There are about 5 things to consider when going for the evil twin of the Aleph 3. After the changes are made we will find that our A3s have specs similar to its older brother the Aleph 5. We get to keep the toasty DC bias of the A3 only toastier. I think this is what keeps the magic in the Aleph 3. Here’s how we can do it:
1. We need to have a beefier PS transformer. About 600-800VA; 2-28V secondary CT. This will deliver about +/-34Vdc loaded with the A3Turbo.
2. We need to increase the built-in current limiter’s current limit. As called in the A3 schematic, the limit is about 4Aac. We will increase this to about 5Aac. The limiter circuit is made up of Q104, R109 and R110. By lowering R110 to about 75Ù the current limit increases to about 5Aac. How it works: At lower load impedances, by virtue of Nelson Pass’s another paradox the Aleph dynamic constant current source, higher voltages and currents develop across and through the output devices and their Source resistors. At about 5Aac the voltage across R122 increases to about 2.35Vac. The voltage divider circuit that includes R109 will drop some of this voltage while about 0.6V develops across R110. 0.6V across the B-E of Q104 will be enough to turn it on limiting input currents to the Gates of the output MOSFETs thereby setting their output currents to a maximum of 5Aac.
3. For the positive going signal to keep up with the increase in negative going signal as the current limit was increased, the 66% of AC current gain (ACIG) as set by R114 will have to be increased. We will set the ACIG to about 78% by replacing R114 with about 634Ù of resistance. Both + and – signals will now start to distort almost at the same time.
4. If we are going to load up our A3Turbo with speakers that dip to 4Ù or 2 Ù at a bunch of frequencies in the audio band you might want to monitor your Source resistors if they heat up rapidly. We should replace them with a higher wattage ones if it did. Start with 5W.
5. As we did with the Source resistors, monitor our heatsinks. The touch and go method as explained by Mr. Pass is effective as it is.
Let’s fire it up. Oops! Nice output signal. That’s SIMetrix though. I’ve tried this at home but only on a computer. You can put into hardware if you find it appealing.
Please feel free to add to it if I'm missing something.
_____________
"For sounding good, I'll take the heat."
1. We need to have a beefier PS transformer. About 600-800VA; 2-28V secondary CT. This will deliver about +/-34Vdc loaded with the A3Turbo.
2. We need to increase the built-in current limiter’s current limit. As called in the A3 schematic, the limit is about 4Aac. We will increase this to about 5Aac. The limiter circuit is made up of Q104, R109 and R110. By lowering R110 to about 75Ù the current limit increases to about 5Aac. How it works: At lower load impedances, by virtue of Nelson Pass’s another paradox the Aleph dynamic constant current source, higher voltages and currents develop across and through the output devices and their Source resistors. At about 5Aac the voltage across R122 increases to about 2.35Vac. The voltage divider circuit that includes R109 will drop some of this voltage while about 0.6V develops across R110. 0.6V across the B-E of Q104 will be enough to turn it on limiting input currents to the Gates of the output MOSFETs thereby setting their output currents to a maximum of 5Aac.
3. For the positive going signal to keep up with the increase in negative going signal as the current limit was increased, the 66% of AC current gain (ACIG) as set by R114 will have to be increased. We will set the ACIG to about 78% by replacing R114 with about 634Ù of resistance. Both + and – signals will now start to distort almost at the same time.
4. If we are going to load up our A3Turbo with speakers that dip to 4Ù or 2 Ù at a bunch of frequencies in the audio band you might want to monitor your Source resistors if they heat up rapidly. We should replace them with a higher wattage ones if it did. Start with 5W.
5. As we did with the Source resistors, monitor our heatsinks. The touch and go method as explained by Mr. Pass is effective as it is.
Let’s fire it up. Oops! Nice output signal. That’s SIMetrix though. I’ve tried this at home but only on a computer. You can put into hardware if you find it appealing.
Please feel free to add to it if I'm missing something.
_____________
"For sounding good, I'll take the heat."
Hi,
just a few things:
1. if you up the voltage from 25 to 34 you´ll have an increase in power dissipation of almost 40%. Watch your fet temperatures!
