Want to build a SS PA

[Steve Conner]

Yeah, think about it dude. There's no winding on a Champ PT that would supply a SS power amp.

I've just done this: built a Champ-style amp with a solid-state "afterburner" on it. The tube power and output transformers are exposed, but there's a second toroidal PT of 50VA or so, hidden inside the chassis, that provides +/-35V to the solid-state section.

I used a discrete power amp with a MJL3281/MJL1302 pair. Massive overkill for 30 watts, I know, but they were sitting there in my junk collection waiting for a use. And that overkill means they should be bombproof: my output stage has inherent current limiting, but no VI limiter type thing that would keep an undersized transistor inside its safe operating area.

I use my own secret current-drive design, basically a modern version of RG's beloved Thomas Vox amps. But Rod Elliott has a number of simple SS PA designs on his site. You might also want to try on the ssguitar forum.

You don't NEED a PCB to use the LM3886. I've seen people solder all of the required components (there are only about half a dozen, after all) straight to the chip's legs. It's not pretty, but it seems to work fine. You could probably use a turret board, for authentic hand-wired mojo.

National also make a thing which is a "LM3886 minus the power stage", for driving your own discrete outputs. The LM4702 and LME49810. But it's not really THAT hard to make a discrete driver, either.

RG: If you're reading this, I got some of those new ThermalTrak power transistors you told me about, with the 5 legs. Still looking for a use for them

[lowell]

Ok i see where you guys are going with this. This is also a pt for the tweed deluxe. That means it can safely handle about 18watts. It's obvious now that I'll need another pt for the 50watt solid state power amp.

[Steve Conner]

Yes, you will...

I recommend a 2 x 24V or 48V CT at about 75VA. Should be able to get a toroid PT of this rating for about 20 bucks.

The toroidal ones aren't mandatory, but they're smaller and easier to hide inside an amp that's supposed to be "tube".

I have a collection of PTs that I pulled from old solid-state amps, whenever I need one I just pick the nearest voltage. Except for this amp none of them was quite right, so I had to buy one from Farnell.

[J M Fahey]

Dear lowell. I do not want to give you the "cooked and ready to eat" answer, because it *may* help you on *this* problem but not on the next 1000.
I'd rather guide/help you, with my best intentions, to solve it yourself.
No,

This is also a pt for the tweed deluxe. That means it can safely handle about 18watts.

is not the point.
You might use a 50 pound, 300W SVT transformer and it would not do, either.
that's why I'll repeat my question, and add a couple more; all 3 point to the same:
1) Check in the datasheet and/or any of the 3886 projects the supply requirements, obviously when driving a speaker to rated power.
2) Post side by side what the windings in a Champ/Deluxe transformer provide/supply (Volts and Amps/mA) and those in one suggested for an LM3886. Just the secondaries; primaries will obviously be correct for your wall voltage.
3) Calculate how much current can come from a 300V supply, through a 220K series resistor.
To simplify calculations, assume the load is shorted, drawing as much current as the supply will furnish.
Thanks.
PS: otherwise, I fear this topic will reach 120 posts, as others did, with about 80 of them being the same, still unanswered question and the unrelated answers.

[R.G.]

RG: If you're reading this, I got some of those new ThermalTrak power transistors you told me about, with the 5 legs. Still looking for a use for them

Yeah, I have lots of stuff like that.

I keep intending to try a pair of those in a discrete power amp. I keep remembering Ben Duncan's comment in his book on power amps that "... the world has pretty much given up waiting for power transistor manufacturers to market devices with internal thermal sensing diodes" or something close to that.

I very much liked the detailed sendup of thermal compensation dynamics in Doug Self's book as well. He points out at some length that the thing which keeps us from keeping as good a bias point as possible is the thermal time lag between dissipation and sensing the chip temperature change. The junctions on the transistor are the obvious way to do this, and one junction per transistor give you the best possible thermal sensing and compensation.

It is possible that it's the thermal lag of compensation in the output stage that is what the Golden Ears crowd are really hearing when they complain about this or that amplifier having better or worse soundstage or separation between notes.

Maybe.

But this is only really applicable to the double emitter follower output stage, where the output transistor temperature is what matters most. In the Sziklai pair, the temperature of the output device matters little, it's the driver temperature that needs compensated.

I keep intending to use a pair of those next time I build a discrete amplifier.

[Enzo]

Just a thought experiment. If a 300v winding could provide 2.5 AMPS of current, how many watts would that transformer have to handle? And does it seem likely a Champ transformer could do that?

[lowell]

JM,
I'll be getting back to you in my next post... I promise.

Enzo,
750watts... uh yeah, I see your point.

[lowell]

"1) Check in the datasheet and/or any of the 3886 projects the supply requirements, obviously when driving a speaker to rated power.

the Po Symbol says:
IV+1=IV-1=35V, Rl=8ohms AND that gives 50W of power.
Not sure if that 35V is nominal supply voltage or if it means + and - 35V for a nominal of 70V. If NOT, then our +/-18v will be similar to these stats.

