Hi Stefano,
The output of the AX100 circuit as published in this thread should be less into 4 ohms than 8 ohms. There is more info on this in the WIKI and the Aleph-X thread. This is the nature of the Aleph CCS style output stage. Something is wrong with the circuit you are simulating if the output is greater into 4 ohms.
This circuit is not the best choice for driving a 2 ohm load. It will sound fine but the output will be quite limited and the efficiency of operation will be really poor.
There are Passlabs commercial circuits that have hung 2 to 12 output devices on a front end just like this one. They all worked fine. The issue is driving the total gate capacitance. According to NP the amps with fewer output devices tend to lean toward sweeter highs. The amps with lots of output devices tend to have a slightly darker sound. Six output devices seems to be the sweet spot. The Aleph 2 and the XA160 both had this number of outputs being driven off the diff pair. Both of these units received serious critical acclaim.
The Aleph-X is a carefully crafted balance of a small number of components. The circuit can be tweeked only so far in any given direction before the sonics or the stability begin to degrade.
Have you read the Aleph-X and Aleph-X builders threads? And have you read the WIKI? If you are seriously going to play with this circuit you need to understand what people have done before.
Graeme
The output of the AX100 circuit as published in this thread should be less into 4 ohms than 8 ohms. There is more info on this in the WIKI and the Aleph-X thread. This is the nature of the Aleph CCS style output stage. Something is wrong with the circuit you are simulating if the output is greater into 4 ohms.
This circuit is not the best choice for driving a 2 ohm load. It will sound fine but the output will be quite limited and the efficiency of operation will be really poor.
There are Passlabs commercial circuits that have hung 2 to 12 output devices on a front end just like this one. They all worked fine. The issue is driving the total gate capacitance. According to NP the amps with fewer output devices tend to lean toward sweeter highs. The amps with lots of output devices tend to have a slightly darker sound. Six output devices seems to be the sweet spot. The Aleph 2 and the XA160 both had this number of outputs being driven off the diff pair. Both of these units received serious critical acclaim.
The Aleph-X is a carefully crafted balance of a small number of components. The circuit can be tweeked only so far in any given direction before the sonics or the stability begin to degrade.
Have you read the Aleph-X and Aleph-X builders threads? And have you read the WIKI? If you are seriously going to play with this circuit you need to understand what people have done before.
Graeme
Thanks for your reply.
i'll post something from the big 3d:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=28336&perpage=10&pagenumber=35
the out power on 4 ohm increased on a bias of 5A.
what i got from the simulation is that the output power was limited, like a sinusoid with the top edge cutted and increasing the bias and the number of output devices, the missing edge was appearing incresing so the output power.
But at this point i'm wondering: what makes this amplifier so worthwhile if it can't be projected to drive an hard load?
Anyways i'm reading the big aleph 3d but it's a really big mess sometimes....i'm tring to go little by little.
Since this morning i got up to page 39.
What i should say is.....accordingly on what i have read so far... in order to get the amp working on 4ohm...for instance...the guys there had to play with the bias.
I 'll read it better.
I would gladly attach my simulation on this forum with the circuit drawn with capture, if it wasn't for a rubber that just stolen yesterday my laptop out of my car...so i have to rebuild the circuit and re-simulate it....then i'l post the results.
In the menatime i'll keep reading the bi post trying to get a better handle on the subject!
i'll post something from the big 3d:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=28336&perpage=10&pagenumber=35
the out power on 4 ohm increased on a bias of 5A.
what i got from the simulation is that the output power was limited, like a sinusoid with the top edge cutted and increasing the bias and the number of output devices, the missing edge was appearing incresing so the output power.
But at this point i'm wondering: what makes this amplifier so worthwhile if it can't be projected to drive an hard load?
Anyways i'm reading the big aleph 3d but it's a really big mess sometimes....i'm tring to go little by little.
Since this morning i got up to page 39.
What i should say is.....accordingly on what i have read so far... in order to get the amp working on 4ohm...for instance...the guys there had to play with the bias.
I 'll read it better.
