Lifting R20 puts the Leds out of the circuit.
You may leave them there or pull them out , itīs the same.
Since you will need to pull everything after R20 anyway, plus the couple parts I mentioned, go all the way and pull them out too.
Pull R18, C18, R21, C8, R20 and short C12, to leave that Op Amp "clean", unity gain.
And everything else that belongs to the power amp.
You can drive your chipamp from Pin 7.
And add a 220K resistor from pin 5 to ground.
Relying on pot wipers for DC return put me nervous.

[chazpope]

thanks, why do you think current will be lower, I thought 'same power/lower voltage = higher current' ? I will do some tests on voltage drop tonight.

the load is 6 ohm and I am stuck with it - so I figured for 68W the voltage needs to be about 33V. Limit wise the chip should be okay up to +-42V, so I guess it's just power/temp I need to watch. The traf looked feeble anyway and I didn't want to take too much of its muscle (coils) away

[J M Fahey]

why do you think current will be lower, I thought 'same power/lower voltage = higher current'

True, but it's not "same power".
It's "same impedance + less voltage=less current"
Ohm's Law.

I will do some tests on voltage drop tonight.

Don't count on voltage drop.
LM3886 is factory specified as +/-32V (idle) for 50W into 8 ohms ... and they know what they're talking about.

the load is 6 ohm and I am stuck with it

Maybe it's 6 ohms DC resistance= 8 ohms nominal impedance.
Consider it 8 ohms.

so I figured for 68W the voltage needs to be about 33V.

More than the datasheet specifies, and that for 50W into 8 ohms.
For maximum power (which will not reach those 68W anyway in the real world) you'd need a 4 ohm speaker and even lower rails, around +/-28V DC.

Limit wise the chip should be okay up to +-42V,

They specify 32 V and 28 V rails, for 8 and 4 ohm speakers.
What they can stand unloaded, at 25šC, coupled to a ninfinite heatsink and in a Lab is interesting data but not a realistic design goal.
Even if it apparently work well in your testbench at higher values, they will often mute in the middle of a show, in a smoky hot humid poorly ventilated Club.
They incorporate an on chip thermal protection which simply turns them off or attenuates signal as much as necessary.
Play it safe.

[chazpope]

I was looking at the Output Power vs Supply Voltage chart (in data sheet SNAS091B) - and there 68W corresponds to 6 ohm 33v or, when like you say it really is 8 ohms then it is about 35V. Then again on top of the data sheet it says: 50W 8ohm +-35 Volts. (not 32Volts ) are we looking at same diagram and which figure is relevant ? I have used this chip only once in another amp and there it was by design 38V from a strong toroidal traf into 6 ohm nominal speaker. So is it really that with 8ohm load it can never reach 68W anyway - then what is the chart for?

[J M Fahey]

I am using the same datasheet as you, but I am not trying to put "the highest power number" on some publicity sheet, rather prefer to design a safe hardworking amp which can take abuse, typical of musical instruments.
So my calculations , to be realistic, assume an 8 ohms speaker, since that's what's fitted at the factory, so I arrive at a *real* 50 W rating (which would mean +/-35V rails), but I also consider that it can sink to 6 ohms at certain frequencies, so I choose the rail voltage accordingly.
From the graph it's between 32 and 33V.
You rounded up, I rounded down. No big deal.
The chart shows real measurements, no doubt, but at the lab they use regulated power supplies to draw them.
No such luxury in a commercial amp.
And I find very difficult that anybody pulls 68W regularly out of them.
Even the factory covers itself against lawsuits and says "68W typical" ... "60 W min".
And for 8 ohms loads they only specify Typical, not Min.

As a side note, I think the idea is not to build an amp which lives up to the 65W claim (which the original one maybe did, but in a horribly unreliable way) but a trusty amp , which can be used forever.
As an example: I have been designing and building amplifiers for 43 years now, lost count but over 10000 of them (not a typo) ... and they keep going 10, 20 and even 30 years later.
Short ago I posted here the pictures of a 200W amp, built around 1974 or 1975, still in continuous use at a Church every weekend.
It even has the original capacitors, probably because by receiving rated voltage every week they never degraded.
Go figure.
And every week I receive mails saying something similar to "I still have one of your amps, bought in 1989/1998/20 years ago/whatever , still going strong, now I want another one for my Son"
As you can see, designing conservatively has always paid , at least for me.
Good luck.
PS: there's another practical consideration: the original amp used 4 x TO220 devices.
Put them side by side, look at the surface area they cover.
Now put the LM3886 besides them: itīs area is close to 2 x TO220, no way it matches 4 of them.
To make matters worse, 4 discrete TO220 (or even 2 of them) can be separated along the heat sink for better power distribution, while those "2 x TO220 equivalent" which mean a single LM3886 (or TDA7294 for that matter) can not, they sit together heating each other, so any safety downrating I apply there surely won't hurt.
jm2c

[chazpope]

yes I know what you mean, back in my teens in '77 I built a '100W' amp and it has 4x2N3055s each on a brick size finned heatsink. No such luxury here and the budget is blown so I have provided supply for a little old fan to help with cooling.

I tried the psu under load (transformer is still not interlaced) - 1.5A takes it down to 30V. Next I have to figure out if this LM3886 I have lying around is in order ..

[J M Fahey]

OK, go ahead.
You are not really abusing it, just getting close to the top.
A small fan, any type, (even those tiny clip-on-the-microprocessor types) work wonders.
Your amp will be fine and sound exactly like the original one.
Post your work when finished.

[chazpope]

Well - I got it playing again, hopefully for longer this time, haven't tested it much cause it was midnight.

This is how I did it with the fan out of the way, its power comes from separate coil and is using one of the unfortunate original bias diodes as rectifier. The LM3886TF (had to buy one) is hooked to the PCB so the headphones jack works as well



And here it is spinning happily, rail voltage seems to be between 37 and 32V depending on volume

[J M Fahey]

Cool !!!
Congratulations !!!
Why the TF version?
It dissipates less than the regular one, plus mica and grease.
Of course, having the CPU cooler on top must definitely help.
I'm making a "flat" power amp to be mounted on the back panel of a speaker cabinet where space (available thickness) is *VERY* limited and am also using 2 thin CPU coolers.

[chazpope]

it was the least hassle and I did not have the mica thing - anyway we played it for 2.5h at the jam and all was ok - hope to not see it again - thanks all who helped