I'm using EI. Magnetic leakage of EI cores is easily shielded with nickel-iron alloy sheeting. The size is the only disadvantage, if you don't have the space. On the other hand, EIs have lower bandwidth and provide better isolation from mains noise, and do not buzz if there's some DC on the mains due to single-phase rectifier distortion etc. which is not that uncommon on household wiring; DC traps are rarely used with EI transformers.
If you're really cautious of magnetic leakage, use R-core; they have significantly less than a toroidal. Also, they have virtually no interwinding capacitance; other transformer types with windings on top of one another should have an electrostatic shield to avoid that.
If you're really cautious of magnetic leakage, use R-core; they have significantly less than a toroidal. Also, they have virtually no interwinding capacitance; other transformer types with windings on top of one another should have an electrostatic shield to avoid that.
Nixie said:
Not true. The loudest, most obnoxious transformer in the house is the EI in the UPS here by my ankle. It goes nuts at least four or five times a day now. Second loudest is the EI in the power supply in the rack in the lab downstairs. I can count on a good buzz around 13:00, plus a few others at random intervals during the day.
All my toroids are much quieter, regardless of manufacturer.
Grey
That's very interesting, and contrary to my own experience, though I've only had two toroidal transformers so maybe my sample size is small. Both would buzz when I turn on some of the power tools in the workroom, which is not even on the same circuit. The EIs usually I rewind myself and I found that dipping the rebuilt transformer in varnish and putting a vacuum to it for about a minute, then drying over a week, removes any buzz so that it's only audible if I put my ear to it.
I forgot to mention, R-cores are also very easy to rewind, since you just spin the bobbin (building a winding machine for this is trivial trivial), which is possible due to the fact that the core is straight through the bobbin and has a round cross section. Adding or pulling a few top turns to/from a toroidal core is possible, but anything more is out of the question. EI cores, of course, have to be completely disassembled, although with some effort it is possible to remove some top turns (I've done it).
wuffwaff said:
Thank you William!! Your thoughtful comments were extremely helpful for me and provided enough understanding to tinker and continue to learn.
BTW, overall you are a great DIYer on this site and have always enjoyed reading your helpful and insightful posts. The contribution of the Aleph and Alephx calculator are as indespensible as Duncan's PSU Simulator to me.
Thank you again,
-David
A and AB
Just to avoid letting to much time go by without raising the biasing issue, here is yet another question about it:
If I understand it correctly the Passlabs X1000 is operating in class A up to a certain output level, where it starts operating in AB, i.e. the bias starts swinging. Would there be any convenient way to make the Aleph-X behave in a similar manner? This just to avoid making a choise of loudspeaker impedance to match the amplifier for. If A conditions would work up to the 15V rails and 4.5 amps bias in the original design by Gray, that would allow some 34W into 8 ohms class A according the The Creator. However, when you some day tinker with a couple of diy speakers that behave badly and suddenly show an impedance of around 2 ohms, things get messy, don't they? Gray said he would use a modified version with about 12V rails and higher current for his nasty tweeter, but that would give less power if used with a 8 ohm speaker, wouldn't it? So, wouldn't it be convenient to have AB as a power reserve for those days when you have your local biker gang over for tea and biscuits and they want to hear the music over the sweet sound of their hogs idling in your living room? Of course I could just parallel enough FETs to handle the heat, have high enough rail voltage for the 8 ohm:ers and high enough current to handle a short cirquit, but then i would soon start to be in trouble with my government, since Swedish officials signed some stupid paper in Kyoto long ago.
Gray,
I just browsed through the first posts in the Aleph-X thread and in one of them you stated to be 44 yrs old. That was in May 2002. If your birthday is later than may, you might, unless I'm mistaken, be coming up on the big five-o. Or is that next year? Sorry for getting personal like this, but I think it would be suitable with a proper diyaudio celebration.
Everyone sends a beer?
Everyone sends a FET?
Enough rambling.
Hilbert
Just to avoid letting to much time go by without raising the biasing issue, here is yet another question about it:
If I understand it correctly the Passlabs X1000 is operating in class A up to a certain output level, where it starts operating in AB, i.e. the bias starts swinging. Would there be any convenient way to make the Aleph-X behave in a similar manner? This just to avoid making a choise of loudspeaker impedance to match the amplifier for. If A conditions would work up to the 15V rails and 4.5 amps bias in the original design by Gray, that would allow some 34W into 8 ohms class A according the The Creator. However, when you some day tinker with a couple of diy speakers that behave badly and suddenly show an impedance of around 2 ohms, things get messy, don't they? Gray said he would use a modified version with about 12V rails and higher current for his nasty tweeter, but that would give less power if used with a 8 ohm speaker, wouldn't it? So, wouldn't it be convenient to have AB as a power reserve for those days when you have your local biker gang over for tea and biscuits and they want to hear the music over the sweet sound of their hogs idling in your living room? Of course I could just parallel enough FETs to handle the heat, have high enough rail voltage for the 8 ohm:ers and high enough current to handle a short cirquit, but then i would soon start to be in trouble with my government, since Swedish officials signed some stupid paper in Kyoto long ago.