I would gladly attach my simulation on this forum with the circuit drawn with capture, if it wasn't for a rubber that just stolen yesterday my laptop out of my car...so i have to rebuild the circuit and re-simulate it....then i'l post the results.
In the menatime i'll keep reading the bi post trying to get a better handle on the subject!
Hi Stefano,
Please keep working on the circuit. I want to encourage you not discourage you. I am also very sorry to hear that your laptop was stolen.
In the example you give the output increased into 4 ohms because the voltage rails were limiting the output into 8 ohms. This another example of carefully balancing the design parameters of the circuit.
Graeme
Please keep working on the circuit. I want to encourage you not discourage you. I am also very sorry to hear that your laptop was stolen.
In the example you give the output increased into 4 ohms because the voltage rails were limiting the output into 8 ohms. This another example of carefully balancing the design parameters of the circuit.
Graeme
The Aleph-X circuit is capable of driving an arbitrarily low impedance as long as the bias is set high enough. As I've noted once or twice, the voltage sets the wattage capability and the bias tells you how low an impedance you can work into. Obviously, there are tradeoffs. Power dissipation gets ugly, fast. MOSFET reliability decreases. Cumulative Gate capacitance begins to be a burden, although different people will give you different answers as to when this happens.
I tried to make the Aleph-X scalable. How large or small you build it is up to you, but please do it intelligently. There are sundry spreadsheets that claim to assist in the process. I've looked at a few. I've seen some that make sense and others that make me wonder what the author was thinking. Simulations? I've said enough about simulations elsewhere. If you've got enough hours in your day that you can rationalize wasting them like that, I suppose it's better than spending the same amount of time at role-playing war games...but only barely.
The essential calculations for an Aleph-X can be done on a hand calculator in about thirty seconds...okay, let's be generous...two minutes. After that, you're over-analyzing the situation. You think you're getting deeper insight into the circuit, but in reality you're just spinning your wheels. Any illusion of progress is in your mind.
Grey
I tried to make the Aleph-X scalable. How large or small you build it is up to you, but please do it intelligently. There are sundry spreadsheets that claim to assist in the process. I've looked at a few. I've seen some that make sense and others that make me wonder what the author was thinking. Simulations? I've said enough about simulations elsewhere. If you've got enough hours in your day that you can rationalize wasting them like that, I suppose it's better than spending the same amount of time at role-playing war games...but only barely.
The essential calculations for an Aleph-X can be done on a hand calculator in about thirty seconds...okay, let's be generous...two minutes. After that, you're over-analyzing the situation. You think you're getting deeper insight into the circuit, but in reality you're just spinning your wheels. Any illusion of progress is in your mind.
Grey
aleph 1.2's to AX100
Grey, do you comments hold true for tranisent bursts of energy.
Example, I am now in the process of sizing a transformer for my Aleph 1.2 to XA100-160 conversion. I found some avel/lindbergs at 1kva @ 25vct. When I do the calculations assuming 32 volt rails and a total bias of 7amps I would get 64 x 7 or 448 watts plus 10% losses through the power supply or as GL said about
500 watts disipation at idle. This theorectically gives me another 500watts of transformer power for transients or just driving the load. This means as per NP that the amp can only drive 7amps additional into the load. DC watts (not rms or ac) would then be 64v x 14 amps or 856 watts total load plus bias.
Am I on the right track? My Aleph 1.2's use 2kva transformers and I am debating what size to get. Theorectially Gl is correct that I can get by with 1kva. I might go political and get two 800va per amp just to be safe (more money is all that is required).
But transient peaks would be protected. I still have the article on the Pass A75 which gives a great description on ps sizing.
dave
Grey, do you comments hold true for tranisent bursts of energy.
Example, I am now in the process of sizing a transformer for my Aleph 1.2 to XA100-160 conversion. I found some avel/lindbergs at 1kva @ 25vct. When I do the calculations assuming 32 volt rails and a total bias of 7amps I would get 64 x 7 or 448 watts plus 10% losses through the power supply or as GL said about
500 watts disipation at idle. This theorectically gives me another 500watts of transformer power for transients or just driving the load. This means as per NP that the amp can only drive 7amps additional into the load. DC watts (not rms or ac) would then be 64v x 14 amps or 856 watts total load plus bias.
