Need help picking out a SS amp to build, need LOUD!

[Zben3129]

I need to make an amp to keep at my friends house where we jam sometimes. Its not often enough to justify keeping my valve junior or firefly there, and I don't want to spend the money to make another tube amp. I do need quite loud though, as we like to play loud there, noone cares 

Anyone have suggestions for an amp to make? Need more than 30 watts

I was looking at the lm3886 amp at GGG, 68w of power, but I couldn't find the lm3886 at mouser. I am assuming it is obsolete now, does anyone know the replacement / another chip that could give 68w+ of power?

Also, on that amp, it is set up as a poweramp, correct? If so I could just add a preamp of choice in front of it? And does it provide 68w clean with +/- 18v?

Zach

[suprleed]

Although I've never built anything with them I believe some of the TDA20xx series of chips offer pretty high wattage output.  I would try a few google searches on this.   GGG may even have a few projects using these.  You could also wander over to www.ssguitar.com and do some searches there for high wattage power amps as that's their specialty.   

[davent]

Zach, Digikey have that chip you're looking for.

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=LM3886T-ND

dave

[R.G.]

The LM3886 is by no means obsolete.

As a practical matter, it will put out about 40-50W into 8 ohms, the advertised "68W" into 4 ohms; both require care in designing the power supply to not exceed the limits of the LM3886 and not to get the chip too hot.

Two LM3886's can be set up in a push-pull configuration with the speaker between them to get 100W; with two LM3886s in parallel on each side of the push-pull, you can get 200W.

However, I'm not a fan of building enormous power amplifiers. I like stacking blocks. One LM3886 can easily drive 30-40W into 8 ohms. So make the power supply and LM3886 part of the speaker cab. Now make another one; you have 60-80W. The next one gets you to 90-120W. But there's no need to stop. Just make a speaker cab with speaker, amp and power supply in it and run a line level cable to it.

In this sense, amplifiers are like nuclear weapons. There is an inverse-cube law affecting loudness just like the inverse-cube relationship of damage on the ground from a nuclear weapon. Two or three 10MT warheads properly spaced will do as much damage as one 100MT, and are much easier to handle and maintain. For power amps, two amps spaced out a bit will give equal loudness at each place as one bigger but remote amplifier. If you had one 30W amp aimed at each member of your band, they'd probably be begging you to turn it down.

Don't forget speakers in your quest for loudness. Speakers vary a lot in how much sound they make from the electricity you put into them. Compare them with the output loudness specification, usually given as db measured with 1W drive at 1meter from the speaker. 6db more loudness rating is the same as doubling your amplifier power.

[Zben3129]

Thanks for the info,

are there any other chips besides the 3886 that would work as a 30ish watt poweramp? As I can't find the 3886 at mouser and I don't feel like making a seperate order somewhere else.

Zach

[R.G.]

are there any other chips besides the 3886 that would work as a 30ish watt poweramp? As I can't find the 3886 at mouser and I don't feel like making a seperate order somewhere else.

OK, so it needs to be LOUD, and also available at Mouser.   

A quick scan of Mouser's parts hunting page turns up only a 50Wx4 channel car audio chip for $21.00, a 45Wx2  into 2 ohms car audio chip, and some TI digital power amp chips for up to over 300W. The difficulty in getting something like this last to run is like building a computer motherboard kit. 

Mouser is simply not cooperating.

By the way, the power amp itself is the simplest, smallest part of your problem. A power amplifier is best thought of as a DC power supply that has some trivial other junk (that is, the amplifier) stuck onto it almost as an afterthought to let some of the power out in a controlled manner. The power supply is where all the money and mechanics is spent.

[Zben3129]

are there any other chips besides the 3886 that would work as a 30ish watt poweramp? As I can't find the 3886 at mouser and I don't feel like making a seperate order somewhere else.

The power supply is where all the money and mechanics is spent.

That is what I was about to ask about, the power supply. I found out about 20 minutes ago that I apparently know nothing about power supplies.

