I would go with a transformer of 500VA or more. The more, the better.
http://www.victoriamagnetics.com
makes custom transformers for a reasonable price. You can get transformers anywhere between 50VA up to 2kVA, for the voltage that you need.
I have ordered several transformers from them:
pic of transformers from them for my aleph2s and bosoz
Good luck with your project.
--
Brian
http://www.victoriamagnetics.com
makes custom transformers for a reasonable price. You can get transformers anywhere between 50VA up to 2kVA, for the voltage that you need.
I have ordered several transformers from them:
pic of transformers from them for my aleph2s and bosoz
Good luck with your project.
--
Brian
Shunt regulators for Aleph
"The load will be pretty near constant" Actually this will not be true for an Aleph type circuit when the output stage current source go into the dynamic current mode when the AC feedback senses enough current though the sense resistors. Shunt regulators are pretty inefficient to use for the output stage of a power amp by the time you bias them with enough current to be useful. The source follower mosfet regulation approach in Mr. Pass's latest Zen variations article works real well and sounds fine on another mosfet amp that I built. It is simple and uses no negative feedback which should avoid most of the active regulation concerns.
H.H.
"The load will be pretty near constant" Actually this will not be true for an Aleph type circuit when the output stage current source go into the dynamic current mode when the AC feedback senses enough current though the sense resistors. Shunt regulators are pretty inefficient to use for the output stage of a power amp by the time you bias them with enough current to be useful. The source follower mosfet regulation approach in Mr. Pass's latest Zen variations article works real well and sounds fine on another mosfet amp that I built. It is simple and uses no negative feedback which should avoid most of the active regulation concerns.
H.H.
Ow!
Thanks for the quick reply Brian. I've searched the forums for info about power supplies but I don't get the reasoning behind the transformer sizes needed.
500VA? I didn't think I would need that much. Is there a link to a thread or site which explains transformer requirements? If my amp dissipates 140W (is this idle or under load?) then what size transformer is adequate? Can an undersized transformer be compensated for with large capacitors if I don't really care about continuous power ratings? I'm trying to minimize costs which is why I ask such things; sorry if I made any bad generalizations. Thanks.
Thanks for the quick reply Brian. I've searched the forums for info about power supplies but I don't get the reasoning behind the transformer sizes needed.
500VA? I didn't think I would need that much. Is there a link to a thread or site which explains transformer requirements? If my amp dissipates 140W (is this idle or under load?) then what size transformer is adequate? Can an undersized transformer be compensated for with large capacitors if I don't really care about continuous power ratings? I'm trying to minimize costs which is why I ask such things; sorry if I made any bad generalizations. Thanks.
Kilentra,
If you go with an inductor input to the filter, the voltage will be lower and the current higher. In this way, you could use a higher voltage, higher VA transformer.
Or you could just use a regulator.
Harry,
I have no preference for shunt regulators, myself, but others do. If they're willing to go to the trouble to build a large enough one, I know of no reason why it shouldn't work.
It's not whether <i>one</i> side of Aleph circuitry varies...we're in agreement on that...but we have two sides drawing in opposition on the same rail. That's what evens out the current draw. I would not recommend regulating the sides separately; not that it wouldn't work, I suppose, but you'd be giving up the more constant current draw that having two sides gives you.
Grey
If you go with an inductor input to the filter, the voltage will be lower and the current higher. In this way, you could use a higher voltage, higher VA transformer.
Or you could just use a regulator.
Harry,
I have no preference for shunt regulators, myself, but others do. If they're willing to go to the trouble to build a large enough one, I know of no reason why it shouldn't work.
It's not whether <i>one</i> side of Aleph circuitry varies...we're in agreement on that...but we have two sides drawing in opposition on the same rail. That's what evens out the current draw. I would not recommend regulating the sides separately; not that it wouldn't work, I suppose, but you'd be giving up the more constant current draw that having two sides gives you.
Grey
What with people posting while I'm writing and getting distracted by other things, I'm not getting all the points answered. Let me try again.
Kilentra,
If you anticipate 140W dissipation, then something along the lines of 300VA will do just fine. If you ventillate well, you can go for 250VA. Don't get me wrong...I like using a larger transformer if I can afford the luxury, but 250VA will do the job nicely if ventillated properly.
