Has anyone considered these? When I got the response back from Wayne, he said I might try the Aleph 2 instead of building three more Zens.
My only problem is those damn matched MOSFETs. The service manual says they need Vgs within 0.01 to 0.1V of each other. That sounds pretty tollerant. Does anyone have an idea of how close any two random MOSFETs will be?
Thanks,
pixie
My only problem is those damn matched MOSFETs. The service manual says they need Vgs within 0.01 to 0.1V of each other. That sounds pretty tollerant. Does anyone have an idea of how close any two random MOSFETs will be?
Thanks,
pixie
Check Pass site for "yield" on matching
As I recall from somewhere on the Pass site, I saw estimat of matching yield which indicated that if you buy 100 units you should get many large good sets.
From memory I think it was 8 sets of 10?
The more you have and the smaller sets you need, the closer to 100% yield you get. I am guessing now, but for the input stage, buying 10 units would probably give you 2-3 pairs.
Check out http://www.passlabs.com -- it is very nice.
As I recall from somewhere on the Pass site, I saw estimat of matching yield which indicated that if you buy 100 units you should get many large good sets.
From memory I think it was 8 sets of 10?
The more you have and the smaller sets you need, the closer to 100% yield you get. I am guessing now, but for the input stage, buying 10 units would probably give you 2-3 pairs.
Check out http://www.passlabs.com -- it is very nice.
Well let's see what Nelson Pass said about matching yield
Since there is actually at least one person who was interested, I decided to dig up the original article http://www.passlabs.com/projects/a75_2_7.htm
For output devices, the master said:
"Within a population of 150 transistors, you can easily get 12 sets [of 12 -- Petter's remark] matched to O.1V VGS at 200mA"
(12*12)/150 is near enough 100%
Note that you don't need to match between top/bottom, only parallell pairs. (((((For the Aleph, you can even get by without matching between top units by setting up drive circuitry for each individual transistor. This would also be possible (simpler in fact -- and preferable I am sure) for a true SE class A design driven by active current sources instead of resistors.)))))
What we need to do is to place a group order from a reputable manufacturer and get the devices matched for us. Does anybody know someone reputable willing to do this and handle payment/shipping?
[Edited by Petter on 02-04-2001 at 10:47 AM]
Since there is actually at least one person who was interested, I decided to dig up the original article http://www.passlabs.com/projects/a75_2_7.htm
For output devices, the master said:
"Within a population of 150 transistors, you can easily get 12 sets [of 12 -- Petter's remark] matched to O.1V VGS at 200mA"
(12*12)/150 is near enough 100%
Note that you don't need to match between top/bottom, only parallell pairs. (((((For the Aleph, you can even get by without matching between top units by setting up drive circuitry for each individual transistor. This would also be possible (simpler in fact -- and preferable I am sure) for a true SE class A design driven by active current sources instead of resistors.)))))
What we need to do is to place a group order from a reputable manufacturer and get the devices matched for us. Does anybody know someone reputable willing to do this and handle payment/shipping?
[Edited by Petter on 02-04-2001 at 10:47 AM]
I allways matched small signal bjt´s in amps and it WAS worth it and very cheap I can say.
When it comes to power output mosfets it isnt the cheapest thing to do. A 100 pieces cost a lot.
About the current source trick I have seen it in an amp working in class A single ended.
It is a good idea. You make separate drive circuits for each power MOSFET which could be cheaper then buying extra power MOSFETs. So I guess you just match the MOSFETs you want for the output and the rest you use in the current sources.
After comparing the Aleph amps I desided to make a version that has 3 output devices in parallel biased at 1.1 amp each at a voltage supply of +30 -30 Volts. I think it will deliver 50 Watts in 8 ohms and more power into 4 ohms.
Something between the aleph 3 and 5. Which is not 4 lol.
Aleph 4 is more powerfull then 5.
When it comes to power output mosfets it isnt the cheapest thing to do. A 100 pieces cost a lot.
About the current source trick I have seen it in an amp working in class A single ended.
It is a good idea. You make separate drive circuits for each power MOSFET which could be cheaper then buying extra power MOSFETs. So I guess you just match the MOSFETs you want for the output and the rest you use in the current sources.
After comparing the Aleph amps I desided to make a version that has 3 output devices in parallel biased at 1.1 amp each at a voltage supply of +30 -30 Volts. I think it will deliver 50 Watts in 8 ohms and more power into 4 ohms.
Something between the aleph 3 and 5. Which is not 4 lol.
Aleph 4 is more powerfull then 5.
Bias current
I'm also building a version of the Aleph series, but with some modifications.
I'm planning to put a switch with different resistor values to change the bias current of the amp.