I took a 23-0-23 vAC transformer, hooked it up to 4 diodes (full wave rectifier) and read the DC output. I got about 43vdc (I didn't use the CT, just the ends).

Then, I tried putting a cap across the DC for filtering (or whatever it does, I see it in every power supply),

+ of capacitor the + of DC, - of cap to ground. 100uf 50v.

When I powered up, I released the mysterious blue smoke. Good as a firecracker (LOUD!!!), bad for electronics. Oh and bad for air freshener 



I guess what I am asking is how do I make a +/- 18vdc power supply 


From my knowledge, I assumed this meant a 18-0-18 transformer, 4 1n4007 diodes (or 1n4004), a high capacitance 50v capacitor, and possibly a 2-3w resistor to drop voltage if neccesary.


Is any of that right? Is there a schematic I could look at? Does it need more filtering? Can you suggest a transformer?


^See I told you I knew nothing 



Thanks

Zach

[davent]

Hi Zach,
This should/might provide some ideas.

http://diyaudioprojects.com/Chip/LM3886_CA/LM3886_CA.htm

Take care,
dave

[slideman82]

Try in electronic-labs.com, there's a 100W (I can't really believe it) in 8 or 4 ohms power amp, with power Darlingtons, there's the PCB desing too! Needs 30-0-30 V AC 3A I think, and I've read it sounds good. But, you'll need a preamp...

[R.G.]

Is there a schematic I could look at? Does it need more filtering? Can you suggest a transformer?

First, go read GEO. In this case, "Power Supplies Basics", where you'll find out why your 50V cap died.

Then come back with questions.

What you need to do is
(a) find out what power supply you need to supply the chip or power amp you use.
(b) design the power supply to fit that need
(c) change the design of the amplifier when you find power supply components that are near but not over the voltages you needed in b.

For the LM3886, this is easy. You want a +/- 30-32Vdc power supply at about 2-3A per power chip. That's a 42 to 44V center tapped power transformer rated at about 3-4A RMS. You'll need about 5000uF/50V for each side of that power supply to ground, and at least 100V/5A diodes to rectify it. The ruts are well plowed for the 3886.

A 3886 costs $5. A 42Vct 100VA transformer costs $25 to $50 if you can't find it surplus. Surplus the transformer will still be $10 or so unless you get very, very lucky - and there is a horde of 3886 builders out there looking for surplus power transformers for them.

[Zben3129]

Is there a schematic I could look at? Does it need more filtering? Can you suggest a transformer?

First, go read GEO. In this case, "Power Supplies Basics", where you'll find out why your 50V cap died.

Then come back with questions.

Man I knew I remembered an article on PS there but couldn't find it! Found it now with the search thing and read it over.

What you need to do is
(a) find out what power supply you need to supply the chip or power amp you use.

I am going with the TDA2040 for cost/availability reasons. This requires an 18vdc bipolar source (+9v, -9v) and draws 4A max (I believe, on the datasheet is says "output peak current: max 4 A. It says it is internally limited, does this mean it will "stop"[shut down?] once draw is 4a to prevent damage?)

(b) design the power supply to fit that need

Math time:
Vsecondary(peak) = (Voutdc +Vripple + 2*Vdiode)
Vsecondary(peak) = (18vdc + 1vdc + 2*(.6v))    ****I am still confused about ripple voltage
Vsecondary(peak) = 20.2v


Vsecondary(rms) = Vsecondary(peak)/1.414
Vsecondary(rms) = 20.2v/1.414
Vsecondary(rms) = 14.3v

Now, if I were to use a regulator,
Vsecondary(rms) = Vsecondary(peak)/1.414
Vsecondary(rms) = (20.2v + 2v)/1.414
Vsecondary(rms) = 15.7v

So, does the above math tell me that if I wanted a regulated 18vdc power supply, that I would need to have a transformer that has a secondary of 16 - 0 -16 approximately?
And what about current? Does it need to be able to handle 4.5A?

http://www.mouser.com/Search/Refine.aspx?N=254453+4294608113+4294608467+4294608276+4294608449&Ns=P_SField%7c0

Is that the kind of transformer I am looking for (some stuff from that list maybe?)

