Mini-A with no negaive feedback?

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
The 2 stage Alephs, including what I take to be the Mini-A,
are diffcult to operate without feedback, as they don't have
a "follower" type output stage, and their damping factor is
derived almost entirely through feedback.

Of course there are examples of other kinds of amplifiers where
this is not so difficult.

Having said that, did anyone ever come up with a Mini-A with
no feedback?

:cool:
 

winslow

Member
2005-08-05 6:27 am
I dunno if anyone has done that. Could you lower the feedback a great deal?

Reason I'm asking is I have a wonderful woman who bought be an Aleph mini for my birthday today. And I have 4 more boards that I need to build, and she's given me the thumbs up to go all out.

So I guess one of the sets will have J-Fets somewhere.

Thanks Papa! Love the amps!

Have to build a preamp one of these days too.
 
A Mini-A with no feedback?
Hmmm...
With no feedback?
Er...well, there's the Aleph current source, which derives its go-juice from a signal derived at the output...itself a feedback loop if you want to look at it that way.
But supposing you kinda blur your vision when your eyes fall on that part of the schematic. Let's assume that the 'no feedback' thing just meant deleting the node from the output to the backside of the differential.
Hmmm...
Okay, the problem breaks down into four parts:
1) DC offset
2) Distortion
3) Bandwidth
4) Damping factor
Did I miss anything?
Ooops...
5) Output stage bias stability
Well, for starters, the MOSFET front end has to go. Too much capacitance. It's been a loooong time since I did the Mini-A, and I don't recall the bare bones behavior as well as I'd like. Assuming that my noggin hasn't slipped too many cogs, I seem to recall that there's something like 20dB of feedback tucked in there. That's a factor of 10x. That implies that the open loop bandwidth is around 10kHz, plus or minus a few kHz. Distortion is going to be unacceptably high open loop. Etc.
Okay, so the MOSFET front end goes. Bring in some JFETs. Less capacitance involved, so the bandwidth will improve. The distortion should be better, too, but not yet acceptable. (Head scratching noises...) I'm not sure how well the distortion could be managed, given that the output isn't as cooperative in that sense as you might want. The bias/DC problem could be handled by a servo-ish kinda thing that injected or un-injected some DC at the Gate of the gain MOSFET, or maybe by...lemme think about that one before I suggest something too stupid to put into print. Damping factor is going to be the sort of thing that would respond best to some manipulation of the output stage topology; i.e. it wouldn't look like a normal Aleph any more.
Okay...here's my semi-official, off-the-cuff, not-for-attribution verdict: No. At least in the sense that I gather you have pre-made PC boards that you want to use. This particular circuit would not respond well to a simple cutting of the feedback loop. With that in mind, go ahead and build them as "normal" Alephs and be happy.
But as to the larger question of whether it's possible to build a lower feedback (as low as zero NFB might prove impossible) Aleph...now that's an interesting question.
It's curious that this topic should come up at this particular moment, given that I'm just returning from an extended sojourn in the land of no-feedback. I have a semi-final amp prototype on the bench which performs decently regarding the usual specs, and am in the process of building a second channel so that I can listen to the silly thing and see if it sounds at all tolerable. That will take a week or two or ten, because my time is limited and I'll have to do some metal working (something I'm ill-equipped to do--wood, yes--plastic, yes--metal, only with difficulty) so as to have another channel's worth of heatsinking. After that, I'd planned to drop another output stage on the other side and goose my 25W amp to 100W or so by bridging it. Then I'd planned to have a go at a phono stage that's been circling, waiting for clearance to land.
But a lower, possibly no-feedback Aleph would be an interesting trick, wouldn't it?
Ah, but how many people would actually go for such a thing, even if it should prove possible? The nattering nabobs of negativity over in the Solid State Forum attempt to burn me at the stake every time I mention no feedback designs. There are few supporters for such a radical idea. In fact, I can count them on one hand and have fingers left over. I am loath to put that much head-scratching into a possible dead end, particularly if no one will build it should the stereo gods chance to smile on my (sometimes) humble endeavors.

Grey
 
my time is limited and I'll have to do some metal working (something I'm ill-equipped to do--wood, yes--plastic, yes--metal, only with difficulty)

Grey--"wood, yes" implies you have a tablesaw. If so, cutting AL is no harder than cutting wood. I would say "shoot me an e-mail if you want more info", but I can't figure out how to turn on the button to allow members to e-mail me.