I+ says quiescent power supply current is 50ma. Not sure if that's what you're asking here.


"2) Post side by side what the windings in a Champ/Deluxe transformer provide/supply (Volts and Amps/mA) and those in one suggested for an LM3886. Just the secondaries; primaries will obviously be correct for your wall voltage."

I can't find a VA rating on a Champ PT... however with the tweed Champ circuit I'm thinking that a VA rating of 15W is about right.
6V6 at 14W 90% dissipation=12.6w
add about 3 watts for heaters, preamp tube plates, and pilot light

LM3886 is a 50watt amp at 50ma with the above specified supply voltages and 8ohm load. I know now thanks to help from Enzo in a previous post about mains fuse values, that watts out = watts in. If the LM3886 amp provides 50watts of power then the power transformer has to be able to supply 50watts of power on its primary AND secondary.

"3) Calculate how much current can come from a 300V supply, through a 220K series resistor.
To simplify calculations, assume the load is shorted, drawing as much current as the supply will furnish.
Thanks."
300/220k=1.4ma (JUUUSST a bit outside)

I hope I've answered your question JM. I'm sure I'm not being as thorough as you want, but I'm not really sure what I'm missing if so.

[Steve Conner]

50mA is the idle current.

The full load current is the idle current plus whatever an 8 ohm speaker needs to produce 50 watts: let's do the math.

P=I^2 * R therefore I = sqrt(P/R) = sqrt(50/8) = 2.5 amps RMS delivered to speaker

2.5A RMS is 3.53A peak, 2.15A average when rectified. We have to divide that average current between the two rails, so each rail has to supply 1.07 amp. Add on the idle current and we have about 1.1 amp.

Sanity check: 1.1 amp * 35V * 2 rails = 77 watts input power. 50/77 = 65% efficiency which is about right for a Class-AB amp at full power.

And +/-35V is 70V total.

[lowell]

Right on. Thanks Steve. Alright, I'm gonna be lazy and impatient at this point and ask if anyone has an LM3886 pcb that I can purchase. I'm gonna use a Princeton 65 power supply for this amp... I got a dead one in the back.

[R.G.]

R I'm gonna use a Princeton 65 power supply for this amp... I got a dead one in the back.

That's a problem. While the LM3886 will work with a +/-42V power supply like that, it will almost certainly be going into thermal shutdown a lot at any kind of higher power. That's because the higher voltage will make it dissipate more power internally for the power it puts out.

And it's in danger of overvolting. The LM3886 is rated for 94Vdc max. If the AC power line gets 10% high, you're there.

[lowell]

I was going to mod the supply a bit to be +-35v. Can't I just slap some 35v zeners on that supply node?

[R.G.]

I was going to mod the supply a bit to be +-35v. Can't I just slap some 35v zeners on that supply node?

Well, not exactly. That would result in massively overheated zeners, or rectifiers, or power transformer or all of the above.

A zener fed more than its breakover voltage lets through a theoretically unlimited current. What limits the current is that the zener burns up if there is no external current limit. So all zeners are always used in a way where the current through them is limited by some other part of the circuit. In power supplies this is usually a resistor. The resistor is calculated so that for the difference between the raw voltage and the zener voltage, the resistor lets through the max load current. When the load current decreases, the zener eats the current the load doesn't want. If the load goes down to zero amps, the zener has to eat it all. For a power amp, the currents are maybe 3-5A, and the load needs almost nothing when there's no signal. So the zener has to be able to dissipate 3A times 35V, or over 100W. The other zener would dissipate the other 100W. 8-|

If you have a princeton 65 that's a hulk, you have the hard part of a PA amp done. Use the power supply and re-convert the power amp from that d&#@*$(#@ silly design back to a stock discrete design. It's pretty simple to cut out all the nonsense and make a stock amp out of it.

Power amps are primarily **power supplies** and heat sinks. The power amp itself is a trivial wart on the power and thermal parts.

[J M Fahey]

You are half way there, PCBs are available, google for them, ot try to get "chipamps" to sell you just one mono PCB.
Worst case, if they sell a pair minimum, you already have another for your next project.
Transformer? : get one of the incredible offers at: miscellaneous.html
Any 40 to 48V C.T., >70VA will do, and mount it feeding the Princeton PSU.
Pull the old transformer and save it for a future discrete power amp.
Or get Rod Elliott's PCB for the 100W guitar amp and power it from the Princeton, as-is (+/-42V).
You do have many options, but match either the +/-42V PSU to a beefier discrete amplifier, *or* the 3686 to a smaller PSU, you can't have it both ways.

[lowell]

Ok I'm gonna go with making my own supply for the LM3886. It looks like the Chia Yu Co Ltd from the miscellaneous.html is the right one... correct?