2. Such a high voltage with such a low peak current will only be suitable for loudspeakers of 8 ohms or more
3. 78% ac-current-gain will probably not be very pleasing to the ear......
4. The mean current through the source resistors will not change with different speaker loads so no need to change source resistors.
William
just a few things:
1. if you up the voltage from 25 to 34 you´ll have an increase in power dissipation of almost 40%. Watch your fet temperatures!
2. Such a high voltage with such a low peak current will only be suitable for loudspeakers of 8 ohms or more
3. 78% ac-current-gain will probably not be very pleasing to the ear......
4. The mean current through the source resistors will not change with different speaker loads so no need to change source resistors.
William
On the wanted list:
For hotrodding & optimal adjustment, would anybody please post a circuit that can use say a automotive amp-meter with needle in centre, (like the X-series meter), so bias / current gain can be tuned "in flight".
As many may have experienced, your 8 ohm speakers are most often some other value...
Arne K
For hotrodding & optimal adjustment, would anybody please post a circuit that can use say a automotive amp-meter with needle in centre, (like the X-series meter), so bias / current gain can be tuned "in flight".
As many may have experienced, your 8 ohm speakers are most often some other value...
Arne K
Hi Uli,
I had wanted to keep the A3 DC bias current...which seems to be the magic number among the Alephs. And also to keep the power dissipation of the mosfets to <50W.
Hi WuffWaff,
"1. if you up the voltage from 25 to 34 you´ll have an increase in power dissipation of almost 40%. Watch your fet temperatures!"
Yes...at 36% increase the fearless DIYer might go for it.
"2. Such a high voltage with such a low peak current will only be suitable for loudspeakers of 8 ohms or more"
As I stated, its specs will be similar to the Aleph 5.
"3. 78% ac-current-gain will probably not be very pleasing to the ear......"
A great number of people love the Aleph 5's sound at 78%.
"4. The mean current through the source resistors will not change with different speaker loads so no need to change source resistors."
Okay..now it'll be an easier upgrade.
I had wanted to keep the A3 DC bias current...which seems to be the magic number among the Alephs. And also to keep the power dissipation of the mosfets to <50W.
Hi WuffWaff,
"1. if you up the voltage from 25 to 34 you´ll have an increase in power dissipation of almost 40%. Watch your fet temperatures!"
Yes...at 36% increase the fearless DIYer might go for it.
"2. Such a high voltage with such a low peak current will only be suitable for loudspeakers of 8 ohms or more"
As I stated, its specs will be similar to the Aleph 5.
"3. 78% ac-current-gain will probably not be very pleasing to the ear......"
A great number of people love the Aleph 5's sound at 78%.
"4. The mean current through the source resistors will not change with different speaker loads so no need to change source resistors."
Okay..now it'll be an easier upgrade.
Hi William,
Please refer to A5 schematic. Here's how I got the numbers:
ACIG=((0.1175/0.333)*(1000/453))*100=77.9%
Here's how I verified it using voltage divider formula at the current limiter circuit:
Vac at ref Source resistor=0.6/(75/(75+221))=2.368V
Imax=2.368V/1=2.368A
ACIG=(1-(0.5/2.368))=78.9% the slight difference in percentages is our assumed 0.6V of turn-on voltage for Q4. It looks more like about 0.57V.
The A5's ACIG is maxed out with it's current limit to let the + and - signals start to distort almost at the same time.
Please point out any miscalculation I made or missed.
Thanks,
Allan
Please refer to A5 schematic. Here's how I got the numbers:
ACIG=((0.1175/0.333)*(1000/453))*100=77.9%
Here's how I verified it using voltage divider formula at the current limiter circuit:
Vac at ref Source resistor=0.6/(75/(75+221))=2.368V
Imax=2.368V/1=2.368A
ACIG=(1-(0.5/2.368))=78.9% the slight difference in percentages is our assumed 0.6V of turn-on voltage for Q4. It looks more like about 0.57V.
The A5's ACIG is maxed out with it's current limit to let the + and - signals start to distort almost at the same time.
Please point out any miscalculation I made or missed.
Thanks,
Allan
Hi Blues,
the A5 will have a bit more than 1.5A. Mine had over 2A with the specified parts.