JJ

PS, sorry for the OT.....but hopefully it is worthwhile.
 

winslow

Member
2005-08-05 6:27 am
Reason I ask this is b/c I have a couple tube amps that I use on my horns that are zero negative feedback designs and they sound awesome and I wanted to make my new Mini-As even more cool.

And when searching for Aleph negative feedback, it seems like the Aleph J (I think it was) had 15 dB less NFB than the prior Alephs.

Some say the J is the best Aleph (for now), and that could be one of the reasons why.
 
I have not one, but two table saws. I don't know if you've ever tried to saw something the size of a decent-sized heatsink, but the blade doesn't extend all the way through. So then you have to roll it over and face the fact that even a well set up saw is going to be a thousandth or two off on the second pass. Aluminum doesn't sand well. And you've scarred the existing anodizing. And it's dangerous as hell because irregular surfaces like the cross section of a heatsink don't always want to slide across a fence gracefully, which means they bind, which means you're at risk of a serious kickback. The idea of catching twenty pounds of heatsink with my teeth doesn't excite me.
Not to mention wrecking my saw.
Can it be done?
Yes.
Are the results acceptable?
No.
At least not to me. I'm a perfectionist. I want it to look right. Note that I'm also not set up to anodize the cut ends, etc. etc. etc. (Yes, I know there are sites with DIY anodizing instructions, but the anodizing won't match.)
I'll do the minimum to get by and try to do battle with a local machine shop later this year. This is something I dread, as I've had awful experiences in the past with metal people not grokking the simple word precision.
I'm between a rock and a hard place. On the one hand there are people who accept mediocre metal work the same way that they accept some pretty uninspiring circuits. On the other hand there are those who live in the Northeast or out West in the US where they can just run down to the corner NASA-spec metal working emporium and have a perfect chassis done to their spec in an hour for the price of a hamburger. I'm unwilling to accept mediocrity and I don't have access to good metal working shops and I have no more space in my shop for a milling machine and all the other specialized items I'd need to do a good face plate and the various other things that I want done.
I can see what I want done in my mind's eye. I've done preliminary drawings. If anyone has noticed a post or two of mine here and there over the last year or two, they might be able to guess the direction I want to go. But getting there ain't going to be easy. You'd think that in a state as obsessed with racing as this one is, reboring cylinders and resurfacing cylinder heads and such from the time they're knee-high to a grasshopper would spawn legions of redneck geniuses at shaping metal. Sadly, it ain't so. All I can figure is that the racing teams here in the Southeast must hire in their shop crews from elsewhere in the nation; they certainly aren't finding them around here.

Grey
 
winslow said:
Reason I ask this is b/c I have a couple tube amps that I use on my horns that are zero negative feedback designs and they sound awesome and I wanted to make my new Mini-As even more cool.

And when searching for Aleph negative feedback, it seems like the Aleph J (I think it was) had 15 dB less NFB than the prior Alephs.

Some say the J is the best Aleph (for now), and that could be one of the reasons why.


What? You're trying to talk me into low-to-no feedback designs?
I'm already there. In spite of the snarls and growls of the I Worship Feedback crowd.
Ancient words of wisdom: The dogs bark, but the caravan moves on.

Grey
 
middle of page - find "Enable Private Messaging?"
well, I did that. On everyone elses posts I see a button at the bottom that says 'email'. I don't see that button at the bottom of my posts.

Grey-
I hope some of this will be useful:

I don't know if you've ever tried to saw something the size of a decent-sized heatsink
Those must be truly massive heatsinks! The biggest I have done are about 2.5 inches tall. A 10" tablesaw can make a cut that deep. I cut the fins in one pass, then flip and cut the thick back in 2 or three passes. Even so, this tends to gouge, so I take a real thin cut the entire height of the heatsink. With lots of lube (I like wd 40) and WD40, it won't gouge.

Aluminum doesn't sand well
Big flat plates sand just fine, but end cuts don't. Use a medium tooth bastard file. You will be amazed at how smooth the end cuts come out. It will take a lot of passes to get all the cut marks out, though.


And it's dangerous as hell because irregular surfaces like the cross section of a heatsink don't always want to slide across a fence gracefully, which means they bind, which means you're at risk of a serious kickback. The idea of catching twenty pounds of heatsink with my teeth doesn't excite me.

