Aleph-X: High-Power Version

Status
Not open for further replies.
Scaling the Aleph-x

Blitz,

Thanx for the info. In fact calculating the value of source resistors as you mentioned should not be a problem as Grey 's formula is very straight forward. In fact, what I'm curious are the values of the OTHER RESISTORS. If you run the circuit at 27.7V(assuming its correct!), it's still more than double the original supply voltage of +/- 12V. If you compare the resistor values of the Aleph 30 which runs at +/- 25V, you will notice that some of the other resistors are slightly different from the original Aleph-X. So if we scale the power supply to +/- 27.7V @ 8.6A, do we just apply the higher voltage with the correct source resistors values only or do we have to scale the other resistors as well?

ckt 😕
 
Blitz said:
I have read Aleph-X threats now for 8 hours and I have to admit that I am not yet through with everything.


I have lots of the same questions. Perhaps someone with a lot of knowledge of the X design can make a cook book.

As input parameter; X watts at Y ohms, number of fets and a few (or a lot) of calculations for the volts, amps of bias, ohms of resistors, watts per fet etc. comes out (excel sheet?) .

This will save us a lot of reading and like me I can use all the help I can get. The Aleph-X thread is too long and has too much non-aleph-X info in it. For my Aleph 4 I used mark finnish his doc a lot.

Perhaps a volunteer?

Edwin
 
Edwin,

This would be great, but I guess that is more than we can expect.

That's why I started this threat: Clarify the open upscaling questions and once this is clear, it easy to make the cals like I started it above. In parallel I am now reading through the Aleph-X-Threat (440 Word-pages) and delete everthing which is only noise / raw information, but does not influence the schematic of the amp. ONce I have a consolidated version of this thrat, I am happy to share this with you.

Best REgards
 
From Mr. Pass:

"The XA200 has a CRCRC (cap/thermistor/cap/thermistor/cap)
filter for the power supply. At about an 8 amp bias, the
remaining ripple is about .02 volts."

Interesting...especially the 8 amps...are these per channel or per side ? I guess per side: If he has 32 mosfets in total, 16 per channel, 8 are current sources, 8 are output devices, 2 sides with 4 outputdevices each, each draws 2 A (most of the data comes from the owner's manual), this would mean 8 A bias for each crcrc-filter, meaning that we have 16 A in one channel, meaning that Grey's calcs on current are wrong, and Audiofreak is right.

Any thoughts ?
 
Sorry guys, just found this from Mr. Pass on the xa-200:


Posted by Nelson Pass on 09-11-2002 08:00 PM:
"Been covered before. 4 banks of 8 matched devices."

So, a xa-200 has per channel 32 mosfets. Mmmmmh. I hope I understand my mistake now: The output is like a circutron, Therfore the current of the two sides don't add each other...When I have 32 mosfets, it come down to 8 output fets per side...each should than only draw 1 A instead of two. The pass-side says 2 A .. I am getting confused. Any thoughts ?
 
Blitz said:
I am now reading through the Aleph-X-Threat (440 Word-pages) and delete everthing which is only noise / raw information, but does not influence the schematic of the amp. ONce I have a consolidated version of this thrat, I am happy to share this with you.
Blitz, good idea. Could you post this on this Wiki?
 
power calculation of an X

Ok,

I´ll try.

First it is very important not to confuse effective and peak (x2^0.5) values cause this makes things very complicated.

Lets say you want 200 watts in 8 ohms this means 40 volts and 5 Amps both effective values.
This is independent of class a,b,c,d,t or x.

Now you must calculate the peak values voltage = 40x 1.41=56.56
current = 5 x 1.41 = 7.07A
These are the values an amp must be able to deliver into an 8 ohm load.

Now for an Aleph X:

Since the amp is sort of bridged each output has to deliver half the voltage wich in this case is 56,56/2=28,28volt. To allow for losses driving the fets and over the source resistors the supply voltage has to be about 2 to 3 volts higher so + and - 30 to 31 volts.

The current that has to be delivered goes through both amps so it still is 7.07 amps. But since you have two amps total current is 14.14 amps.
Nice thing about the active current source is that it can actually double it´s output so total bias for an Aleph X is 14,14/2 = 7,07 amps.

This amp will dissipate 7,07 x 31 x 2 = 438 Watts for one channel.

More dissipation will be in the transformers, diodes, chokes etc. so 500-550 watts per channel is not unrealistic.

This means the transformer has to be about 1000-1500VA depending on personal taste. The secondary voltage depends a lot on wich diodes and/or chokes are used but won´t be much less than the supply voltage (so 28-31 volts). It is very important to calculate this before ordering!

I think it equally important to read Pass Labs user manuals very carefully and look at all the specs as nice but not neccessarily precise information.😎

william
 
Yes. That's it.

