Any discrete transistor based amps that are Ruby like?

[dennism]

I've built Rubys and the other 386 chip based amps.   I wondered if there is an amp similar in complexity that doesn't utilize any chips?  In the same 1/2 watt range.  It would make for a cooler turret or eyelet board build than a chip amp.   Links to schematics or layouts of any kind would be greatly appreciated.   Thanks in advance!

[ItZaLLgOOd]

I remember there being a DOD?? micro amp schematic around here some where but I believe it still used an IC with a pair of output transistors.

edit: here's what I was talking about

http://www.diystompboxes.com/smfforum/index.php?topic=75400.0

[MikeH]

Beavis had a project like that on his site, but he said it sounded awful.

edit: here it is: http://beavisaudio.com/projects/HalfWattClassA/

[CynicalMan]

It does sound awful, but in a good way.  

[WGTP]

Joe has a couple of interesting things.   

http://www.diystompboxes.com/analogalchemy/

[PRR]

> an amp similar in complexity that doesn't utilize any chips?

What do you want? A guitar-cord booster/fuzz? Or actually drive a speaker?

There's lots of simple fuzzes.

Driving a speaker is hard work. Doing it well means more than a few parts.

Sure you can build a discrete loudspeaker amp. People used to do it. I've done it by stripping mil-surplus logic boards for transistors and caps, flea-clipped on perfboard.

The whole point of the LM386 is that you can not come close to that result with so few discrete parts. Once it came out we stopped building part-Watt amps.

Here's a part-Watt loudspeaker amp. It is very loosely related to LM386 topology.



Sim says 1/4W in 10 (or 8) ohms at 15% THD, more or less, give or take. That's "pretty poor" by most standards. However improvements take more parts. And the bentness of this marginal design may be euphonic.

The 2N3904/6 parts are awful marginal for the output stage. You may burn a few in heavy playing. A shorted speaker will probably pop one. Since these are 20 cent parts, and you wanted simple, I'm not inclined to "improve" its robustness. But if you have the high-gain high-current parts like 2N4401 and its complement, this is the place to use them.

C4 is not necessary if the power supply is very-very clean. IMHO you probably want it.

R4 may not be "essential". It helps protect Q2 in a dead-short, but only after Q4 has failed. Since both are the same part-type, they could live or die together. R4 also "improves" linearity, which is why you may want to skip it.

Overall gain is set by R9. For guitar into "loud"speaker it will want to be around 100 ohms. You can short it out for maximum gain and nastiness.

Input impedance at high gain is quite low, under 100K. That's what you get with just three gain stages.

C1 will probably be a film-cap but I marked "+" in case you wanna use a small (~~2uFd) electrolytic.

Diodes are any Silicon diode or rectifier. 1N4148, 1N914, 1N4007.....

[slacker]

That looks interesting Paul, I'll have to give it a try.

I built this using 2N3904s and 2N3906s http://schematicwiring.com/simple-electronic-applications/the-circuit-schematic-diagram-of-simple-500-mw-3-transistors-audio-amplifier/ which to me looks like a similar thing. With a pedal in front of it to provide some boost it sounded Ok, like Paul mentioned though the output transistors got pretty hot.

There's some other simple amps here http://www.rason.org/Projects/transaud/transaud.htm I haven't tried any of them though and I think they would need a booster in front of them for guitar use. Dano built the mosfet one and said it was Ok http://www.beavisaudio.com/

[dennism]

I appreciate all the replies.   Yes, I was looking to drive a speaker and make it into the usual thing that I've done with 386 amps like cigar box amps.  However, I may have underestimated just how much complexity is already packed into that little chip.   Sounds like it might be more trouble than it's worth to try to avoid the 386.   I like the way the 386 amps sound, I just wanted to make a cooler looking point to point style build and the chip kind of ruins that look.   Perhaps I should rethink what I was planning on doing and just find a way to build the circuit where the chip is more hidden.   Anyway, thanks to all who replied.

[PRR]

> other simple amps

Many ways to build power amps. Not many ways to do a good "simple" power amp. And coming up from guitar-level is a long way to go.

> I think they would need a booster in front of them for guitar use.

Estimate the current gain required: the ratio of output current to input current.

Using round numbers: For guitar to speaker we must transform from 100K at the input to 10 ohms at the output. At the same voltage, this implies 100,000:10 or 10,000:1 current gain.

For guitar to speaker we actually need a voltage gain of 100 or so. The higher output voltage means more output current.

So 10,000 * 100 = 1,000,000:1 current gain.