I am assuming I need to look for a "36v CT @ 5A, 18v @ 10A"

Or is it "36vCT @ 2.5A, 18v @ 5A"
[/quote]

As for the caps, I am still unclear on choosing a value, voltage rating, etc.

It looks to me like 1n4007 diodes would pass for the bridge

I am still looking for a regulator


Thanks

Zach

[Nasse]

I dont know how much it does cost today but when I did build 3886 amp I got 35 amps diode bridge quite cheap. On another 3886 kit amp 3 A diodes were supplied with the kit.

+1 for the speaker efficiency. They are not all the same, even if they do look similar. Wattage is not so important than efficiency. And remember to filter out lows in the amp/preamp design. 80 Hz is very low for guitar

[R.G.]

I am going with the TDA2040 for cost/availability reasons. This requires an 18vdc bipolar source (+9v, -9v) and draws 4A max (I believe, on the datasheet is says "output peak current: max 4 A. It says it is internally limited, does this mean it will "stop"[shut down?] once draw is 4a to prevent damage?)

Actually, the datasheet shows +/- 18V, total of 36Vdc, +18 and -18. Let's do the math.

Vsecondary(peak) = (Voutdc +Vripple + 2*Vdiode)
Vsecondary(peak) = (36vdc + 1vdc + 2*(.6v))   
Vsecondary(peak) = 38.2V
Ripple happens because the rectifiers only provide pulses of current near the peak value of the AC voltage on the transformer secondary. This charges the caps up to Vpeak minus one diode drop. The capacitors then supply the DC to the circuit until the next charging pulse. The cap voltage runs down as this happens, so the run-down on the cap voltage from one AC peak to the next is the ripple voltage. If the circuit is using I amperes, then Vripple = I*dt/C, where dt is the time between charging pulses. You can set this to anything you want to pay for by buying bigger capacitors; C is in the denominator. If you cheap out on caps, the ripple voltage gets big. Put in huge ones and the ripple gets small. One common rule of thumb is 1000uF per ampere. I prefer rather more than that.

We need to know I to calculate C. We need to know how much power is going out of the power supply at full warp drive. Looking at the charts in the datasheet, Figure 1 shows output power as 27.5W at 4 ohms load, +/- 18V supply. However, the chip is also getting hot, using some power. Figure 9 shows chip dissipation as 12W near that output power and load. So the power supply has to provide 39.5W (about) that gets eaten up as heat.

With 36V available (18+18V) the power out of the power supply is at least 39.5/36 = 1.1A on average, at max power out.  Now we can calculate caps.

For 2% ripple, Vripple per cap is Vripple = 18*0.02 = 1.1A*8.3mS/C. And C = 1.1A*8.3mS/.18V = 51mF or 51kuF. Ouch. That's huge. So we can't afford 2% ripple. Let's settle for 20% ripple, which is 5000uF. Not perfect, but affordable. (Trust me here, the results will be OK in the end.) Looking at Mouser, there are multiple 4700uF/50V caps for $2.50 -$3.50. Assume you'll get them for $2.75 each, total of $5.50.

So we need 36Vdc, plus two diode drops for the bridge. What about that Vripple thing? That's only needed if you'll be regulating from the raw DC. In the case of power amps, DC is almost never regulated. It's possible for this small amp, but I suggest you ignore it. Besides, we're tolerating 20% ripple, so the no-load DC would be higher, and that endangers the chip by making the voltage too high for it at no load.

We need a transformer that produces 37.2V peak. Doing the peak-to-RMS thing, we get Vrms = 37.2/1.414 = 26.3Vrms.