I just wanted to tell you that 78% ac-current-gain won´t sound very nice so be prepared to dial in some more bias.
William
P.S. I still don´t understand peoples obsession with symetrical clipping.......
the A5 will have a bit more than 1.5A. Mine had over 2A with the specified parts.
I just wanted to tell you that 78% ac-current-gain won´t sound very nice so be prepared to dial in some more bias.
William
P.S. I still don´t understand peoples obsession with symetrical clipping.......
Blues said:
We don't seem to be operating from the same schematic, but:
The first formula does not include the limited gain of Q5 circuit
and the transconductance of the output devices.
The result is that the current source gain is less than your
calculation (I arrived at the original figures by comparing the
actual AC current through the output vs the current through
the current source and not through calculation).
Regarding the limiting circuit, the original circuit already used 75
ohms in the divider, so I'm not sure why you need to change
the resistor. The original limiting on the negative side is about 7
amps peak, which is your 5 amp average figure.
Or have I miscalculated?
Nelson Pass said:
We don't seem to be operating from the same schematic, but:
The first formula does not include the limited gain of Q5 circuit
and the transconductance of the output devices.
The result is that the current source gain is less than your
calculation (I arrived at the original figures by comparing the
actual AC current through the output vs the current through
the current source and not through calculation).
Regarding the limiting circuit, the original circuit already used 75
ohms in the divider, so I'm not sure why you need to change
the resistor. The original limiting on the negative side is about 7
amps peak, which is your 5 amp average figure.
Or have I miscalculated?
Mr. Pass,
The schematic is from the Aleph 5 manual. I don't know how to factor in the transconductance and Q5 gain into the formula as you said. How do we do that?
The change to 75ohms is for the Aleph 3 which originally has 100ohms as we are trying to "hotrod" the A3 as the thread title referred.
William,
We can also keep the ACIG to 60-68% on the A3T and just increase a bit the current limit if you are concerned with the rather high ACI gain I specified. We might not match the power specs of A5 though but maybe close.
Symmetrical clipping or close to it is one way of getting all the juice from a fruit without getting the seeds
You're right - I looked at the A5 manual. duh.
In any case, I don't bother to factor in the npn gain or the
transconductance - I just adjust the resistance feeding the
base of Q5 until I get what I want. The gain of the current
source is simply measured as a percentage of what goes
through the current source versus what goes to the load.
You can set this arbitrarily, but 50% is the most optimal viewed
from an efficiency and performance standpoint. If you go higher
or lower, you should increase the bias to compensate. At twice
the bias, you can raise the gain to 100% or reduce it to 0%.
100% measures better than 0%, but 0% sounds better.
As usual, the best solution lies in the middle path.
In any case, I don't bother to factor in the npn gain or the
transconductance - I just adjust the resistance feeding the
base of Q5 until I get what I want. The gain of the current
source is simply measured as a percentage of what goes
through the current source versus what goes to the load.
You can set this arbitrarily, but 50% is the most optimal viewed
from an efficiency and performance standpoint. If you go higher
or lower, you should increase the bias to compensate. At twice
the bias, you can raise the gain to 100% or reduce it to 0%.
100% measures better than 0%, but 0% sounds better.
As usual, the best solution lies in the middle path.
Okay. Thanks!! I'll try to thread the middle path.
For the DIY Aleph 3 owners who have over-heatsunk we still can get about 60Wrms into 8ohms by just changing the power supply rails to +/- 34V. We keep the DC bias, the current limit as is at about 4A and the ACIG at about 66% (as per basic formula). By decreasing R114 from 750ohms to 732ohms (68% ACIG) we can maybe make it to 75Wrms at 4ohms. Give it a listen and decide if the increase to 68% is worth it.
If you want to experiment with different ACIGs =(((Rtotal 124-127)/(Rtotal 120-121))*(R115/R114))*100% refer to A3 for component designations.
Increasing the rails is an old trick for more power. Next time we'll try for more DC bias and rails for a more legit A3 Turbo, capable at lower impedances while staying in the middle path too...
