The Mini-A

Number Two: He is exactly like you in every way...except one-eighth your size.
Doctor Evil: Breathtaking...I shall call him...Mini-A

Yes, the exciting, new, pocket-sized Aleph. Use it for your Walkman, use it in your car, you can even use it for a hand warmer! Approved by none other than that arch-fiend, Doctor Evil.
What Austin Powers thinks, we aren't sure. He's off testing his mojo with a fembot.
But seriously, folks...
A couple of months ago someone brought up the idea of a small amp. Now, yes, you could build a Zen, or you could build a Son of Zen, or you could build...oh, never mind, it was build a miniature Aleph or bust. I was okay until I realized that I actually could find a use for such a creature. (Really I was. No, really. This twitch? I always twitch like this...) So I started thinking about a self-powered, biamped speaker using a couple of nice drivers that I've got sitting on the shelf gathering dust.
And as I've said before, I'm happy as a clam until I start getting to thinking about <i>how</i> I might go about something. Once that happens, I start muttering to myself, scratching numbers on scraps of paper, wandering off to work without clothing on, that sort of absent-minded nonsense.
Blast it, the Mini-A grabbed hold of me.
I hereby release a Version 1.0 schematic of the Mini-A.

And now, a word from our sponsor...

The usual disclaimers apply:
1) This circuit is covered under US Patent number 5,710,522. For those of you bright enough to pound sand, you'll realize that this means that Nelson Pass owns this circuit just as surely as if he had penned it himself. If you build this critter for profit, seventy-seven lawyers will knock on your front door tomorrow morning and cause you more grief than you can shake a stick at. By the time they're done with you, you'll wish they still did things like they did in the good ol' days when they were content to visit a plague of locusts on your tomato plants, or sic a rabid Chihuahua on your ankle. Nope, we live in modern times now. There are new, high-tech ways of making you suffer.
So don't even think about it.
2) This is a first pass (no pun intended) at producing a little feller. I recalculated some of the obvious things, but left a few others alone. Suggestions for improving the circuit will follow. I've built a breadboard prototype. I tested it on the bench. There's no circuit board to go with the schematic. There's no warranty, expressed or implied. Your mileage may vary. In other words, I drew the schematic a month ago, finally got time to start building it yesterday afternoon, finished this afternoon, and have found that it works. I'm sure that there are a few more things that can be optimized, but it'll run as-is.

So why build a dinky version of an Aleph? Doesn't everyone want more power, not less? Aside from the powered speaker possibility I mentioned above, let me kinda rattle your cage with the following suggestions:
--The suggested rails are +-15V. It draws approximately 1A. Remember that transformer in the corner you've been trying to find a use for? The one that wasn't big enough for anything else? Smile, little orphan trannie, it's your day in the sun...
--Okay, so you plunked down the big bucks for a bag full of output MOSFETs for another project, and now you've got leftovers that you don't know what to do with. Yeah. You guessed it. It's showtime for those little puppies.
--Want to experiment with battery-powered amps? Two ordinary car batteries and you're set to go.
--Sick and tired of the device matching problem? It sometimes seems like it's harder to get a date for a MOSFET than it is for a human. The output devices for this amp DO NOT need to be matched. Cool, huh?
You get the idea.
All right, let's discuss the circuit.
It's a cut-down version of an Aleph 3. I arbitrarily chose 15V rails, but it will work nicely as long as you're within shouting distance of that voltage. 18V? No problem. 12V? No problem. Just keep in mind that it's a class A circuit and it will demand roughly an amp at idle, every day, all day. Don't try to run it off a wall wart. (Yes, I know it's tempting. Sorry to spoil your fun.)
So what's it put out?
About 5W without being refiddled. My calculations suggested that it should swing a little over 9V RMS (this would translate to about 10-11W into an 8 ohm load). In fact, it comes out almost exactly that as long as I use my oscilloscope for a load, but comes up a bit scant (around 6V) when I put an 8 ohm load resistor on it. A little more bias current would probably be a good idea. Go back and look through the threads, there's been plenty of discussion of bias in Alephs.
Other things you might consider doing include putting Zeners on the input to stop little lightning bolts from eating your front end's lunch (my little test rig here likes living dangerously--you ought to see him dressed up in his little black leather jacket, ready for a night on the town). You might be able to re-jigger the feedback ratio (R6 vs. R8) to optimize distortion (this will also change your closed loop gain, so be careful). You might wish to recalculate R13. Unless I've slipped a cog (not unlikely, actually, it's late and I'm tired), this sets the threshold at which the protection circuit kicks in. It may be right, it may not. I didn't really take the time to mess with it. And last, but not least, you could graft the whole Volksamp-Aleph front end onto the thing.
I trust you can handle the math to put an LED on the critter so that you will know that it's on.
If you've studied the Aleph circuits at all, this circuit should be pretty much self-explanatory. The idle voltages will all read about the same as the real Alephs:
--About 4 or 5 volts across R4
--9V across D1
--Roughly +4 volts at the sources of the front end differential, relative to ground
--About 4 or 5 volts across R3
--About 4 or 5 volts across Q4
Power supply?
What have you got on hand? Just stand back and throw some parts at it. It'll work.
I'm embarrassed to tell you what I've got on this thing. Let's just say that it's running fine on a small transformer (estimated to be 100VA--not even warm to the touch) with a junk box bridge rectifier, and some pretty small caps (30000 uF each rail). More would be better, and I've got some fairly hefty caps on hand (about this time it will belatedly occur to someone that they can use those monster 1 Farad caps like you put in cars for this circuit), but it isn't something I'm going to spend a lot more time on.
Why?
Shhh...
Listen...
Hear that?
I hear an Aleph-X calling. Gotta go.