A single BJT transistor gives hFE current gain of roughly 100. We can pick devices with higher hFe; we also can't use all the current-gain for the audio, some must be used to set up DC bias conditions.

So we need 100*100*100=1,000,000 or three sequential stages to get the current gain.

A resistor-loaded power amplifier has 6% efficiency. For the suggested 1/2W output this leads to 8 Watts of heat! This may be distressing. A push-pull output stage can give efficiency over 50%, much less than a Watt of heat to handle, at the edge of what "small" (TO92/plastic) devices can manage.

> I built this

The schematicwiring.com plan has push-pull but only two stages of current gain. The input impedance is shockingly low (well under 1K) and the voltage gain is small (maybe 10; depends on the source). It probably works great from a headphone or iPod source. It won't be any use with passive guitar or most pedals. You probably used a fairly strong (low impedance output) booster.

Selmer SolidState Mercury 5w

This amp had just three transistors (and the '3053/'3055 pair could be replaced with a TIP120). It does have an awful lot of resistors. It was surely the cheapest hunk of junk they could sell (a "beginner's amp", something to grow out of real fast). It does add an audio transformer (half as good as push-pull). We don't have the B+ voltage or OT impedance, probably 30V and 50-100 ohms.

> Dano built the mosfet one

MOSFET does not have the leaky input of a BJT. You could say that input impedance is "infinite", but that's not true for audio. The input is a capacitance; IRF510 is pretty small capacitance. Voltage gain for a single stage at low voltage is low: maybe 4. That particular plan is not optimized for high input impedance (another iPod booster): add a 1Meg resistor between R1 and MOSFET Gate, it may play better. Cute, but not fabulous.

> how much complexity is already packed into that little chip.

There's obviously 11 transistors; actually more. You could do it with less. But some of the "extras" actually reduce the external part-count (notably the near-ground inputs which need no input coupling cap) or the size/cost of some external part (no giant cap on the NFB).

Count the legs. Several to a dozen 3-leg transistors, 10 or 20 2-leg parts, there's 40-50 legs to solder for a guitar-to-speaker amp. The brilliance of the LM380/LM386 is that someone did that "once", on a sheet of plastic, and then duplicated it on Silicon millions of times. Result is that you only have to mess with 5 or 7 legs to whip a speaker.

[PRR]

me> you only have to mess with 5 or 7 legs to whip a speaker.

I can't count. There's 15-20 legs to solder, '386 plus the passive parts.

That's roughly 1/3rd of any discrete amp which will raise a guitar to a speaker.

Can you stand three times as much work? When there was no alternative, we did it. Hobbyists still build far more complex all-discrete audio amps.

[mac]

I posted this some ago, schem link is dead but it is in my gallery now,



mac

[brett]

Hi
You can find Class AB designs on the web that will drive (push-pull) an effecient speaker through a pair of 2N3904/2N3906 (or BD139/BD140 or TIPs if more volts are used and more power is needed).  You can get about 0.2W clean from 9V or 0.4W from 12V (=Ruby's power).
Here's a complete amp schematic.  Delete Q1 and Q2 for a "power" amp. http://www.ke3ij.com/amp.htm.  There are dozens of similar designs.  Most are good.
cheers

[Brymus]

Would there be any benifet in replacing the two diodes common to these designs with a 2n3904 configured as a diode amp ?
Would that increase the efficiency or fidelity any or is that just not a good place for a diode amp?

[PRR]

> replacing the two diodes

Audio Power Amplifier Design Handbook
Douglas Self
Chapter 15
46 pages.

All of Doug Self's books are EXCELLENT.

However, in "this context", K.I.S.S. Do the diodes. It works.

[brett]

Hi again
+1 for diode biasing of the output transistors. It's simple and works well.
If you are using > 12 volts and going for > 1W , mount the diodes on the transistor cases.  That'll reduce overheating (due to thermal runaway).
These simple amps can sound quite good and cost next to nothing.  A DC wall wart makes a good power supply, as batteries don't last long.
good luck

[Joe Kramer]

Wouldn't the way to go with a simple discrete power amp be to use a small output transformer, like the output sections in old portable solid state tape players, etc.?  Mouser has transformers that would work. . . .