Thanks for this summary. After now going to numerous threats, I know that this is the coreect way to calculate it and you have even made us aware of other variables in the equasion. Thanks a lot.

Before I now calcualte the new, right values for the different setups (for the lazy bones): What would be your recommendation on the number of mosfets ? Let's take your example: 200 Watts in 8 ohms:

Power Dissipation only in the mosfets would be: 28,3*7,1*2= 400 Watt. When using 16 Mosfets (Let's assume 044Ns), we have 25 Watt / device at 28,3 V and 0,88 A, Correct ?

Now, from a power dissipation point of view this seems to be feasable. I read somewhere that the fets sound better the more current we put through them tough. Some put 2 A amps through them. Is it really the current which make them sound better or the power dissipation ?

Theoretically this example should result in the same values the XA-200 has. Well, still the 2 A through each Mosfets is not feasible yet or will the peak current be 0,88A * 2 ( I thought the factor 2 was already inculded in the equation above )?

Best Regards
 
William,
Nice and concise! It's about time someone posted that...
I have been reading the x thread from the beginning, and it took me a while to sort through the calculations on my own... it's sort of a tricky circuit- with the active current sources and the bridged output stage.
One thing that nobody seems to talk about when considering dissipation is how much it will go up under normal operation when the active current sources "kick in". I assume that most listening will be using around 1-2 watts maximum, and that the peak demands will be very short in duration. Any thoughts on this?
I have calculated my heatsinking to have a 19° C temp rise at idle, but I'm wondering how much heat I will really see in actual operation.
My amp will be a 40 watt/ 8 ohm- 12 output devices / channel... max of about 60W- current limited into lower impedances. Can't wait to see some of the higher powered versions! Unfortunately, my heat sinks just can not handle any more, or I would crank the bias up.
-NS
 
NS,

dissipation in the output fets is at it´s highest when no power is delivered. So actually the amp will cool down when delivering real power (this power is used to heat up your speakers).

With the X the two active current sources will work together in such a way that when one doubles the current, the other one will shut of so total current draw will be constant all of the time independent of input signal and/or load.

william
 
William,
I did not know that about the current sources! Thanks for the info...
I am tempted to increase the bias now, considering this new information. My speakers have a minimum impedance of about 3.2 ohms, so any increase in bias will help me out. Thankfully, they are about 92.5 db / 1 watt, so I don't need a whole lot of power.
-NS
 
Blitz,

yes, power dissipation per fet is roughly total dissipation/number of fets so in this case about 438/16 = 27.4 watts per fet (the error made here is total bias/(total fets) x (source resistor value).

I don´t think 27 watts per fet is too much but it depends on heatsink temperature and method of mounting the fets. I like to keep dissipation per fet at 20-25 watts. This allows for quite high heatsink temperatures (summer....).

Also the more fets the lower the voltage drop over the source resistors giving a bit higher efficiency and more power into lower impedances.

william
 
Here I found an alternative way of calculationg the bias, which basically seems to adress the changing impedance with frequency of most speakers, but results in much higher bias as calculated above I guess (Upto 8/3 times higher if you calculate for a 8 ohm speaker with a 3 ohm minimum) to keep full power over the whole frequency band (while we could question that assumption as I guess most speakers have there minimum at the lower end where they need real power):

"Now, if you used 8 ohms as your reference load, then there are a couple of different ways of going about the calculations for lower load impedances, depending on what you want to achieve. They will give different results... let me elaborate:

You have two choices:
1. set up the amp for 8 ohm loads, then calculate what kind of power you will be able to get into a 4 or 2 ohm load.
2. Set up the voltage rails for the desired peak power into 8 ohms, then calculate what current will be required to drive this voltage into a lower impedance load, and set the bias current accordingly.

Method 1 just tells you how much power you can get into a lower load impedance. If you attempt to drive a low value load with an amplifier set up this way, you will not be able to drive peak voltage into the load... your amp will be current limited into loads under 8 ohms.

Method 2 allows you to drive 8 ohms at the desired power level, and ensures that your amp will be voltage-limited into lower loads, down to 4 ohms or whatever you used to calculate the bias current. This method of course results in much higher idle currents, since it will be necessary to drive lower loads to the same voltage (and hence much higher power levels) than an 8 ohm load.

IMHO, method 2 is superior when you will be driving real loudspeakers, since most 8 ohm speakers have impedance dips to 4 or 6 ohms at some frequency range. Since speakers are generally designed for voltage drive, you'll want to be able to give them the full voltage your amp is capable of without being current limited.