Gray,
I just browsed through the first posts in the Aleph-X thread and in one of them you stated to be 44 yrs old. That was in May 2002. If your birthday is later than may, you might, unless I'm mistaken, be coming up on the big five-o. Or is that next year? Sorry for getting personal like this, but I think it would be suitable with a proper diyaudio celebration.
Everyone sends a beer?
Everyone sends a FET?
Enough rambling.
Hilbert
The way I figured to do match a wide range of loads was simply to have multiple voltages and bias settings that can be switched. My transformer has multiple taps on the primary, so the voltage change is easy. Bias can be changed by switching a different bias resistor into place. This seems to me the best solution.
Hi Hilbert,
since the X is a kind of balanced single ended design (here the single ended part is important) it needs to be operated in class A.
Thanks to the active current source the efficiency is already twice as good (half as bad) as a normal single ended amp.
The only thing you could do to have a bit more power is to raise the ac-current-gain but this will have other (distortion) disadvantages.
So what you could do is go for a 20-25V supply (as recommended by Mr. Pass himself) and enough heatsinking for 7-8A of bias.
As long as your speakers are 8 ohms and behave nicely you can reduce the bias to around 5A. If you change thenm for something a bit more nasty and have problems with the available power you can up the bias to 8A.
Of course you could also build some 102 dB/w 12 Ohm speakers and recuce the bias to 2.5A........
William
P.S. dissipation 250, 400, 125 watts respectively
since the X is a kind of balanced single ended design (here the single ended part is important) it needs to be operated in class A.
Thanks to the active current source the efficiency is already twice as good (half as bad) as a normal single ended amp.
The only thing you could do to have a bit more power is to raise the ac-current-gain but this will have other (distortion) disadvantages.
So what you could do is go for a 20-25V supply (as recommended by Mr. Pass himself) and enough heatsinking for 7-8A of bias.
As long as your speakers are 8 ohms and behave nicely you can reduce the bias to around 5A. If you change thenm for something a bit more nasty and have problems with the available power you can up the bias to 8A.
Of course you could also build some 102 dB/w 12 Ohm speakers and recuce the bias to 2.5A........
William
P.S. dissipation 250, 400, 125 watts respectively
Nixie said:
I've been thinking about that too, but I was hoping for a solution that didn't mean a lot of dismounting and soldering, which I guess would be the case for the bias resistors. Also, maybe there would be even more tinkering to get the amp to stabilize to the new settings?
wuffwaff said:
So, going absolutely nuts: Couldn't the AX be 'wrapped' in a complimentary pair, that would kick in only when the AX starts clipping, letting the clipped tops of the waves extend into a somewhat distorted but still feasible complimentary buffer?
I suppose this might be ridiculous. I know I haven't got the first idea how to do it.
Maybe I'll just go for the 102 dB/w speakers. Hm... could one build a full range pizeo speaker?
Maybe a distributed mode bending wave panel driven by piezo patches...
Oh, well...
Oh, Mr Rollins, I just realized I spelled your name wrong in my entire last post. I cannot tell you how stupid I feel. I humbly beg your pardon.
unfortunately still,
Hilbert
Heat and class A are two sides of the same coin. If you want to have class A, you're going to have to put up with the heat. Attempts to reduce the heat penalty incurred with class A go back to prehistoric times. They've unearthed Neanderthal circuits that attempted to use class A to cook venison steaks, then switch down to class B as soon as dinner was done.
Didn't work. Either the steak was undercooked (a common problem in those days), or it was overcooked (an equally common problem). Neanderthals, however, were tough...they'd eat it anyway. And yes, the family's pet saber toothed tiger still burned his snout nosing around the cooling fins. This led to an outraged set of fangs lashing out at all in the vicinity and brought an abrupt end to the experiment.
In more recent epochs, people have attempted sliding bias circuits (Pass, et. al.), and little class A circuits riding atop class B circuits (Walker, Pass, et. al.), etc. The idea is tempting, for obvious reasons, but you just can't cheat Mother Nature. Some circuits were more highly regarded than others, but none have exactly set the hi-fi world on fire...if you'll pardon the expression.
Switching rail voltages is workable over a limited voltage range, but I'd advise against doing it on the fly. Better to shut down and reboot, so to speak. If you're going to try drastic voltage changes, there are some resistor values that you might want to change.
Changing bias current can also be done, within limits. There's a resistor in the Aleph current source that can raise or lower the bias a bit. That one can be changed with the amp hot--I've done it.
Trying to manhandle this particular circuit into doing class AB isn't going to produce very satisfactory results. Better to build another amp. It'll give you an excuse to build something else, like an A-75--it's more amenable to this sort of thing.
hilbert,
My name is what it is. I get Greg, Gary, Gray, and all sorts of other things...and that's not even counting the words your mother used to spank you for saying. It was hell growing up, but works to my advantage now that I'm an author. As far as I am aware, there is only one person alive on Earth today with my name, and I'm it. Reduces chances people will confuse me with, say, Stephen King, a name which is actually fairly common. Of course, the downside is that no one is likely to mistakenly send me Stephen's royalty checks, either.