For the DIY Aleph 3 owners who have over-heatsunk we still can get about 60Wrms into 8ohms by just changing the power supply rails to +/- 34V. We keep the DC bias, the current limit as is at about 4A and the ACIG at about 66% (as per basic formula). By decreasing R114 from 750ohms to 732ohms (68% ACIG) we can maybe make it to 75Wrms at 4ohms. Give it a listen and decide if the increase to 68% is worth it.
If you want to experiment with different ACIGs =(((Rtotal 124-127)/(Rtotal 120-121))*(R115/R114))*100% refer to A3 for component designations.
Increasing the rails is an old trick for more power. Next time we'll try for more DC bias and rails for a more legit A3 Turbo, capable at lower impedances while staying in the middle path too...
Mr. Pass,
My design goals for a higher power A3 is to keep the existing A3 circuit, increase DC bias and voltage rails and maybe make it to about 60Wrms into 8 ohms and about 85Wrms or so into 4 ohms. I'll just mess with the output/CCS circuit. Do you see any problems achieving these goals? Aside from excessive heat and running output devices to the brink will there be other problems? I'll heed your advise of keeping mosfets at <50W dissipation.
Do you think I'll be able to keep the A3's signature sound doing this?
Thanks for your continuing support.
My design goals for a higher power A3 is to keep the existing A3 circuit, increase DC bias and voltage rails and maybe make it to about 60Wrms into 8 ohms and about 85Wrms or so into 4 ohms. I'll just mess with the output/CCS circuit. Do you see any problems achieving these goals? Aside from excessive heat and running output devices to the brink will there be other problems? I'll heed your advise of keeping mosfets at <50W dissipation.
Do you think I'll be able to keep the A3's signature sound doing this?
Thanks for your continuing support.
Here’s my second take on the A3T. My goals for the A3T is for it to keep its Aleph 3 circuit, increase rail voltages and bias the heck out of it, get about 60W at 8ohms and possibly more at 4 ohms. For those wanting more power and hopefully keeping the sweet A3 sound, here’s an option worth trying out.
1. The PS transformer will need to be changed-out to one rated at about 800VA with 2-28V secondary center-tapped per channel. This will yield about +/-34V on the PS rails.
2. Increasing the bias current to about 2.7A will require replacing the Source resistors to a lower value. Each set of CCS and output circuit (two in the A3/ch), about 1.35A of bias will be available. I chose this value so we won’t have reliability issues like if we went over 50W of dissipation on each of the Mosfets. In this case dissipation will be about 46W. If you’ll be using reference NPN transistors with a rather high-ish turn-on voltage, Vbe=0.7V I’d recommend 0.51 ohm Source resistors. ZTX450s have about 0.6Vbe so Source resistors will have to be about 0.455 ohm or two 0.91ohm resistors in parallel. I think MPSA18s have about 0.5Vbe so it’ll have to be 0.375 ohm or 0.75//0.75. Use 3W resistors.
3. To maximize this increase in bias current we will increase the current limit. Based on the subjective experience of Mr. Pass and WuffWaff, among others, an acceptable value for an AC current gain is between 50% to 65%. I choose 60% so as to yield about 7A in current limit for a 2.7A bias. Replacing R110 with 140 ohms will yield just that on the output circuit.
4. On the CCS circuit we will adjust for 60% ACIG. By replacing R114 with 845 ohms the CCS circuit will now work with the output circuit to symmetrically clip.
5. As with all Alephs, good heatsinking is a wise investment. On these 150W rated mosfets, Mr. Pass’s recommends working them at <50W each. With 4 at about 46W each, a channel of A3T will be burning almost 200W/ch. So a substantially finned heatsink rated at about 0.15C/W is recommended for a 30C rise at an ambient temp of 25C.
The A3T based on the simulations I made, make it to 60W/8ohms and 85W/4ohms. It’s Class A all the way to its bias current. Beyond that it’s Class AB.
It has been another learning experience for me. Thanks for all your inputs and comments.
1. The PS transformer will need to be changed-out to one rated at about 800VA with 2-28V secondary center-tapped per channel. This will yield about +/-34V on the PS rails.