Oh, and I agree with Harry that Nelson's regulator is a good choice. I'm planning a more elaborate one, but just for the fun of it.
Sonny,
By all means post results.
I think I'm caught up now...maybe...
Grey
Kilentra,
If you anticipate 140W dissipation, then something along the lines of 300VA will do just fine. If you ventillate well, you can go for 250VA. Don't get me wrong...I like using a larger transformer if I can afford the luxury, but 250VA will do the job nicely if ventillated properly.
Oh, and I agree with Harry that Nelson's regulator is a good choice. I'm planning a more elaborate one, but just for the fun of it.
Sonny,
By all means post results.
I think I'm caught up now...maybe...
Grey
Thanks Grey, I can go to sleep now. I was going to use some nice 120mm fans along with heatsinking that would be 'barely acceptable' by itself. The transformer should be cool enough and there will be at least .1 farad after it. Power is not my main concern and I really don't mind if my amp is restricted to 50 watts or something because of an inadequate component which I can upgrade later if I want. My problem was just that I was not really ready to pay for custom made transformers and couldn't find a larger 12V one.
I'm just curious as to how much money these things are costing with all the expensive components I see here. Mine should total under $400 for a pair of monoblocks (just the parts, I will use a very simple chassis or none at all) but that will probably go up as usual when I can't find good prices on heatsinks, fans, caps, and stuff like that. It happens.
I'm just curious as to how much money these things are costing with all the expensive components I see here. Mine should total under $400 for a pair of monoblocks (just the parts, I will use a very simple chassis or none at all) but that will probably go up as usual when I can't find good prices on heatsinks, fans, caps, and stuff like that. It happens.
I noticed already that it might be quite a trouble to obtain high power transformers with such low voltages, unless of course you want to go custom. My favourite Chinese store which stocks Plitron has 300VA transformers with double 22V secondaries and the 500VA with 20V secondaries. Too high for my needs.
I have 700VA and 1000VA transformers and could use Variac with them, but it's too much hassle.
So for now, I'm finishing Aleph 5 monoblocks I started previously.
I have 700VA and 1000VA transformers and could use Variac with them, but it's too much hassle.
So for now, I'm finishing Aleph 5 monoblocks I started previously.
Attachments
Aleph X power supply observations
"Harry,
I have no preference for shunt regulators, myself, but others do. If they're willing to go to the trouble to build a large enough one, I know of no reason why it shouldn't work."
Actually I might know of a couple reasons they are not practical in this case. A typical shunt regulator goes in parallel with the load instead of in series with it. It gives up current draw as the load increases its current draw to keep are constant current draw from the rest of the supply. The source feeding the parallel combination of the load and shunt regulator are usually a current source or a resistor much larger than the output impedance of the shunt regulator to form a voltage divider for noise from the raw supply. It works best when the bias current for the shunt regulator is greater than the ac variations of the load current. This being for the standard reasons for Class A operation. With the large current demands from a power amp one can see this quickly becomes very impractical. Shunt regulator are great for preamp and low level circuits but I have never seen one for the output stage of a power amp.
"It's not whether one side of Aleph circuitry varies...we're in agreement on that...but we have two sides drawing in opposition on the same rail. That's what evens out the current draw. I would not recommend regulating the sides separately; not that it wouldn't work, I suppose, but you'd be giving up the more constant current draw that having two sides gives you."
This rational would hold if the circuit connections to the positive rail were constant current sources but not for the Aleph current sources. As the load current out exceeds the DC bias current of the current sources, the dynamic current source is pulling one side of the to the positive rail while the other side of the load is being pulled toward the negative rail by the output mosfet. The net effect is an increase in current drawn from both supplies. Seperate regulators for each output would definitely be unwanted for the negative rail which benefits from a good low impedance connection between source resistors with a short low resistance trace or wire.
H.H.
"Harry,
I have no preference for shunt regulators, myself, but others do. If they're willing to go to the trouble to build a large enough one, I know of no reason why it shouldn't work."