Being a student and therefore living in a very small space, 300-400W can get really hot in the summer, I want to turn the bias current down when I'm not listening loud, and up ,for less distortion, when I want.
Does anyone have any idea what the formula for the value of the resistor vs bias current?
I think the resistor I want to change is R19 in the Aleph 5 schematics, and R26 in the Aleph 4. Is this correct?
-Kaj
I'm also building a version of the Aleph series, but with some modifications.
I'm planning to put a switch with different resistor values to change the bias current of the amp.
Being a student and therefore living in a very small space, 300-400W can get really hot in the summer, I want to turn the bias current down when I'm not listening loud, and up ,for less distortion, when I want.
Does anyone have any idea what the formula for the value of the resistor vs bias current?
I think the resistor I want to change is R19 in the Aleph 5 schematics, and R26 in the Aleph 4. Is this correct?
-Kaj
In the Aleph 4 and Aleph 5 you have to change R 40-42 to change the bias current. It isnt so easy to do I think. You need a bigger value for smaller bias current. The Voltage drop across those resistors must be 0.5 V so thats how you select the resistors. But you have to change all resistors not only the one the bias circuit takes voltage drop off.
OK?
But in the service manual for Aleph 5 it says:
"R19 is a fixed resistor which trims the DC current value. R21 and C10 adjust the current against output current as sensed by the voltage across R22-R27 and R100"
And in Aleph 3, where this guy http://www.institute.uni-bremen.de/~carnivor/aleph/aleph.html has done a modification to do the same, it says:
"R113 is a fixed resistor which trims the DC current value. R114 and C103 adjust the current against output current as sensed by the voltage across R124-R127"
Sounds pretty much the same to me, or am I way off?
-Kaj
But in the service manual for Aleph 5 it says:
"R19 is a fixed resistor which trims the DC current value. R21 and C10 adjust the current against output current as sensed by the voltage across R22-R27 and R100"
And in Aleph 3, where this guy http://www.institute.uni-bremen.de/~carnivor/aleph/aleph.html has done a modification to do the same, it says:
"R113 is a fixed resistor which trims the DC current value. R114 and C103 adjust the current against output current as sensed by the voltage across R124-R127"
Sounds pretty much the same to me, or am I way off?
-Kaj
I think it sounds ok.
I just read it in the site.
So if it works there why not.
It´s a good idea too usually when we listen to music it´s low. Specially when I am home working on the computer or just hanging around the music is mostly backround music.
So it´s a good idea to be able to turn the bias down when you want to.
Good luck with your project.
I hope I find the time and money to start soon.
I just read it in the site.
So if it works there why not.
It´s a good idea too usually when we listen to music it´s low. Specially when I am home working on the computer or just hanging around the music is mostly backround music.
So it´s a good idea to be able to turn the bias down when you want to.
Good luck with your project.
I hope I find the time and money to start soon.
Optimizing Alephs for speaker impedance and heat generation
Hi again, guys!
The Aleph and any other SE amp should be optimized for speaker impedance. This will also optimize heat generation.
Step 1: Choose impedance of speakers (you want lowest impedance under about 2-3KHz or 20KHz depending on how rabid you are). let us say this is 4 Ohms
Step 2: Choose required Amplifier Power. Let's say this is 64W RMS
Step 3: I1=sqrt(P/R)=4A RMS
Step 4: For an SE design with static current sources you need 4*sqrt(2)A standing current minimum. For an Aleph type you need 2*sqrt(2)A standing current minimum. For a resistor (Zen) type you need 8*sqrt(2)A standing current.
Step 5: Work out required voltage. This is (peak current or I*sqrt(2) ) in each direction so +-rails at that voltage. Rail to rail is thus 2*(peak current)*(minimum speaker impedance) for Aleph. A Zen model is 4 times less efficient. A static source driven is half as efficient as the Aleph.
Step 6: Work out power dissipation = V(rail to rail)*I_quiescent=P (always for Class A amps). I_quiescent is the standing current you have going through transistors.
So now you know that you need to change the voltage as well if you wish to make the units power efficient. This is also the reason why the max power of Alephs reduces with impedance load under the current limit. Normal amps only increase "power" with lower impedance until they blow up.
If this actually makes for the best sound is another matter. I leave it as an exercise to work out what speaker impedance Aleph amps are optimized for yourselves (hint look at current. You can also work this out backwards by looking at the specifications to see when the amplifier output power reduces with lower speaker impedance).
[Edited by Petter on 02-06-2001 at 11:46 AM]
Hi again, guys!
The Aleph and any other SE amp should be optimized for speaker impedance. This will also optimize heat generation.