Please don't ever do that! (how could you type those quick short posts missing several fingers??). Take 20 minutes and make a small panel cutting sled. I have one that I made for speakerbuilding. It has some play in the rails. Even so, I cut 12" x 18" panels square enough that I can't measure the out-of-squareness with my carpenter's square.

Even with the above, it probably won't satisfy the perfectionist in you if you want a highly finished piece of work. I hope it will make doing the minimum a bit more tolerable, especially the sled for panel cutting and crosscutting.

(sorry, I have no help for chipping anodized aluminum -- none of my stuff is anodized)

JJ
 
winslow said:
Choky/Zen Mod, mind if I PM you a few questions?

Also, if you can't do completely away with the GNFB, is there a way to lower it considerably?


you can always PM me ; that's why is my "PM" buton enabled :rofl: ;)

you can decrease amount of overall NFB with decreasing OLG, but that must be compromise between -what your ears hear and what your CRO tells you ........ easy - with source degeneration in input LTP
 

Babowana

Formerly "jh6you". R.I.P.
2006-07-13 4:23 pm
www.aheadamp.com
Zen Mod said:
you can decrease amount of overall NFB with decreasing OLG, but that must be compromise between -what your ears hear and what your CRO tells you ........ easy - with source degeneration in input LTP


Mmm . . . decreasing OLG, decreasing NFB
. . . . . . . . increasing OLG, increasing NFB

and, if no strange thing happens in both ways . . .

It's just a matter of scaling up and down with the same basic charactor, isn't it?

I'm unsure myself . . .



:darkside:
 
jupiterjune said:


Those must be truly massive heatsinks!



Heat is an unavoidable part of the class A experience. If I use the water-cooled system, I can dump a tremendous amount of heat, but the circuit is no longer portable. If I go with passive heatsinks, I need serious hunks of metal, not toys. I confess that I am somewhat more conservative than Nelson, in that I prefer to be able to leave my hand on the heatsink indefinitely (even if I have scorch marks afterwards). Add to this the fact that my design goal for my circuit is that it not current limit driving my Magneplanar ribbon tweeters (ca. 2.5 Ohms) and you've got a recipe for quite a bit of heat. Given these various factors, the available heatsink profiles are limited to perhaps four or five models, such as the Aavid-Thermalloy 66975 or 60265.
Anyone who tries to advance the idea of a really wide, flat heatsink is advised to consider the number required and the resulting size of the enclosure. I don't want to run long lengths of wire from the front end to some of the more remote output devices.
There's a juggling act here, in that I want a rack mount chassis. That limits the width of the body to around 17" or so. I want the chassis to be 24" deep or less (which I think qualifies as a pretty deep chassis in most anyone's book). The transformer I'm using is 5" in diameter and I can get the circuitry on boards 5-6" wide. That leaves roughly 12" of cumulative depth that I can use for heatsinks.
Please don't try suggesting that I just make the heatsinks taller until you've done your homework on heatsink length vs. efficiency. If nothing else, go play with the online calculator at Aavid-Thermalloy. A heatsink that rates .075 degrees/Watt at 3" is not .0375 degrees/Watt at 6". Actually, you're well into diminishing returns if you go much above 6" tall. I've seen numerous posts where people show pictures of heatinks 12-15" high and seem to think that is accomplishing something useful. It's no better than a heatsink 6-8" high; the rest is just eye candy--wasted as far as heat transfer goes. Etc. etc. etc. I don't want to get off on a rant, here, it's just that few people seem to think about heat very deeply.
All I can say is that I appear to see my work differently from the way others see theirs.
Okay, back to the Mini-A feedback thing. Lowering the feedback isn't difficult. All you have to do is raise the value of the series resistor coming from the output to the backside of the differential or reduce the value of the resistor that goes from the backside to ground. Or both. What I tried to indicate before was that there will be consequences. I'd suggest setting limits as to how much DC you will tolerate at the output, and possibly distortion (assuming you have access to a distortion meter) and the other parameters I mentioned. Then play with the circuit and see what you can do. Bear in mind that a lot of what happens will be specific to your circuit alone. As you reduce the feedback the circuit itself becomes more important. Matching components carefully will help considerably.
I've spent the last year or so putting together a zero feedback circuit. I certainly don't claim to have the ability that Nelson, John, or Charles bring to the table, but I've spent a little time thinking about it, and all I can say is that it isn't necessarily easy. Don't just snip the feedback loop and think your work is done.

Grey