Aleph-X builder's thread.

Yes, Peak power will occur between 3-4 ohms. I am thinking towards my future speakers which will be in the +/-4 ohm range. With my current speakers (8ohm), this amount of bias shouldn't hurt either as I am assured to reach voltage limit before I reach current limit. 100Watts is more than enough. Finally, my basement is cold!!!!

P.S. Nelson said that increasing the bias increases the sonic performance, up to a point. I'm not sure where that point is but would like to find it.
 

Attachments

  • photo 12.jpg
    photo 12.jpg
    17.2 KB · Views: 3,652
The sinks are greatly enhanced by fans, but not required if they are large enough. Hense the need for a tow motor.

I designed these based on my old text books from college. With natural convection they should remove over 500 watts per side with 30 degree C rise. I derrated them 50% considering there has to be good thermal contact between the plates and the spacer. I will use thermal grease when I final assemble it.

I did a test with my wifes iron,(I thought it was a good idea until my wife saw it) and I was amazed at how well it kept 1200W. After two hours I could still hold my hand on the base plate. I understand that it is not a "scientific or quantitative test"

The sinks are made by 1/4" base plate, 1/2x1/2 spacer, 1/8" plate, spacer, 1/8 plate, etc. Size is 8" by 24".

If it doesn't work in real life I'll get a fan.

Scott
P.S. I now know that it is not proper to respond to the wifes question "What re you doing with my iron" is NOT "I'm just borrowing it, beside you never use it anyway"
 
promitheus said:
What would be the lowest voltage that could be used without having to sacrifice quality?
I have no idea but a few things come to mind. The voltage should be high enough to make the mosfets operate in their most linear region. Datasheets can be handy here.
Look at the graph in the spreadsheet. Changing voltage and current have a tremendous influence on the output power vs. impedance of the load. Choose appropriate values for your speakers.
Finally, as Nelson says: Burn heat as much as your heatsinks can take. ;)

/Hugo:)
 
I think you're in for it now!

At least with answers like that. Maybe something along the lines of "Just making sure this ol' thing still works right. They are on sale over at ?????'s you know." But that could bring even more problems!

Anyway, I'm guessing you are going to be asked again and again about the heatsinks, so I'll shoot first.

I understand the general construction of laminated plates with 1/2" spacers in between. My question is how many spacers? Are they located right at the mounting points of each Mosfet?

Overall, it does seem like a smart way of assembling a clean looking heatsink. The downfall of so many junctions to be smooth and thermally greased does seem real, but your de-rating seems conservative, which may just be fine. With a size like that, though, I'd guess you would want wider or more spacers. Just a guess.

Can you provide a sketch or something of your arrangement.

Regardless, good thinking. I hope upon completion your are sonically satisfied and have adequate cooling.

Sandy.
 
heat sinks and better compound

Those are cool looking heatsinks, but the idea of using that much thermal grease gives me the willies. Try Digi-Key #HS243-ND instead. That is Aavid Ultrastick Thermal Compound. It is non-greasy, waxy material with better heat transfer characteristics than silicone grease, and is removable. Best of all, it doesn't "travel" or spread oily crap everywhere. It undergoes a phase change at 130 or 140 or so (use your wife's oven, since you already burned up the iron). It expands at that point to fill all available space.

I used it to bond built up sections of my A-75 heatsinks; u-channel to base plate, base plate to finned sections. It works great. Downside: it costs $17.13 per stick. But stick is good for 1000 sq. inches used correctly. I'll try to attach a photo, but might exceed bandwidth.

Larry Wright
Seattle area;)
 

Attachments

  • aavidthermalstick.jpg
    aavidthermalstick.jpg
    45.3 KB · Views: 3,204
Heat Sinks

Thanks for the info on the Ultra Stick. Looking at the data this material is far superior than the thermal grease I was going to use. I had nightmares on how I was going to neatly assemble them with grease. I make a horrible mess just mounting FETS.

I will post the drawing of the sinks and amp when I figure out how to export the file into a form that this board will accept.
Scott
 
Aleph-X power supply

Hi Everyone,

I've been following the Aleph-X threads and making notes as I go, but I can't seem to find what I am looking for in my notes and the search turns up 9 thousand hits...

