The Aleph-X

jacco vermeulen said:
You likey crazy bird food then ?


Naah.

me clever coyote .

(ya can buy drekload of J309/310 for 10 greens)
 

Attachments

  • c.jpg
    c.jpg
    38.6 KB · Views: 715
I’ve added some of the embellishments requested and attach a new schematic. Differences between this and version 2.1 are as follows. Standard stuff - added zener protection (D1-4) for output MOSFETs. Improved the bias for the cascodes as suggested using current regulator diodes J507 (my software doesn’t have the right symbol for these so I have substituted a current source in the schematic) and quiet voltage references LM4040 instead of the simple voltage divider resistors. Personally I suspect this is overkill but I doubt that it will make the circuit sound any worse and it may even be better. I’ve placed a resistor across the top trim pot although I’m not convinced it is required. I’ve also added C9 and C10 to tailor the high frequency response although they probably will not be required in practice. The arrangement for +ve current to the drivers for the SE bias has been improved by the addition of R5, R9 and C3, C4. Finally I’ve added some on-board PS decoupling in the form of C5-8. None of this will make a huge difference.

Sadly the software that I am using doesn’t know about some of the components I would like to use and hence I will make some suggestions here. Q12, 17 would be better as ZVP2110G, likewise Q9, Q18 should be ZVN2110G. A possible drawback is that these are surface mount parts but they do have much better power dissipation. The ZTX450 parts would be better as ZTX451 but as both are getting harder to obtain, perhaps we should consider BC639 or similar?

I remember somebody being concerned about damage to the input JFETs. I’ve noticed that most other circuits using these parts do not appear to include any form of protection (unlike when using static sensitive devices such as MOSFETs) so I’ve not included anything here. Anyone have any suggestions for something suitable if it is really required?

Now we must get down to the nitty gritty decisions that need to be made before we can actually construct something and begin experimenting in earnest. The main considerations here are what PS voltages to use, how much bias for the output devices and how much should be SE. The latter will also determine the ratio of SE devices to the other output devices. Unfortunately this is also the hardest decision as I don’t have any data on how to determine what level of SE bias might be appropriate, whilst the rest is easier being related mainly to the required output power and associated load.

Other PS details such as transformer size and regulation for the front end also need to be decided. I was hoping to get around to this but have run out of time this weekend. Other considerations include the number of PCBs (assuming we are going this route) as this project will likely require the output devices on their own dedicated boards for heat sinking reasons. I’m thinking of building mono-blocks so it may be convenient to split the output board in two if heatsinking is to be on two sides of the case. So many decisions…

Ian.
 

Attachments

lumanauw said:
Sorry, I'm not following this thread, but why the AlephX (or is it XA?) becomes full complementary differential? It's quite different than post #1

That is a good question that can almost certainlly be answered better than I can manage. What I will say is that the commercial XA.5 has gone this way and hence the DIY effort here is pretty much just following along. If I had to guess I would say the benefit has something to do with better performance with single ended inputs and perhaps lower common mode noise due to cancellation.

Perhaps someone more expert than I can enlighten us further?

Ian.
 
lumanauw said:
Basic things has changed? 3stages instead of 2 stages? Single ended becoming pushpull?

Yes, indeed. If you read back up this thread from the end you will see that this is the direction taken by the Master and that the commercial XA and X amplifiers are converging in terms of basic topology. People that have heard the new XA.5 amplifiers are saying that they sound better than the old so, as usual, the Master must be on to something.

Ian.
 
lumanauw said:
Hi, Ian,

Looking at your schematic, I couldn't resist this dumb question. Will the X150 (classAB, already with UGS front end), jacking the bias to full classA (1/2 current needed for full rail swing), making it XA.5?

Not really for me to say as I don't know the intimate details of either design. That said, I think it would be close. The main difference I suspect will be the amount of SE bias which is probably higher in the XA.5. The other obvious issue is a purely practical matter, you would need many extra output devices to handle the current demands of a fully biased class A output. Beyond that I don't know although I have to admit I have spent plenty of time thinking about it.

Ian.
 
What I will say is that the commercial XA.5 has gone this way and hence the DIY effort here is pretty much just following along

Ian--are you sure if this is correct? I thought I read somewhere that the ".5's" were an incremental change over the original X topology, which I assumed had a single ended front end as shown in Petter's X100 thread and on Nelson's patent.

Exactly when did the change to dual-diff. front ends occur??

JJ
 
I dont think that the Aleph-X.5, which is being developed here needs to be an excact copy of the factory made .5. I do like this layout here, a lot:
http://www.diyaudio.com/forums/showthread.php?postid=1240345#post1240345
To me it seems to be the "next level", compared to Zen's and the stock Aleph's/Aleph-X's. The frontend looks great and very promising indeed. On another note, I remember that its been said that the "old" Pass Lab's XA's up to at least 100-160 Watt's always had a J-fet frontend. That didnt stop people saying that the Aleph-X was a XA clone (well more or less at least)😉 What's more; the Aleph-X with the "crude" 610/9610/ input managed to sound great anyway😉 , but thats another story. The dual-diff pair's were seen on a picture in another thread. (GL's Aleph-X thread). I think that Ian's schematic is pretty close to what is beleived to be a XA.5 clone. Certainly close enough, and I think Ian has made a tremendous job 🙂
Something else, regarding the pcb layout (if it comes to that!) I think that its a good idea to split the output-boards in two, making it suitable for monoblock's as suggested earlier. They could still be mounted in a stereoamp if thats what you want.