Am I on the right track? My Aleph 1.2's use 2kva transformers and I am debating what size to get. Theorectially Gl is correct that I can get by with 1kva. I might go political and get two 800va per amp just to be safe (more money is all that is required).
But transient peaks would be protected. I still have the article on the Pass A75 which gives a great description on ps sizing.
dave
Quick rule of thumb for any of the Aleph variants, whether single-ended or bridged: Transformer VA = 6 x RMS power output. If you're looking for 100Wrms out, think in terms of a 600VA transformer. There's enough fudge factor that you need not worry if the transformer is 550-575VA, particularly if it's a good quality transformer.
All this fiddling down to fractions of a volt is wasted time. Why? Because calculating your anticipated VA needs to 1% is goofy when transformers only come in 10% increments.
When in doubt, round up. Life's too short to be anal about these sorts of numbers. Save it for calculating like RIAA eq in phono stages...something where it might be worth your while.
Grey
All this fiddling down to fractions of a volt is wasted time. Why? Because calculating your anticipated VA needs to 1% is goofy when transformers only come in 10% increments.
When in doubt, round up. Life's too short to be anal about these sorts of numbers. Save it for calculating like RIAA eq in phono stages...something where it might be worth your while.
Grey
aleph 1.2's to AX100
Grey, thanks, I will look into 600va xformers. Two per amp with one running the positive rail and one running the negative rail. This to make sure that they are not working to hard and causing the cores to vibrate. I will check with my scope the dc on the primary side of the 2kva transformer that I am using now. However the 350.5 doesn't buzz.
GL, I am assuming that your 800va xformers are quiet and that you get no 120hz buzz in the speakers when you listen at the cones?
GL, I reread NP's article Zen 8 where he goes into setting the Aleph ccs at 50% gain. In your thread when you talk about this and how you did it I am assuming that it was with respect to the input differential pair. I am assuming that the ccs for the outputs is set for the bias current. I am not sure which resistors to play with in that regard. I probably will start out using the same values as the Aleph 1.2. I am stuyding the schematic while laying on my lawn chairs in the sun with a beer or two. Too hot for wine.
I re read your thread to see what you did.
Grey, thanks, I will look into 600va xformers. Two per amp with one running the positive rail and one running the negative rail. This to make sure that they are not working to hard and causing the cores to vibrate. I will check with my scope the dc on the primary side of the 2kva transformer that I am using now. However the 350.5 doesn't buzz.
GL, I am assuming that your 800va xformers are quiet and that you get no 120hz buzz in the speakers when you listen at the cones?
GL, I reread NP's article Zen 8 where he goes into setting the Aleph ccs at 50% gain. In your thread when you talk about this and how you did it I am assuming that it was with respect to the input differential pair. I am assuming that the ccs for the outputs is set for the bias current. I am not sure which resistors to play with in that regard. I probably will start out using the same values as the Aleph 1.2. I am stuyding the schematic while laying on my lawn chairs in the sun with a beer or two. Too hot for wine.
I re read your thread to see what you did.
gl said:
Sorry, i haven't understood what you meant.
May you explain this concept, if possible, a little bit more?
Thanks,
Stefano.
Hi Dave,
There is no hum, no noise (no cones - Magnepans) with my ear to the speakers.
The 50% current gain is set for the output stage. Note the resistor designators quoted. They are all in the output stage. The CCS for the input diff pair is the exact same as in Greys original.
Hi Stefano,
The output power is dependent on the current available AND the voltage swing available from the power supply. In the example you show, the power supply the voltage rails aren't high enough to allow the output current to reach the level of the bias current. This is Ohms law in action.
Graeme
There is no hum, no noise (no cones - Magnepans) with my ear to the speakers.
The 50% current gain is set for the output stage. Note the resistor designators quoted. They are all in the output stage. The CCS for the input diff pair is the exact same as in Greys original.