Can someone post a schematic of the Aleph-x power supply for me, please? I am starting to look at transformers and don't know how many secondaries I need. Do I need a transformer with two secondaries for each channel, like the Aleph 2 design?

Thanks for your help!
Eric
 
Calculating Power of Q6 and r48

I have been searching the associated threads for calculating power dissipation of Q6 to no avail. Considering that my philosophy for the amp is to use as quite of a CCS as I can, I have chosen the Q6b, ZVP3310A option. Seeng that the MAX dissipation of the device is .6W I need to calculate the appropriate value for r48.

Given that my rail voltage is +/- 23V
subtract (VR2,R24,R26 network) voltage (about 4V)
subtract Vgs of diff pair Q5,Q7 (about 3.5V)
Leaves 15.5 Volts Vds across Q6

Power Dissipation would be (0.02A*15.5V) = .31W

To reduce power to lets say .2W and keep it linear I need to lose 8 volts which would set r48 to around 400 ohms. This puts the Vds of Q6 at 7.5 Volts.

I assumed Zero Volts on speaker out to back calculate Vds on Q6.

Additional note, I have chosen J1b and a J509 current regulating diode to give 3mA across D1b. R17 would be 10k.

Does this seem to be reasonable thinking?

regards
Scott
 
Eric,
A single secondary will do in a pinch. You can always create a virtual ground using resistors, since very little current flows to ground in this circuit. A center tapped secondary will be better, and dual secondaries would arguably be even better still.
Start by determining how much wattage you want, then work backwards from there, as that will give you the rail voltage and an indication of where you want your bias set. If you're buying a new transformer, then you'll very likely end up with a center tap or split secondary. If you're buying surplus/used, don't be afraid to go with a single secondary if it'll save you money.
Scott,
I'm trying to visualize the front end current source in my mind...your thermal calculations look good to me. I can't remember where the J509 would go, as that's not something I put in (Vref for the CCS, perhaps?). Perhaps someone who has a schematic at their elbow can answer that one.
Low noise is a good thing, no doubt about it, but the fact that the front end current source noise is common mode once it hits the differential, it'll cancel out for the most part. My two cents worth is that a TO-220 device (i.e. the IRF9610) can easily do a half-watt with no heat sink. If you want to use another package, then flip a coin and decide how close to the Pd limit you want to run. I tend to run things a little cooler than some people--I like being able to hold on indefinitely.

Grey
 
Unless you need more watts than the original schematic with 15V rails
It's actually quite an easy build. Just follow the BOM and buy
parts that match the original schematic.

Where it's not so plum easy is getting the DC Offset within working
limits. But with very good matching (.01) she'll work, and the offset
will come into line... just have patience (that's what i keep telling myself!)

I mean, if i can do it anyone can!
 
Q6 and Power cont.

Grey,

Thanks for the reply. For clarification, the J509 is Q12b which is an option on the Group purchase boards and not in your original circuit. The option actually calls for a J505 (1mA) current regulating diode but I thought it would be better to increase the current with a J509 (3mA) to operate d1b(lm329) further away from it minimum. Could have selected a j511(5mA) as well.

I failed to remember that the CCS is common mode which reduces the importance somewhat. I also agree that operating Q6 in the to92 pakage at 40% of its max capacity isn't my cup of tea either. However out of all of the devices this is probably the easiest to replace.

Lets consider this, r48=0 Q6=irf9610 using a precision reference of the type lm329 as d1. With this example I calculate that the resistor network (r24,r26, and vr2) would have to be around 130 ohms to get sufficient Vgs for linear characteristics. As you stated earlier, Q6 would be well within its power limits and vref would be much quieter than the original zener.

Maybe I am being to short sighted, I will have to search for a precision 9.1V voltage reference and these questions will become mute.


In the end I may just end up with the original version, however, I'm having fun trying to learn this stuff and satisfy my desire for improvements with the minimum amount of smoke. Not that smoke scares me mind you.

Regards
Scott.
 
CCS Calculation Spreadsheet

Attached is a spreadsheet that I have been using for calculating the CCS components for the various configurations. I am not an electrical engineer so be careful with the results until somebody on the web that is more qualified can verify it. The results seem reasonable.
 

Attachments

  • ccs calculator.zip
    4.1 KB · Views: 576