Bummer.
And, yes, I'll turn 50 on September 10th.
Grey
Didn't work. Either the steak was undercooked (a common problem in those days), or it was overcooked (an equally common problem). Neanderthals, however, were tough...they'd eat it anyway. And yes, the family's pet saber toothed tiger still burned his snout nosing around the cooling fins. This led to an outraged set of fangs lashing out at all in the vicinity and brought an abrupt end to the experiment.
In more recent epochs, people have attempted sliding bias circuits (Pass, et. al.), and little class A circuits riding atop class B circuits (Walker, Pass, et. al.), etc. The idea is tempting, for obvious reasons, but you just can't cheat Mother Nature. Some circuits were more highly regarded than others, but none have exactly set the hi-fi world on fire...if you'll pardon the expression.
Switching rail voltages is workable over a limited voltage range, but I'd advise against doing it on the fly. Better to shut down and reboot, so to speak. If you're going to try drastic voltage changes, there are some resistor values that you might want to change.
Changing bias current can also be done, within limits. There's a resistor in the Aleph current source that can raise or lower the bias a bit. That one can be changed with the amp hot--I've done it.
Trying to manhandle this particular circuit into doing class AB isn't going to produce very satisfactory results. Better to build another amp. It'll give you an excuse to build something else, like an A-75--it's more amenable to this sort of thing.
hilbert,
My name is what it is. I get Greg, Gary, Gray, and all sorts of other things...and that's not even counting the words your mother used to spank you for saying. It was hell growing up, but works to my advantage now that I'm an author. As far as I am aware, there is only one person alive on Earth today with my name, and I'm it. Reduces chances people will confuse me with, say, Stephen King, a name which is actually fairly common. Of course, the downside is that no one is likely to mistakenly send me Stephen's royalty checks, either.
Bummer.
And, yes, I'll turn 50 on September 10th.
Grey
GRollins said:
Ah! Too bad they gave it up just before the ice age. I blame the tigers. Stupid cats with mammuth complex. I'd never trust any one with a dental ornament that size. That's why they invented braces in the first place.
But I want both!! Both dammit!!
A friend is working at the national super computing center, and they subscribe to central cooling as well as central heating. A powerplant nearby sends them heat if they need it and coolant if they need it. They send the luke warm coolant back and get a discount. Maybe I could subscribe as a private person and cut down on the bill too.
I give up and will simply stand the heat.
Thank you George, it won't happen again!
Now for the grand prize: How do we plan a surprise for you on a public forum?
Suggestions anyone?
Hilbert
ps. Maybe Stephen King (any of them) could send you a royalty check for your 50th birthday to make it up to you?
Water cooling is quite cheap compared to air cooling and works so well that you may find it difficult to get the system sufficiently warm to grill dinner. Incidentally, this method of cooling has the endorsement of STAG (Saber Toothed Amplifier Group).
I did a thread on water cooling once upon a time. The original system is still in use. Really, the only complaint I have is that you can't just pick up an amp and tote it over to a friend's house for a listen. It's your place or no place.
Very handy if you've got an eye on a particularly fetching cavewoman. "Sorry, dear, I just can't move the thing. Took two mammoths and a whole tribe of those new-fangled Cro-Magnon critters to get everything in place. But I do have some particularly nice mead if you'd like to have a sip."
Grey
I did a thread on water cooling once upon a time. The original system is still in use. Really, the only complaint I have is that you can't just pick up an amp and tote it over to a friend's house for a listen. It's your place or no place.
Very handy if you've got an eye on a particularly fetching cavewoman. "Sorry, dear, I just can't move the thing. Took two mammoths and a whole tribe of those new-fangled Cro-Magnon critters to get everything in place. But I do have some particularly nice mead if you'd like to have a sip."
Grey
GRollins said:
Are they a part of the Feline Amplifier Society (FAS)?
I just saw a nice pic of some copper blocks for water cooling over at passdiy.com. Using such cool (sorry) blocks, a minature version of a household radiator and a silent circulation pump, I could even get approval from my felix domesticus. To up the dissipation further I could mount the FETs on Peltier elements through which I run the bias. Might as well put that current to work, right?
Considering the fact that mead always should be served warm (with a table spoon of honey and a drop of lemon), I could as well run my mead tap through my amps too.
I think I have to see a man about some copper tubing...
Nixie said:
Oh, sorry, I though you meant a switch just for the transformer. Maybe it would be possible to have a relay (or a few of them) to choose between both rail voltages and bias resistors. Not on the fly perhaps, as noted by Grey, but I guess it would be possible to have a "4ohm - 8ohm" switch that only should be used when the amp is powered down.
A really clever person (i.e. not me) could probably come up with some logic cirquitry that would prevent switching between impedances as long as main power is switched on.
jacco vermeulen said:
Hey, nice! Thanks! Now I can start chasing back-issues instead of writing my PhD thesis.
Hilbert