Hi Stefano,
The output power is dependent on the current available AND the voltage swing available from the power supply. In the example you show, the power supply the voltage rails aren't high enough to allow the output current to reach the level of the bias current. This is Ohms law in action.
Graeme
aleph 1.2's to AX100
GL, I spoke to Meloney at Avel-lindberg xformers. She recomended their 1kva trannie with two 25vac secondaries rated at 20 amps each. If this is the case and I want a class A safety factor of two--7amps bias plus 7amps output plus losses at 10% this comes to 16 amps which leaves me a little head room. This also meets Grey's criteria of 6 x the output watts or 6 x160 = 960va. Since I will use this to biamp the tweeter section of my line arrays, I should be ok with the power. This is the cheapest solution. I can also stack two 625va's but that is $120 more.
dave
GL, I spoke to Meloney at Avel-lindberg xformers. She recomended their 1kva trannie with two 25vac secondaries rated at 20 amps each. If this is the case and I want a class A safety factor of two--7amps bias plus 7amps output plus losses at 10% this comes to 16 amps which leaves me a little head room. This also meets Grey's criteria of 6 x the output watts or 6 x160 = 960va. Since I will use this to biamp the tweeter section of my line arrays, I should be ok with the power. This is the cheapest solution. I can also stack two 625va's but that is $120 more.
dave
Hi Dave,
There are lots of "rules of thumb" for class A operation. Mine is to take the total idle power dissipation in watts and multiply by 2. Thats the nominal minimum. IIRC that was the recommendation given by NP way back in the A40 article. If you're a little over or under I don't think you'll hear a huge difference. That being said there are lots of people since then (particularly on this site) that claim that three times or more will give better sonic results.
Comparing the DC amps of bias with the AC amps out of the secondary is not comparing apples to apples. They aren't equivalent.
For your application 1000VA should be just fine. If I was in your place I'd do it.
Graeme
There are lots of "rules of thumb" for class A operation. Mine is to take the total idle power dissipation in watts and multiply by 2. Thats the nominal minimum. IIRC that was the recommendation given by NP way back in the A40 article. If you're a little over or under I don't think you'll hear a huge difference. That being said there are lots of people since then (particularly on this site) that claim that three times or more will give better sonic results.
Comparing the DC amps of bias with the AC amps out of the secondary is not comparing apples to apples. They aren't equivalent.
For your application 1000VA should be just fine. If I was in your place I'd do it.
Graeme
GL thanks, one decision down and on to the next. I have decided to use PT to PT wiring for the new circuit board and I hope to build the test cirucuit this weekend for testing the new diff front end xistors. I have a bunch left over from previous projects. dave
gl said:
I think you'll find that my rule of thumb leads to exactly the same conclusion as Nelson's. My approach has the advantage that people are more likely to know the RMS output of the amplifier than the quiescent power dissipation...and since the Aleph variants' power dissipation tends to run roughly three times the RMS power output, followed by a factor of two to keep it cool...thus by the math I learned at my mother's knee is 3 x 2 = 6, which tells me that the transformer VA should be about six times the RMS output.
As for going over on VA, there's an easily understood reason why it might produce audible results: The transformer secondary will have lower impedance. Anything that allows the amplifier circuit to access current faster and more easily is all to the good.
Of course, bigger transformers cost more, but that's up to the individual.
Grey
I agree; the final results are very close.
However, I stand on my comments. Using total dissipation is easier for me to figure and includes a number of variables that total output doesn't predict.
Graeme
However, I stand on my comments. Using total dissipation is easier for me to figure and includes a number of variables that total output doesn't predict.
Graeme
Sorry if i'm interferiring on this conversation!
I have problems to catch this rule of thumb of "6".