Actually I might know of a couple reasons they are not practical in this case. A typical shunt regulator goes in parallel with the load instead of in series with it. It gives up current draw as the load increases its current draw to keep are constant current draw from the rest of the supply. The source feeding the parallel combination of the load and shunt regulator are usually a current source or a resistor much larger than the output impedance of the shunt regulator to form a voltage divider for noise from the raw supply. It works best when the bias current for the shunt regulator is greater than the ac variations of the load current. This being for the standard reasons for Class A operation. With the large current demands from a power amp one can see this quickly becomes very impractical. Shunt regulator are great for preamp and low level circuits but I have never seen one for the output stage of a power amp.
"It's not whether one side of Aleph circuitry varies...we're in agreement on that...but we have two sides drawing in opposition on the same rail. That's what evens out the current draw. I would not recommend regulating the sides separately; not that it wouldn't work, I suppose, but you'd be giving up the more constant current draw that having two sides gives you."
This rational would hold if the circuit connections to the positive rail were constant current sources but not for the Aleph current sources. As the load current out exceeds the DC bias current of the current sources, the dynamic current source is pulling one side of the to the positive rail while the other side of the load is being pulled toward the negative rail by the output mosfet. The net effect is an increase in current drawn from both supplies. Seperate regulators for each output would definitely be unwanted for the negative rail which benefits from a good low impedance connection between source resistors with a short low resistance trace or wire.
H.H.
Shunt regs
<b>HarryHaller said
Shunt regulators are pretty inefficient to use for the output stage of a power amp by the time you bias them with enough current to be useful.</b>
Agreed Harry. I forgot to mention in the earlier post I meant to look into these for the input stage only. With a well designed lowish Z supply, would a reg for the output make a lot of difference if the front end was well controlled? I've never used them for output stages.
<b>HarryHaller said
Shunt regulators are pretty inefficient to use for the output stage of a power amp by the time you bias them with enough current to be useful.</b>
Agreed Harry. I forgot to mention in the earlier post I meant to look into these for the input stage only. With a well designed lowish Z supply, would a reg for the output make a lot of difference if the front end was well controlled? I've never used them for output stages.
Kilentra,
I'm planning to build a stereo Aleph-X amp using a single 1kVA toroidal transformer. The trannie has two 27V center tap secondaries (13.5V - 0V - 13.5V) rated at 16A each. I don't know how well that fits into your plans, but I have more than one of these trannies, so perhaps we could work something out. It would be best if you live in the US since the trannie has only a single 120V input and due to shipping costs. Please feel free to contact me privately if you have an interest.
Jim
I'm planning to build a stereo Aleph-X amp using a single 1kVA toroidal transformer. The trannie has two 27V center tap secondaries (13.5V - 0V - 13.5V) rated at 16A each. I don't know how well that fits into your plans, but I have more than one of these trannies, so perhaps we could work something out. It would be best if you live in the US since the trannie has only a single 120V input and due to shipping costs. Please feel free to contact me privately if you have an interest.
Jim
Output Device Question
I've been curious about this since this project started at the beginning of the year...
I know that Nelson has posted that the current product line is using the 240's.
I figure that the IRFP040's and 044's will also work...
Here's the question:
Since the rail voltages on this amp are so low, can we use smaller (voltage wise) output devices? I'm asking this because I remember reading posts stating that the lower voltage devices sound better...
Secondly, are any of you familiar with other devices that might better either the 040's or the 240's for this application? I know the 040's are hard to find but if they are the absolute best for this application, then I'm willing to try to find them... Also, do the 044's sound the same as the 040's?
I'm thinking:
IRFP040
IRFP044
IRFP140
IRFP240
Thanks in advance for your thoughts,
Steve
I've been curious about this since this project started at the beginning of the year...
I know that Nelson has posted that the current product line is using the 240's.
I figure that the IRFP040's and 044's will also work...
Here's the question:
Since the rail voltages on this amp are so low, can we use smaller (voltage wise) output devices? I'm asking this because I remember reading posts stating that the lower voltage devices sound better...
Secondly, are any of you familiar with other devices that might better either the 040's or the 240's for this application? I know the 040's are hard to find but if they are the absolute best for this application, then I'm willing to try to find them... Also, do the 044's sound the same as the 040's?