Grey
 

Attachments

  • mini-a.zip
    21.5 KB · Views: 10,634
HI, GRollins

thanks for tyhe scaled down mini-A amp.... could be a nice Headphone Amp....

I have a general question about this topology:

When I design stuff with SPICE I always get lower distortion figures out of the 2nd leg of the LongTailPair.... why do people always use the first leg of the LTP for voltage gain?

This could very well be a very stupid question, please don't hate me for it ...

greetings,
Thijs
 
Thijs,
It's a question of phase, really. The signal comes in on the front end of the differential. As long as you're using a large value resistor or a current source, the signal will then be present at both of the plates/drains/collectors, but in opposite phase (note that if you intend to use the output from the backside of a differential, you're going to need a resistor for the gain device to work against). If you have no intention of using negative feedback, you are free to choose either output. See the schematic for Nelson's BOSOZ for an example of this. If, however, you have any intention of using NFB, you're pretty much locked into using an arrangement like you're describing.
Your goal is to provide the differential with a scaled-down replica of the output signal. Once you've done that, the differential does the rest, ignoring the parts that are the same (clean, pure output), and using the parts that are different (i.e. distortions of one sort or another) to try to tidy up the bad stuff that the amp did to the signal.
Sit down with a schematic that has a differential and trace the phases of the signal as it goes through various stages in the circuit. You don't need to worry about frequency, voltage levels, or volume levels. It's all a question of phase.

Grey
 
HI Grollins,

Thanks for the explanation! I understand I could use both leg's for the gain but the feed but must be changed to avoid positive feedback..

I've designed or build a multi stage amp containing a LongTailedPair, so it didn't cross my mind, but it it obvious to me now...

But still a intriging questions arises: how come the other leg of the LTP always gives significant better linearity (with slightly less gain)?

Can this feature be uses if you appply feedback trough a series resistor in the input ? (like the X-series of amp )

greetings,
thijs
 
janey,
I'd consider bumping up the bias a bit before trying to run 4 ohm speakers. With the voltage across the output devices being as low as it is, you can run fairly high bias currents and still be well within the SOA. Try 1.5-2A.
hifi,
On 15V rails, the circuit runs at about 30W Pd. A 30VA transformer would be kinda close to its ratings unless you wanted to back off on the bias and use it for headphones.
thijs,
The reason that the back side of a differential functions differently is that the signal is entering via the source/emitter/cathode rather than the gate/base/grid. The capacitance of the device is less of a problem, for one thing. Think of it as though the front of the differential is functioning as a follower, so you get lots of current, but not much voltage gain (in fact, you don't even get back the same amount of voltage that you put in). Then you're inserting the signal into the back device via its rump end, which just happens to be a perfect match for the output of the follower.
(If you really want to get picky, the front MOSFET is operating in a mode known as common drain, whereas the back MOSFET is being used in common gate mode. But the combination of the two of them in tandem is known as a differential.)
You can apply the feedback to either side of a differential (or both at the same time, for that matter) as long as you keep in mind that the phase must be proper for that stage of the amplifier. Elseways you end up with positive feedback and the result isn't pretty.
If you're looking for more ways to insert feedback, you can always insert it through the source/cathode/emitter, but this won't work with a differential; you'll need a stand-alone gain stage to do this. Tube folks do this all the time, but it's not quite as common in solid state circuits. Possibilities abound. Use your imagination.
The 'recipe' for R1 & R8 has been covered several times. Look back through the threads.