By finally adjusting and trying to trade rail voltage- power dissipation on each mosfet-frequency response - drivability ....i came up to the following configuration that i would like to share with you guys:
+-26V rail voltage for the power supply
32 mosfets per channel (8 mosfets per quadrant)
1.4A Bias on each mosfet in idle condition
36W each mosfet on idle total power dissipation (576W)
I will use 2 big heatsinks 300 X 200 mm coefficient 0.1C/W (for the monoblock, so it would be 8 mosfets on each heatsink, which is about 29C over the ambient temperature)
i can deliver on 8ohm 90Wpk - 180Wpk on 4 - 350W pk on 2 (undistorted) at 2.6Vpk at the input
if i rise the input up to 3.2V i can reach 180W pk on 8 ohm, but on lower load the output power is distorted at thi sinput level.
(i'll check this values later to make sure the exact values)
The point is....how big should the transformers be for the monoblock?
If i would use the rule of thumb of 6 times, i'd need solar panel and the niaagara's falls energy 🙂 ....well i'll set the amplifier by the falls....not a big deal 🙂
Just kidding!
the only thing tat pushed me to head toward this monster is the willing on having the beast driving a very hard load and i think this would be a good characteristic.
Could anybody give some good advice for power supply's issues?
I have problems to catch this rule of thumb of "6".
By finally adjusting and trying to trade rail voltage- power dissipation on each mosfet-frequency response - drivability ....i came up to the following configuration that i would like to share with you guys:
+-26V rail voltage for the power supply
32 mosfets per channel (8 mosfets per quadrant)
1.4A Bias on each mosfet in idle condition
36W each mosfet on idle total power dissipation (576W)
I will use 2 big heatsinks 300 X 200 mm coefficient 0.1C/W (for the monoblock, so it would be 8 mosfets on each heatsink, which is about 29C over the ambient temperature)
i can deliver on 8ohm 90Wpk - 180Wpk on 4 - 350W pk on 2 (undistorted) at 2.6Vpk at the input
if i rise the input up to 3.2V i can reach 180W pk on 8 ohm, but on lower load the output power is distorted at thi sinput level.
(i'll check this values later to make sure the exact values)
The point is....how big should the transformers be for the monoblock?
If i would use the rule of thumb of 6 times, i'd need solar panel and the niaagara's falls energy 🙂 ....well i'll set the amplifier by the falls....not a big deal 🙂
Just kidding!
the only thing tat pushed me to head toward this monster is the willing on having the beast driving a very hard load and i think this would be a good characteristic.
Could anybody give some good advice for power supply's issues?
Something about your math looks wonky, but I don't have time to backtrack and see what went wrong.
I hadn't intended my previous posts to be a reading comprehension test, but please note that I said RMS, not peak wattage.
If you're intending 26V rails, you're going to end up with something on the order of 120-130W into 8 Ohms. Six times 120W is 720VA for a normal Aleph variant. Done.
If you're obsessing over the numbers any further than that, I'm going to start thinking that you've got a sexual fetish about pushing buttons on calculators. If that's how you get your jollies, then be my guest, but you've left the realm of audio. Perhaps there's another website out there that caters to those who want to whisper sweet nothings whilst caressing their computers, but I don't know the URL.
I've said it before and I'll say it again...all this nitpicky silliness about calculating the VA down to the nearest .001 is for people who have more time than sense. Setting aside the possibility of a custom order transformer, you're not going to find a manufacturer who offers 700VA, 705VA, 710VA...what you'll find is something like 600VA, followed by 750VA. So you go with 750VA.
Now, your post indicates that you're wanting more output devices and more bias than normal. If that is truly your intention, you'll have to do your own math. It's not really all that difficult:
(26V rails) * (32 devices) * (1.4A bias) = 1165W power dissipation at idle.
Rough it off and call it a 2kVA transformer per channel. They're expensive, but if you're determined to build an Aleph-X to that spec, that's what you'll need.
It took all of twenty seconds to calculate and half of that was taken up by me having to stop and answer a question for a guy here at work.
See? Easy.
Grey
I hadn't intended my previous posts to be a reading comprehension test, but please note that I said RMS, not peak wattage.
If you're intending 26V rails, you're going to end up with something on the order of 120-130W into 8 Ohms. Six times 120W is 720VA for a normal Aleph variant. Done.