I'm thinking:
IRFP040
IRFP044
IRFP140
IRFP240
Thanks in advance for your thoughts,
Steve
... 044 and 040 are similar but i'd still prefer 040's if available, sigh......... i'll probably be using 044's.Harry,
Actual measurement at full power into 8 ohm load, 1kHz signal: 90mVAC on the rail.
The thing you're not taking into account is that as one current source draws more current, the other side is releasing a symmetrical amount. Result=0. Or pretty darned close. I attribute the residual to slight mismatches in the devices and the fact that I haven't retuned the current sources properly since I swapped out some parts values.
Like I said, I'm not advocating shunt regulators, per se--never tried one, so I have no opinion--but if the load is that close to DC, then the waste on a properly designed regulator is going to be reduced considerably, since you won't have to be allowing for large AC swings coming back up the rail.
On the other hand, if the load is that near DC, I'm not sure that any sonic differences between series and shunt regulators are going to be audible, as just about any regulator can do the DC dance with no sweat.
Nonetheless, I'm still planning on a series regulator. Just need some time (and parts), that's all.
vinay,
Your limit isn't the circuit so much as the dissipation on the devices you use.
If you start reducing the bias current through the devices, you're likely to end up with an amplifier that will only drive really high impedances; in other words, it will current limit.
You need to balance the voltage swing and the current available in order to come up with an output that will work well into realistic speaker loads.
Taken to extremes, I suppose you could use the circuit to direct drive some electrostat panels, but it's not a project I'm prepared to undertake. Running ESLs straight off an amplifier has stumped minds a lot keener than mine.
(That's not to say I haven't built a prototype, but that was long ago in a galaxy far, far away. Yes, it worked. No, it didn't blow up. How long it would have run in the real world...I dunno...)
Grey
Actual measurement at full power into 8 ohm load, 1kHz signal: 90mVAC on the rail.
The thing you're not taking into account is that as one current source draws more current, the other side is releasing a symmetrical amount. Result=0. Or pretty darned close. I attribute the residual to slight mismatches in the devices and the fact that I haven't retuned the current sources properly since I swapped out some parts values.
Like I said, I'm not advocating shunt regulators, per se--never tried one, so I have no opinion--but if the load is that close to DC, then the waste on a properly designed regulator is going to be reduced considerably, since you won't have to be allowing for large AC swings coming back up the rail.
On the other hand, if the load is that near DC, I'm not sure that any sonic differences between series and shunt regulators are going to be audible, as just about any regulator can do the DC dance with no sweat.
Nonetheless, I'm still planning on a series regulator. Just need some time (and parts), that's all.
vinay,
Your limit isn't the circuit so much as the dissipation on the devices you use.
If you start reducing the bias current through the devices, you're likely to end up with an amplifier that will only drive really high impedances; in other words, it will current limit.
You need to balance the voltage swing and the current available in order to come up with an output that will work well into realistic speaker loads.
Taken to extremes, I suppose you could use the circuit to direct drive some electrostat panels, but it's not a project I'm prepared to undertake. Running ESLs straight off an amplifier has stumped minds a lot keener than mine.
(That's not to say I haven't built a prototype, but that was long ago in a galaxy far, far away. Yes, it worked. No, it didn't blow up. How long it would have run in the real world...I dunno...)
Grey
Confound it, missed replying to posts again.
Regarding choice of output devices...any of the usual suspects should do quite nicely. Just watch your device ratings.
I'm still waiting for reports on stability from the field. I suspect that it's going to come down to one of two things: stray capacitance on these push boards I'm using or parts choice. The gate capacitance, for instance, may be critical for stability. Choose a MOSFET (or an NPN, for that matter) different from what I've got here, and you may have to put in a cap across the NPN. If someone goes for something like an IRFP240, let me know what happens. (Or a Zetex for the NPN...)
To date, I've yet to see any instability here, but someone, somewhere will manage to trigger it if it can be done.
Grey
Regarding choice of output devices...any of the usual suspects should do quite nicely. Just watch your device ratings.
I'm still waiting for reports on stability from the field. I suspect that it's going to come down to one of two things: stray capacitance on these push boards I'm using or parts choice. The gate capacitance, for instance, may be critical for stability. Choose a MOSFET (or an NPN, for that matter) different from what I've got here, and you may have to put in a cap across the NPN. If someone goes for something like an IRFP240, let me know what happens. (Or a Zetex for the NPN...)