Grey
 
Vince,
Changing R18 & R19 to .33 ohms would get you in the ball park. Be prepared to adjust the value of R11 to fine-tune the bias a bit after that.
At this low a rail voltage, it's unlikely that you'll exceed the current ratings of any of the most likely output devices, but watch your device dissipation, as you'll be running around 30W through each MOSFET. Not undoable, of course, but you'll need decent heatsinks. The hotter you choose to run the devices, the more often you'll see failures. Leave yourself a little elbow room on the heat and you'll be fine.
Some minor adjustment of R12 might be needed to re-balance the output & current source, but it should run okay as-is.
I'm at work and don't have all my notes and such, but I'd think that R20 and 21 should also be decreased, probably to .33 ohms each.

Grey
 
Stole a few minutes time and fiddled with the Mini-A. Results as follows:
--Redid the current sensing resistor array, as it was just tagged together before. Still tagged together, for that matter, but better. Redressed the wiring harness. I'm now getting just shy of 10W output out of the circuit, which is closer to what I'd calculated, so this makes me happy.
--Put a 10 pF cap across R8 (the feedback resistor) to trim the high frequency response. 10kHz square wave ringing now reduced. Call this the Version 1.01 if it makes you happy.

Grey
 
Compensation Cap

I would put a 0.001uF compensation cap from the gate resistor of the output/voltage gain transistor to the output if I were you!
Go look at the Aleph 3 schematic and you will see. This a real good idea since the amp has no output zobel network and could get unstable with very capacitive speaker cables or other reative loads. Also for fun try leaving out C3(or lifting one end) to defeat
current boost for low impeadance load. You give up some output current and voltage swing but get rid of the feedback loop for the output current source. I did this on my hotrod Aleph 3 and am willing to give up some signal swing for the icreased resolution.
This topology also reponds to the best parts you can find. With few mods you are getting into the openess of the exotic tube amps. I used Vishay Holco and Dale wirewounds for the resistors,
IXYS diodes for the bridge, and Panasonic HFQs and Rubycon Black gates for caps. I also used Cardis jacks, silver solder, and silver plated teflon wire. Sounds Way Cool, runs pretty hot.
H.H.
 
Harry,
I don't remember what the standard torture test load is, but I put a 1 uF cap across an 8 ohm load and...nothing happened. The amp just kept right on chugging. Note that I did this 'live' so the amp hit the cap at full power.
Okay, so full power is just shy of ten watts...
Point being, I'm not worried about stability.
As for "hotrod" versions of the circuit:
If you disconnect C3, then R12 doesn't do anything, so you might as well remove it. If R12 isn't there, you might as well remove R11, too. Okay, at that point R20 and R21 are actually causing more harm than good by mucking up the damping factor, so you might as well take them out.
Guess what?
You don't have a hotrod Aleph 3, you have a Zen with a front end on it.
By definition, it's the extra stuff on the current source that makes it an Aleph. Once you remove that, there's nothing particularly unusual about the circuit. There are scads of variants of this idea on the Web even as we speak. For that matter, there are variants being sold commercially, too. Perfectly legally, mind you, since a single-ended amp with a current source is old hat. I have no doubt that the amp sounds good, but it's not an Aleph any more.
Oh, Vince, almost forgot...
Keep in mind that you can jack up the bias by increasing R11. The stock value for the Mini-A is 47k. That gives you about an amp of bias. Just for fun, I tossed in a 220k resistor, and ramped up the bias to something like 1.6A.

Grey
 
Aleph Again

Well...... It looks like an aleph 3! I leave discussions on semantics
and definitions to the writers. As an engineer and audio designer, I will stick to technical issues. Removing the two 0.47 ohm resistors will do very little to the damping factor since they are inside of the feedback loop and the amplifier has a hign open loop output impedance due to the nature of it's topology. I would
also not really call to Aleph amps"a Zen with a front end." I have seen several of the copys of the Aleph type circuits and some on pretty neat, but not very practical or reliable. I still maintain that
the 0.001 uF cap is A GOOD IDEA for stability. Go run some 10K square waves and try 1000pF and 0.1uF on the output and look
at the output on a 100Mhz scope. Phase margin and stability are one of the most overlooked factors in amp design. (particularly amateur constructors) Go read the article by Mr. Pass on how very capacitive cables can even blow some amps up! I offered
the sugestion of removing the feedback loop as an easy experiment in the spirit of simplifying the circuit. I think even Mr. Pass might aprove......