Steen🙂
 
jupiterjune said:


Ian--are you sure if this is correct? I thought I read somewhere that the ".5's" were an incremental change over the original X topology, which I assumed had a single ended front end as shown in Petter's X100 thread and on Nelson's patent.

Exactly when did the change to dual-diff. front ends occur??

JJ

Only the One and Only would know when the change over to dual-diff occurred but I think it is fairly common knowledge that the change was made some while ago (look at some related threads here). In any case the original patent showed dual-diff versions as well. So the X series had been this way for some time (using the UGS front-end) and the X.5 is indeed an incremental change to this varient.

Ian.
 
Steenoe,

Thanks for kind words and I agree with you and others that we don't need to aim for a precise XA.5 copy. I would like to keep to the spirit of the XA.5 design though and I believe the focus to be more about the output stage than the input. And this is where is gets difficult in terms of the detail. We know from previous hints supplied by the Master that this is based around the classic Aleph0 and hence this is where I started. What we don't know is how much SE bias is used (although I suspect it will be considerable to keep to the spirit of the original Aleph which is really a SE amplifier) and hence how much of the ouput stage (read output devices) must be devoted to this and how much to the PP. I'm currently at a bit of a loss to explain why the XA100.5 apparently uses 40 devices in one monoblock as this seems to be more than required even allowing for the SE CCS. Once we can make some basic decisions here it should be possible to get on and start building...

Since I will probably go the PCB route I will indeed split the output board at least in two and maybe more.

Ian.
 
Usually the factory made Pass amps uses more outputfets than the average DIY'er would use. A stock Aleph-X 100 watter can be made with only 12 powerfets in all. Some even use only 8, but thats pushing your luck I guess. It should be possible to cut down on the 40 powerfets (which is a lot) and still retain the same bias-current. Regarding the SE bias, the japanese spec.sheet says the current draw is 300 watts for a monoblock. If its max current or idle current I am not sure. Maybe someone that understand japanese could tell? Might lead to some conclusion.
http://www.electori.co.jp/PASS/XA100.5.pdf
The spec. sheet also says 53 deg. C above ambient! I hope thats a mistake.

Steen🙂
 
steenoe said:
Regarding the SE bias, the japanese spec.sheet says the current draw is 300 watts for a monoblock. If its max current or idle current I am not sure. Maybe someone that understand japanese could tell? Might lead to some conclusion.
http://www.electori.co.jp/PASS/XA100.5.pdf

Steen🙂

It is continuous power (see Pass Labs webs site) which tells us what the overall bias is but not how much of it is SE. I am beginning to suspect that most of the bias may be SE which would make it more like the older XA.

Ian.
 
steenoe said:
Usually the factory made Pass amps uses more outputfets than the average DIY'er would use. A stock Aleph-X 100 watter can be made with only 12 powerfets in all. Some even use only 8, but thats pushing your luck I guess. It should be possible to cut down on the 40 powerfets (which is a lot) and still retain the same bias-current. Regarding the SE bias, the japanese spec.sheet says the current draw is 300 watts for a monoblock. If its max current or idle current I am not sure. Maybe someone that understand japanese could tell? Might lead to some conclusion.
http://www.electori.co.jp/PASS/XA100.5.pdf
The spec. sheet also says 53 deg. C above ambient! I hope thats a mistake.

Steen🙂

Look closely at the specs that matter:

XA100
Output Voltage Maximum plus, minus approx. 40 volts
Output Current Maximum plus, minus approx. 5 amps
Damping factor 30 ref 8 ohms nominal
Distortion 0.05% @ 10 W, 1% @ 100 W @ 1 KHz approx.

XA100.5
Output Voltage +/- 43 volts
Output Current +/- 20 amps
0.01% @ 10 W, .11% @ 100 W @ 1 KHz approx.
Damping factor 150 ref 8 ohms nominal

With the 4 x drive current (into low impediance) one would imagine it to be a v. high transconductance design.

That invarably means lots of parrellel devices.

The Master has already mentioned that more devices results in lower distortion..all things considered and this appears to be the case here.

I also suspect the Pass inventory is well stocked.


🙂 Ian
 
I doubt that I will ever be good at japanese😉
Counting the fets (source resistors actually) there is 32 in each side of the pic. If the picture in the japanese pdf is in fact a XA100.5, that amounts to a total of 64 mosfets. I could have sworn that Mr. Pass said 40 somewhere?

Steen🙂