If you're obsessing over the numbers any further than that, I'm going to start thinking that you've got a sexual fetish about pushing buttons on calculators. If that's how you get your jollies, then be my guest, but you've left the realm of audio. Perhaps there's another website out there that caters to those who want to whisper sweet nothings whilst caressing their computers, but I don't know the URL.
I've said it before and I'll say it again...all this nitpicky silliness about calculating the VA down to the nearest .001 is for people who have more time than sense. Setting aside the possibility of a custom order transformer, you're not going to find a manufacturer who offers 700VA, 705VA, 710VA...what you'll find is something like 600VA, followed by 750VA. So you go with 750VA.
Now, your post indicates that you're wanting more output devices and more bias than normal. If that is truly your intention, you'll have to do your own math. It's not really all that difficult:
(26V rails) * (32 devices) * (1.4A bias) = 1165W power dissipation at idle.
Rough it off and call it a 2kVA transformer per channel. They're expensive, but if you're determined to build an Aleph-X to that spec, that's what you'll need.
It took all of twenty seconds to calculate and half of that was taken up by me having to stop and answer a question for a guy here at work.
See? Easy.
Grey
it's an italian manufacter...in order to order from them you need to have a company...i mean you can't order as a private person.
Stefanoo said:
I think what Jacco is trying to do is to help you determine if your data is correct.........so if I were you I'd provide a link 😉
Magura 🙂
GRollin...thanks for your reply....always very coloured thought.
Anyways.... since i came across at the same result 1.1KW dissipated on idle state...i was wondering what it would be 6 times this value...that's all 🙂
If you are telling me...that a 2KVA transormer will be fine.... (i was going to consider the 1.2KVA) ...i 'll go with this value.
I thought that for a class A the worse condition in power term would be set by the idle codition.
Anyways since Pout=Vload * Iload = Vload^2 / R = I^2 * R
please..correct me...if i'm saying something completely out of sense:
if 26V is the rail voltage, the output will swing, at maximum, at Vrail-Vsource ....so, it will be a little less than 26.... so by substituiting the values obtained i will obtain the output power.
To complete the reason, please correct it if it's wrong....so may be i can cach up eventual faluts on my line of reasonings.
if i want the amp to be able to drive a 4 or 2 ohm load i will have to deliver a current that it's I = V / R .
If i'm not wrong, the amp is able to deliver, at maximum, the bias current superimposed.
Therefore, i will have to bias the output stage, at least, a little higher than the value of current absorbed from the minimum value of load that i supposedly want to drive.
So i had to use more output devices, in order to trade this current demand and the power dissipation' s condition on each mosfets.
It's too late.....i'm probably saying such a stupid things..... so must better go off to sleep... 🙂
Anyways.... since i came across at the same result 1.1KW dissipated on idle state...i was wondering what it would be 6 times this value...that's all 🙂
If you are telling me...that a 2KVA transormer will be fine.... (i was going to consider the 1.2KVA) ...i 'll go with this value.
I thought that for a class A the worse condition in power term would be set by the idle codition.
Anyways since Pout=Vload * Iload = Vload^2 / R = I^2 * R
please..correct me...if i'm saying something completely out of sense:
if 26V is the rail voltage, the output will swing, at maximum, at Vrail-Vsource ....so, it will be a little less than 26.... so by substituiting the values obtained i will obtain the output power.
To complete the reason, please correct it if it's wrong....so may be i can cach up eventual faluts on my line of reasonings.
if i want the amp to be able to drive a 4 or 2 ohm load i will have to deliver a current that it's I = V / R .
If i'm not wrong, the amp is able to deliver, at maximum, the bias current superimposed.
Therefore, i will have to bias the output stage, at least, a little higher than the value of current absorbed from the minimum value of load that i supposedly want to drive.
So i had to use more output devices, in order to trade this current demand and the power dissipation' s condition on each mosfets.
It's too late.....i'm probably saying such a stupid things..... so must better go off to sleep... 🙂
- Status
- Not open for further replies.