2. Increasing the bias current to about 2.7A will require replacing the Source resistors to a lower value. Each set of CCS and output circuit (two in the A3/ch), about 1.35A of bias will be available. I chose this value so we won’t have reliability issues like if we went over 50W of dissipation on each of the Mosfets. In this case dissipation will be about 46W. If you’ll be using reference NPN transistors with a rather high-ish turn-on voltage, Vbe=0.7V I’d recommend 0.51 ohm Source resistors. ZTX450s have about 0.6Vbe so Source resistors will have to be about 0.455 ohm or two 0.91ohm resistors in parallel. I think MPSA18s have about 0.5Vbe so it’ll have to be 0.375 ohm or 0.75//0.75. Use 3W resistors.
3. To maximize this increase in bias current we will increase the current limit. Based on the subjective experience of Mr. Pass and WuffWaff, among others, an acceptable value for an AC current gain is between 50% to 65%. I choose 60% so as to yield about 7A in current limit for a 2.7A bias. Replacing R110 with 140 ohms will yield just that on the output circuit.
4. On the CCS circuit we will adjust for 60% ACIG. By replacing R114 with 845 ohms the CCS circuit will now work with the output circuit to symmetrically clip.
5. As with all Alephs, good heatsinking is a wise investment. On these 150W rated mosfets, Mr. Pass’s recommends working them at <50W each. With 4 at about 46W each, a channel of A3T will be burning almost 200W/ch. So a substantially finned heatsink rated at about 0.15C/W is recommended for a 30C rise at an ambient temp of 25C.
The A3T based on the simulations I made, make it to 60W/8ohms and 85W/4ohms. It’s Class A all the way to its bias current. Beyond that it’s Class AB.
It has been another learning experience for me. Thanks for all your inputs and comments.
There would be advantages to doing this to an Aleph 3 but you are REALLY pushing the limits on the output devices and dissipation of them. While the heat sinks may be able to dissipat that much heat the outputs cannot do it reliably over long periods. So then you add outputs so thise in ther don't start to cry and you are at an Aleph 5!
Mark
Mark
Guys!...Where's your sense of adventure? Isn't this what DIY audio is all about? Discovering NP's designs was an adventure...let us push on and discover what's at the edge and learn at the same time.
"Kilowatt"...the word wouldn't have been coined if people hadn't pushed on.
Aleph 5? Well that's for another thread...A5 Turbo will be the thread name.
"Kilowatt"...the word wouldn't have been coined if people hadn't pushed on.
Aleph 5? Well that's for another thread...A5 Turbo will be the thread name.
Mark,
I'm rambling about the limits of the Aleph 3 circuit. Aleph 5 adds another parallel set of CCS and output circuit. You just have to look as far as the Zen v4 biased with 2A...if Q1's Drain is adjusted for half the rails Q1 and Q2 will be at 44W of dissipation. Mine is at 24V at the drain so Q1 is at 48W. My amp operating at what I think is not enough heatsinking (60C @2hrs) has every bit been damn reliable. The gunfight on the movie "Open Range" was very scary.
This effort of doing the math just proves the scalability of the Aleph 3 design as is. One day I'll build it. Maybe I'll smoke a few IRFs but as always with DIY it's lifetime warranty
4A of total bias on the A3 by the way will take you to 30Wrms pure Class A into a 4-ohm load. No need to raise the rails...that's A3 on top fuel!
I'm rambling about the limits of the Aleph 3 circuit. Aleph 5 adds another parallel set of CCS and output circuit. You just have to look as far as the Zen v4 biased with 2A...if Q1's Drain is adjusted for half the rails Q1 and Q2 will be at 44W of dissipation. Mine is at 24V at the drain so Q1 is at 48W. My amp operating at what I think is not enough heatsinking (60C @2hrs) has every bit been damn reliable. The gunfight on the movie "Open Range" was very scary.
This effort of doing the math just proves the scalability of the Aleph 3 design as is. One day I'll build it. Maybe I'll smoke a few IRFs but as always with DIY it's lifetime warranty
4A of total bias on the A3 by the way will take you to 30Wrms pure Class A into a 4-ohm load. No need to raise the rails...that's A3 on top fuel!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.