To date, I've yet to see any instability here, but someone, somewhere will manage to trigger it if it can be done.
Grey
Another FET to consider ...
Apogee,
another choice is IRF054. I haven't used one but the data sheet and spice model imply it's equivalent to two 044s - about twice the gm, about twice the input capacitance and I expect it's about twice the price.
(One advantage of using 2x 044s would be that each dissipates half the power and the die temperature will be less).
What is so special about 040 - I cannot find a datasheet?
Regards
13th Duke of Wymbourne
Apogee,
another choice is IRF054. I haven't used one but the data sheet and spice model imply it's equivalent to two 044s - about twice the gm, about twice the input capacitance and I expect it's about twice the price.
(One advantage of using 2x 044s would be that each dissipates half the power and the die temperature will be less).
What is so special about 040 - I cannot find a datasheet?
Regards
13th Duke of Wymbourne
040's ceased production quite a while ago and probably the only chance of getting them is finding someone who has a private stash of them left from their production days .... they are the nicest, lowest distortion, lowish voltage output stage mosfets around. 044's/140's/240's are the usual suspects these days generally speaking the lower voltage types are sonically better if they are suitable for the circuit you wish to use. I generally wouldnt use x5x devices such as 054, 150, 250 etc as 2 x4x devices are almost always better due to die temp etc.
The thing you're not taking into account is that as one current source draws more cur
I don't know what "full power" means.... please put it terms of volts peak to peak across the load resistor so we can see how much AC current is being drawn by the load. It might be easiest to measure from ground to one output and multiply by two.
What is the actual DC bias current through each leg of the output stage? Your schematic shows 0.22 source resistors which would give a bit under 3 amps. Is this correct? Also, I assume the 10K pots V1 and V3 there to fine tune the bias adjustment. I notice Mr.Pass doesn't include the resistors (that the combination of the 47.5K resistors and the pots make) in the latest Aleph current source on the Zen variations article. Maybe with good transistor and source resistor matching they could be ommited?
I think with a 4 ohm load and enough signal for the amp to leave Class A operation to enter Class AB ( the real point of having the Aleph current sources as I understand it ) you will see significant increase in current draw from the supplies. I guess we will have to wait for Mr. Pass to return from his vacation to enlighten us, unless an ambitious Spice modeler out there might want to take a shot at this one.......
Back to work on the Tex-X. Thanks for your time and interest Grey.
H.H.
I don't know what "full power" means.... please put it terms of volts peak to peak across the load resistor so we can see how much AC current is being drawn by the load. It might be easiest to measure from ground to one output and multiply by two.
What is the actual DC bias current through each leg of the output stage? Your schematic shows 0.22 source resistors which would give a bit under 3 amps. Is this correct? Also, I assume the 10K pots V1 and V3 there to fine tune the bias adjustment. I notice Mr.Pass doesn't include the resistors (that the combination of the 47.5K resistors and the pots make) in the latest Aleph current source on the Zen variations article. Maybe with good transistor and source resistor matching they could be ommited?
I think with a 4 ohm load and enough signal for the amp to leave Class A operation to enter Class AB ( the real point of having the Aleph current sources as I understand it ) you will see significant increase in current draw from the supplies. I guess we will have to wait for Mr. Pass to return from his vacation to enlighten us, unless an ambitious Spice modeler out there might want to take a shot at this one.......
Back to work on the Tex-X. Thanks for your time and interest Grey.
H.H.
Sound vs. specs
I was going through the specs of various irfp's... low voltage mosfets have lower on resistance, higher in-out cap, marginally longer time responses nothing surprising here. I thought that the capacitance was the somewhat critical element in the sound performance. I guess not.
Can anyone shed some light on the topic?
PS. The A-X board is coming along nicely, I will post in a couple of days. I am trying to make it "super-symmetrical".
I was going through the specs of various irfp's... low voltage mosfets have lower on resistance, higher in-out cap, marginally longer time responses nothing surprising here. I thought that the capacitance was the somewhat critical element in the sound performance. I guess not.
Can anyone shed some light on the topic?
PS. The A-X board is coming along nicely, I will post in a couple of days. I am trying to make it "super-symmetrical".