[brett]

Hi

small output transformer

That's a good, but somewhat complicated idea.  To match the output impedance of 4 and 8 ohm speakers, you's want something like a 48:8 ohm OT.  They aren't easy to find.  Driver transformers (e.g. 1K:1K) are easy to find.
The best small OTs can be wound yourself by dismantling a wall wart and re-winding.  For a 48 ohm primary and a roll-off frequency of 160 Hz, you'll need 48 milliHenries of inductance (easy to remember, huh? 48 = 48).  For a 3/4 inch x 3/4 inch core, you'll need about 200 turns for the primary.   (In many cases you can use the old secondary as the new primary, because it's designed to be efficient at a low frequency (60 Hz)).  For 200 turns on the primary, you'll need 80 turns on the secondary.  The primary wire can be thinner than the secondary.  The DC resistance of the primary should be less than about 5 ohms, and for the secondary should be less than 1 ohm (ie less than 10% of the AC resistance).

A one stage I put a photo tutorial on the net showing how to re-wind a wall wart.  Have a search, it might still be there.
cheers

[PRR]

> use a small output transformer

What Brett said, only I been pounding nails while thinking.

> the way to go...

DEFINE "the way to go". What goals? Simplicity? Economy? Power? Size? LM386 wins all of these easily, so "discrete" has to be "a goal". And you should justify your goals, at least quantitatively. How much more work/cost/size/hassle will you allow to meet this "discrete" goal?

Look at the relationships of speaker impedance, battery oomph, transistor limits.

When you can't make it work directly, use a transformer.

> old portable solid state tape players, etc.?

A transistor radio from the 1960s would be scaled for 100mW output; the wheezy batteries of the day had too-short run-time if power was higher.

Taking 50% efficiency (78% is the ideal we never reach), that's 200mW from the battery. At 9V that is 22mA. The output stage should load the battery like a 9V/22mA= 409 ohm resistance.

A totem-pole output stage reflects the audio load out to the supply as about 6 times higher. 8 ohm load, the output stage acts like 50 ohms to the supply.

So a totem-pole loaded in 8 ohms sucks 9V/50= 180mA. Which is too much for older 9V batteries.

It actually works out if we had a 65 ohm speaker. But the voice coil wire gauge gets annoyingly small, fragile, costly.

We could totem-pole into a 65:8 transformer. But we still need the output cap. And totem-pole is not especially simple. And usually needs both polarities of transistor.

The classic answer is a 800CT or 1KCT winding with two same-polarity devices. No output cap. simpler biasing, common emitter resistor. However you need two drives, complementary voltage, asymmetric current flow. This is most conveniently done with another transformer. Usually a 10K:2KCT.

Sure, it's a good plan. And you can get the parts. However you still need nearly the same number of transistors. Three to get from 0.5V 10K (AM radio detector) to speaker. Five to get from 5mV 50K (tape head) to speaker. Perhaps four from 30mV 100K (guitar). While a common-emitter stage has higher power gain than a common-collector (emitter follower), you need high impedance to use that gain. The driver impedance is limited by transformer parasitics. The output impedance is set by supply voltage and desired power, and is not real high.

When small transformers were very much cheaper than transistors, and particularly when 9V batteries were expensive yet lame, the transformer makes sense. Transformers were wound on kitchen tables by piece-working housewives; transistors had to be made in a foundry. So even very slight electronic advantage from a transformer would be justifiable.

Today a transistor is 1/10th the cost of a transformer. And batteries improved enough so that life was not a big problem even with the heavy load of a totem-pole into an 8 ohm load. (Also the expected output rose from 100mW to 250mW, and more for anything with a motor.)

And of course the transformers are not THAT expensive in DIY context. How much do 50-cent transformers cost today? $3? $7? You could buy a baggie of PN4401 for that price; or you could buy a fine beer and piss it away.... we DIY-ers don't fret about a buck like a mass-production operation has to.

The 3-transistor audio path used in transistor radios has about 30 legs to solder. But add the preamp needed for guitar, more like 45 legs. So the labor will be similar to the 4-transistor transformerless plans.

[Joe Kramer]

To match the output impedance of 4 and 8 ohm speakers, you's want something like a 48:8 ohm OT.  They aren't easy to find.

http://www.mouser.com/ProductDetail/Xicon/42TU048-RC/?qs=LQJGOuQCHKR6gBOQxy1apw%3d%3d

DEFINE "the way to go". What goals? Simplicity? Economy? Power? Size?

Sound?  Fun?  Interest?

[Boprikov]

Wouldn't the way to go with a simple discrete power amp be to use a small output transformer, like the output sections in old portable solid state tape players, etc.?  Mouser has transformers that would work. . . .

Legendary Pignose uses output transformer. There is rather clean schematic on a Russian guitar forum:

http://www.guitarplayer.ru/forum/index.php?topic=72219.0

I don't know if it´s any better than Ruby, but it sure is legendary.