So, to continue the design example using method 2:

100W into 8 ohms dictates 22V rails, giving roughly 40V peak into the load. If the loudspeakers have an impedance dip to, say, 4 ohms (you'll have to verify this with your own speakers), then 40V/4R = 10A. This will be the peak current to the load, and each half of the Aleph-X will have to idle at half this current, for a total idle current of 10A.
__________________
Chad.
"

Any thoughts ? How practical is that ? In a real world, the speaker may have his dip in the bass (I know that it is different with some electrostatic speaker though). Additionally we should assume that we need probly real power especially in the bass. So, If I am voltage limited in the Mids/ Highs (let's assume we have there 8 ohms), does it matter ? Will we see distortion or only limited power ( which we anyhow don't need that much in that frequency area ) ? In other words: will it be audible if we stay with or limited voltage calculated for 3 ohms or should we design the voltage for 8 ohm and the current for 3 ohms ?
 
The appended little spreadsheet may helps in the configuration of an Aleph / Aleph-X (see tabs). I hope the file is self-explanatory, simply start with the input on the left side (yellow fields) and thereafter with your input on the right side (blue fields).
May I ask Grey, Chad, Peter… to have a look on the results and let us know if it’s okay or even not. Needless to say, I’m not an EE…

Jens
 

Attachments

Blitz,

to calculate bias values into lower impedances just change the numbers. The calculation stays as it is.

Problem (as always) is the heat dissipation. You´ll have to decide beforehand wich voltage you can live with and then go for the largest possible bias. For my Thiels I will use 16V rails and 10A of total bias cause they are below 3 Ohms most of the time.

Jens,

I can´t really see what you´ve done cause the formulas are not open for examination ("blattschutz"). One mistake you´ve made is to calculate the ac current by dividing by 1,3 wich in this case is too much.
What do you mean with "bias rms" the bias will always be the peak value so 7,07A for 5A rms

Dissipation won´t be 32x5x2 but 32x7.07x2.

william
 
Blitz said:

There it is stated that the xa-200 powersupply with +-35 Volts at 30 Amps ! This is quite different from the 30 Volts at 5 Amps stated somewhere in the threat.
The writer of that was talking about the current rating of
the supply, which he computed from 2KVA / 70 volts. It
is not the draw of the circuit.
 
Again: Excellent

William,

This hits the nail on its head. By playing around with it I found my personal favorite BIAS ( I have a speaker with 5,5 Ohm):

voltage 19,5 volts
bias 6,3 amps
ac current gain 50% 50% means half the current by the source
number of fets 8
Peak current 6,3 amps
power 8 Ohms 76,6 watts
power 6 Ohms 102,1 watts
power 5 Ohms 99,2 watts
power 4 Ohms 79,4 watts
power 2 Ohms 39,7 watts

Dissipation 245,7 watts
Dissipation per fet 30,7125 watts

Number of fets is counted without the current fets, right ? So in reality I need double the amount of fets, in my case 16 per channel.


I guess Mr. Pass is somewhere in this ballpark:

voltage 31 volts
bias 8 amps
ac current gain 50% 50% means half the current by the number of fets 16
Peak current 8 amps
power 8 Ohms 210,3 watts
power 6 Ohms 192,0 watts
power 5 Ohms 160,0 watts
power 4 Ohms 128,0 watts
power 2 Ohms 64,0 watts

Dissipation 496 watts
Dissipation per fet 31 watts

Still, the only thing which makes me wonder is that the current and dissipation is still quite different from what is stated in the documentation. Clearly, I don't want ask for secrets of the XA-200 ( I would love to, but would not expect an answer as this is part of the game. On the other side, it is christmas...), but be sure that we have no flaws in the calcs. Any comments, Mr. Pass (thanks for taking the time to be with us)?

May be this is my last post for a few days. In that case: Happy christmas to everybody ! And thanks for the input ! This was more and faster than I have anticipated !
 
Yes it would be nice to nail down the number of devices needed.
People converting Aleph 2's would have 12 per channel, and it would be a pain to get new ones to match. So I now realize that I'm (and other Aleph 2 folks) am looking for the specs to make an x aleph with 12 Mosfets, whatever power that turns out to have, and biased the same relative amount as the Aleph 2 which Mr. Pass claims sounds most pleasing to him due to the biasing the 2 has.
 
Wait...

I can understand that we can all learn from thread like this one, especially for newbie like myself, but after a few posts that I read here all I can think of this is getting very dangerous here. Just a warning guys it is all well for having interaction among a few of us to kick the ball around a bit but we do humblely require the old hands to guide us thru, and any not so experienced DIYERs out there reading this thread should have the patient to wait and see/watch what the other more milage guys stated clearly what is safe and what is not before you act upon. That's why I (I mean I) always try to stay away from the serious chats among experience fellow members when they have something to discuss instead of posting questions that interupted them or basic/newbie questions should be asked when it is appropriate.

We have no dead line to meet, just try to do a good job and enjoy/reward ourselves with a fine product.

From a humble newbie Happy Season,
Chris

I hope I am not stepping onto anybody's toes.
 
Status
Not open for further replies.