Step 1: Choose impedance of speakers (you want lowest impedance under about 2-3KHz or 20KHz depending on how rabid you are). let us say this is 4 Ohms
Step 2: Choose required Amplifier Power. Let's say this is 64W RMS
Step 3: I1=sqrt(P/R)=4A RMS
Step 4: For an SE design with static current sources you need 4*sqrt(2)A standing current minimum. For an Aleph type you need 2*sqrt(2)A standing current minimum. For a resistor (Zen) type you need 8*sqrt(2)A standing current.
Step 5: Work out required voltage. This is (peak current or I*sqrt(2) ) in each direction so +-rails at that voltage. Rail to rail is thus 2*(peak current)*(minimum speaker impedance) for Aleph. A Zen model is 4 times less efficient. A static source driven is half as efficient as the Aleph.
Step 6: Work out power dissipation = V(rail to rail)*I_quiescent=P (always for Class A amps). I_quiescent is the standing current you have going through transistors.
So now you know that you need to change the voltage as well if you wish to make the units power efficient. This is also the reason why the max power of Alephs reduces with impedance load under the current limit. Normal amps only increase "power" with lower impedance until they blow up.
If this actually makes for the best sound is another matter. I leave it as an exercise to work out what speaker impedance Aleph amps are optimized for yourselves (hint look at current. You can also work this out backwards by looking at the specifications to see when the amplifier output power reduces with lower speaker impedance).
[Edited by Petter on 02-06-2001 at 11:46 AM]
McKajVah,
I downloaded the circuits (Aleph 4 and 5) and, at a first glance, they seem to have the quiescent current controlled by Ib and HFE of the MPSA18 (Aleph 5 - Q5) (VCE=34V-6.25K*HFE*Ib), where Vce is aprox. equals VGS of IRF244 and times GM of this device equals Isourcequiescent.
But the designer says R19 (wich controls Ib with R20) is only for trimming, and not for big changes in the quiescent current. He says the circuit has many of their characteristcs controlled by characteristics of the components and there are no adjustments to be done, as usual in simple circuits like these; they can't allow big changes.
Regards
I downloaded the circuits (Aleph 4 and 5) and, at a first glance, they seem to have the quiescent current controlled by Ib and HFE of the MPSA18 (Aleph 5 - Q5) (VCE=34V-6.25K*HFE*Ib), where Vce is aprox. equals VGS of IRF244 and times GM of this device equals Isourcequiescent.
But the designer says R19 (wich controls Ib with R20) is only for trimming, and not for big changes in the quiescent current. He says the circuit has many of their characteristcs controlled by characteristics of the components and there are no adjustments to be done, as usual in simple circuits like these; they can't allow big changes.
Regards
Petter,
If you change the quantity of output devices you must deal with the load impedance, Vcc and/or the source resistors to maintain the performance.
As you can see in Aleph 5 and 4, the designer has changed VCC and Rsource to change the power using the same load impedance on both designs.
Regards
If you change the quantity of output devices you must deal with the load impedance, Vcc and/or the source resistors to maintain the performance.
As you can see in Aleph 5 and 4, the designer has changed VCC and Rsource to change the power using the same load impedance on both designs.
Regards
McKajVah,
I think you can use the following equations for a initial setup of bias resistors (R26 and R19).
R26=1000*(11.5 - 0.5*Iq)/(.780 - 0.5*Iq) ohms
R19=1000*(8 - .33*Iq)/(.711 - .33*Iq) ohms
They are dependent on hFE and Vbe of ZTX450 and MPSA18. The transconductance of IRF244 is important too.
I used datasheets from Harris, Zetex and Fairchild to get the data.
The Iq must be lower than 2.15A for Aleph 5 and 1.5A for Aleph 4. It's because Rsource is defined so as to the voltage over it.
Anyway, the performance of the amplifier will suffer, because Vcc, Rsource and Rload weren't changed. I think this approach is valid only for little Iq changes over the value defined by the designer.
I wish it could help you
Regards
I think you can use the following equations for a initial setup of bias resistors (R26 and R19).
R26=1000*(11.5 - 0.5*Iq)/(.780 - 0.5*Iq) ohms
R19=1000*(8 - .33*Iq)/(.711 - .33*Iq) ohms
They are dependent on hFE and Vbe of ZTX450 and MPSA18. The transconductance of IRF244 is important too.
I used datasheets from Harris, Zetex and Fairchild to get the data.
The Iq must be lower than 2.15A for Aleph 5 and 1.5A for Aleph 4. It's because Rsource is defined so as to the voltage over it.
Anyway, the performance of the amplifier will suffer, because Vcc, Rsource and Rload weren't changed. I think this approach is valid only for little Iq changes over the value defined by the designer.
I wish it could help you
Regards
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.