Heatsinking the input diff pair?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello!

Currently I'm designing an Aleph-X. Now its time for the PCB-layout and a questions has been coming up: What about the heatsinking of the input diff pair labeled with Q5 and Q7 in Gray's schematic? Is a heatsink necessary? Somewhere I read a statement of Nelson Pass that heatsinking of the input diff pair isn't necessary but looking at pictures of AlephXs there are the diff pairs mounted on heatsinks. So, which thermal resistance for the heatsink should I choose?
How should the Mosfets be mounted? On the Wiki-page you can read, that they should be mounted back-to-back but then mounting on a heatsink would be a problem.
Would it be a solution to mount the mosfets side by side on the same side of the heatsink or isn't it really optimal?
Another solution could be to mount them back-to-back with a heatsink between the mosfets but then the air flow won't be optimal because I haven't seen a suitable heatsink yet because the PCB will be mounted vertical in the housing. All heatsinks where I can mount the mosfets back-to-back are desinged for horizontal PCBs like in JLMs Aleph-X shown in the gallery on the passdiy-website. I hope you can imagine what I mean.
What are your suggestions?

AudioAngel

P.S. And what about Q6?
 
The heatsinking is necessary to ensure both devices are at same temp.( otherwise thermic distortion ) , and it will ensure your DC offset to be more stable .

In the XA 200 pictures from Nelson the diff pair is mounted on same heatsink :devilr: so it is good to do the same :smash: and even if "it is not necessary" . Not necessary but desirable ...

best regards

Anael
 
AudioAngel said:
Hello!

Currently I'm designing an Aleph-X. Now its time for the PCB-layout and a questions has been coming up: What about the heatsinking of the input diff pair labeled with Q5 and Q7 in Gray's schematic? Is a heatsink necessary? Somewhere I read a statement of Nelson Pass that heatsinking of the input diff pair isn't necessary but looking at pictures of AlephXs there are the diff pairs mounted on heatsinks. So, which thermal resistance for the heatsink should I choose?
How should the Mosfets be mounted? On the Wiki-page you can read, that they should be mounted back-to-back but then mounting on a heatsink would be a problem.
Would it be a solution to mount the mosfets side by side on the same side of the heatsink or isn't it really optimal?
Another solution could be to mount them back-to-back with a heatsink between the mosfets but then the air flow won't be optimal because I haven't seen a suitable heatsink yet because the PCB will be mounted vertical in the housing. All heatsinks where I can mount the mosfets back-to-back are desinged for horizontal PCBs like in JLMs Aleph-X shown in the gallery on the passdiy-website. I hope you can imagine what I mean.
What are your suggestions?

AudioAngel

P.S. And what about Q6?


mabey you could redesign the diff circuit so that at idle, the power dissapation from both transistors are equal, cancels out temp. coefficient
 
You can use devices other than the IRF9610. Note that--unless you're willing to turn the whole project upside-down--you'll need a P-device. Or you could capacitor couple the front end to the outputs. But then you'll need to provide for bias, which will set you back a few minutes while you figure out a bias scheme.
Changing devices will change the sound of the amplifier. Sometimes radically. As an example, if you go with a bipolar front end, you're likely to end up with more gain than the stock MOSFET front end. More gain will in turn lead to more feedback, which in turn will alter the distortion spectrum, damping, etc.
You might even feel that it's an improvement, sonically.
Just be careful not to unbalance the relationship between the front end DC offset across the load resistors. This sets the bias for the back end.

Grey

EDIT: The front end doesn't get really hot. If you use too much heat sink all you do is create 'sails' that catch every vagrant breeze, which makes the devices drift. Paired devices are useful for this problem. There are other strategies for keeping the front end stable, thermally.
 
To avoid the vagrant breeze problem I mounted the diff pair on a small piece of aluminum, with electrical isolation of course, and enclosed the whole shooting match in a small steel cover that I scavenged from the TV tuner section of a PC video card. You can also find these cages in WiFi access points and so on.
 
Administrator
Joined 2004
Paid Member
I do the same thing as markp. Many commercial manufacturers do likewise. It's a simple, inexpensive way to get the job done. In Adcom amplifiers, they bond a small piece of copper across the top of the diff. pair. These run a little warmer,so that's a good solution too.
-Chris
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.