Here the real world intrudes. If you want to find a 26.3Vrms transformer, you will look long and hard and still maybe not find one. But the world is practically littered with 24Vac transformers and there are quite a few 25vac and 25.6Vac transformers. I suggest you grab a 24 to 25Vac transformer and use it. That gives us a voltage of 24*1.414 - 1.2 = 32.7Vdc, and a power output of 22W into 4 ohms. You will not be able to hear the difference between 22W and 27W, and you'll actually be able to build it because you can find a transformer. Notice that +/- 16V, which is what a 24vac transformer gives you, is the recommended operating condition they list on the spec sheet for the normal operating conditions.

The peak current out of the amplifier is 16V into a 4 ohm speaker or 4A. The chip only provides this output at the peaks of the output signal. Actually, it's somewhat less because the chip internally drops a volt or so in its output transistors.

If you're buying a retail transformer from Mouser, this transformer will work:
553-F107Z (24Vac/48VA) chassis mount $17
553-F192X (24Vac/2A) chassis mount, $17

There are others at higher cost.

Notice that in buying a transformer, the transformer sees an RMS current larger than the DC output current because it's supply in big pulses of current, not steady current. The RMS heating value of the pulses in the transformer may be 1.6 to 1.8 times the DC output. So for 1.1A out, you could need 1.7 to 2A rating for the transformer. There is even more to this than I can put in here, but for constant high power use, go with the 2A rating.

The advice on diode bridge assemblies is good - they're about $2.50 for a 20A or more bridge. More is better here.

Do NOT scrimp on the AC power side. Use an IEC power inlet with fuse (about $4) and a decent power switch breaking both sides of the AC power line ($2).

So the power supply side of things is up to about $32.00. As an aside, getting under $1/watt is average for power supplies unless you can buy in high quantity or get surplus parts. This is a good number so far. The TDA2040 is $3.50. Then you have to do something to get 12W of heat out of the chip (you need a heat sink) and something to hold it all (i.e. a metal box; do NOT try to put this in a wooden box or leave it open to enquiring hands, pets, and beers.)

How're we doing so far?

[Mark Hammer]

You need loud.

Wattage is what the amplifier is potentially capable of delivering in terms of current to a load of a given impedance.  If you stuck a 100W 8R resistor on the output of an LM3886-based amp it would be plenty quiet.  Why?  No speaker.  So let's shift focus to the speaker here.

Speakers vary in their "efficiency".  That is, the sound pressure level they are capable of producing with a given output signal from an amplifier.  Typically, bigger speaker cone area = more air movement = greater SPL, so a 12" speaker will be "louder" at any given wattage than the same voice coil on a 10" speaker, 8" speaker, etc.  But even so, some speakers of the same impedance and surface area are more efficient than others and deliver greater SPL for the same wattage.  Often, though not always, this is a byproduct of the magnet size and the gap that the coil sits in.  You can generally always see speakers rated in terms of "sensitivity" or efficiency, in terms of db per1W input power at one meter's distance.  If you look on this sample chart - http://www.eminence.com/guitar.asp?speaker_size=12 - you can see a bunch of 12" speakers from the same manufacturer that range from producing 98.8db SPL to 103.5db SPL at one watt.  That may not seem like much, but consider that 3db = double the loudness, and one of those is about 3 times louder than the other.

So, you want to keep an eye out for efficiency/sensitivity so as to milk the most out of your 30W or whatever.

Efficiency is not only a function of the speaker, but of the enclosure the speaker is in.  Some enclosures have a way of conserving most of the energy the speaker produces, and some have a way of losing a lot of it.  What you want to do is to stick that speaker in a cabinet that squeezes the most volume out of it.  Closed back or variations of ports often helps in that regard.  Certainly you don't want too small a cabinet.

Picking the right cab and speaker can make your seemingly modest 30W solid state amp every bit as loud or louder than something that seems to be more powerful...on paper.  Doesn't hurt top perch the amp on a stool or stand either.  Your bandmates are listening with their ears, not their knees.