Group Build: Dynafet

[luvdunhill]

Jacob or Marc,

What's new on the DynaFet stuff?

Pete

I have received all the parts for the two prototype boards, just waiting on the board themselves from Jacob. I have received all of the small device heat sinks from Hong Kong as of yesterday.

In other news, I may pass on the group buy for the passive parts. I have a very large order to place to Parts Connexion and will go ahead and order a bunch of the parts then and just match them myself.

I'm hoping to have time this weekend to work on my case. I had a chance to cut the aluminum angle brackets and mark the holes on the heat sinks already. With this much heat, we need to pay close attention to the thermal junctions, hence I'm looking into various types of isolators and grease. I think I have a very workable solution and will be testing it soon in another design. I'm hoping to get a 5-6 degree temperature drop, which is pretty significant.

Stay tuned, we're getting oh so close.

[digger945]

It will be very interesting to see what you come up with concerning the different thermal solutions, TIM's vs. grease, or combinations.

[Pars]

Another product line that might be of interest:

THERM-A-GAP 600 Series Gap Fillers

I have some tubes of this at work (T644; not sure where they came from and what we used this for, and they are expired).

[Icarium]

Jacob or Marc,

What's new on the DynaFet stuff?

Pete

Sadly J4cbo is locked into finals right now.. they should be over this week (Started last week), but I think immediately he is moving out here to norcal to start his internship. He had hoped to complete the board-stuff prior to shit going down, but alas :/

Once that evens out though we should be back on track.

[luvdunhill]

Hey guys, are you planning on tapping the side of the angle bracket that secures the output devices, or instead using a nut on the bottom of the PCB? I'm wondering if the latter will be able to get as much force on the devices. My bracket is around 0.2" thick at this point. I'm also looking at #6 screws to attach the bracket to the heat sink, as I'm not sure how many threads I'd get out of a #8 screw. Again, the base of my heat sink is 8mm and bracket is ~ 0.2" thick, so should I consider going up to #8?

[digger945]

I plan to use just 2 #6 to attach my .125"x1" angle to the HS(Conrad, ≈8mm thick base).

Last go 'round I used 3 #8's but I think that's not really necessary with smooth mating surfaces and a dab of thermal gravy.

Since I can't seem to drill holes in a perfect line no matter what, I will probably go large on the holes to secure the devices to the angle so as to leave some wiggle for good alignment.

I understand what KG meant on this page when I took mine apart to make some changes recently. Far easier than removing the entire HS assembly from the amp.

Basically it will be a near repeat of the Dynamite only this time no broken bits or taps>. I must say that I originally thought hard about tapping the angle for device mounting and then, if it didn't feel stout enough, I could always use a longer screw and put a nut on the bottom of the pcb.

A few pics.

[luvdunhill]

Cool, thanks for the info. I think I'm going to use a 9/64" bit on the flange where the devices attach and a 5/32" bit on the flange that attaches to the heat sinks. I think this should give enough wiggle room (one screw size difference), while not over sizing the hole around where the devices attach to ensure as much contact between the devices and the angle bracket.

[luvdunhill]

The guy at the Toshiba plant must have been slacking on the 2SC2240s

sorry the graphics are small... just squint, it's all decently legible.

[luvdunhill]

yawn...

[Smeggy]

That is one serious collection of goodies..

[Fitz]

Oh my.

[Pars]

Someone has been busy...

[luvdunhill]

what, matching hfe for every transistor to the same value on all four boards isn't unreasonable, is it?

By the way, this amp needs some more compensation at gain of 4... at gain of 7 I didn't need any compensation.

What gain do other typical four channel amps use, like the dynamight and beta22?

[n_maher]

What gain do other typical four channel amps use, like the dynamight and beta22?

For balanced operation I can't imagine a single-ended gain of more than 4 (making the balanced effectively unless you're operating with a extremely low output source (less than 1V) or you have the world's most inefficient headphones.

[luvdunhill]

For balanced operation I can't imagine a single-ended gain of more than 4 (making the balanced effectively unless you're operating with a extremely low output source (less than 1V) or you have the world's most inefficient headphones.

I remember wanting more gain with steve's beta22, though I'm not sure what gain it was set at. Perhaps it was 5 per board? This was using a 1.99 Vrms source and my Fostex T50. I'm likely going to be using a custom attenuator one way or another, so my plan is to just add more attenuation here. I know he was using the JT, so I figured more gain (7- with another -10dB to -15dB attenuation should do the trick nicely.

[n_maher]

I can't recall exactly, but will try and look some time soon to be sure but I think Steve's beta is configured as 2/4, which still seems like it should be enough with a 2V source and any headphone but I guess it could be close on some single-ended cans. Personally I feel no shame using the upper half/third of the pot but that does tend to bother some poeple.

[luvdunhill]

Personally I feel no shame using the upper half/third of the pot but that does tend to bother some poeple.

It's not a matter of shame, it's a matter of having a blank four deck 42-step attenuator arriving soon Seems like an excellent use of the added steps to get more attenuation.

Yeah, I remember listening to his beta22 at full bore with his Pearl into the Fostex*. I guess I could do the math and get the exact output, but I know my CD player is 1.99vrms balanced and I wanted a bit more gain for sure.

(*) Yeah, that was the "fated" meet where the hubris of the TRS jack hit in a big way.

[n_maher]

Yeah, I remember listening to his beta22 at full bore with his Pearl into the Fostex*. I guess I could do the math and get the exact output, but I know my CD player is 1.99vrms balanced and I wanted a bit more gain for sure.

IIRC the pearl setup that was being used then was seriously low output, like ~.5V or something along those lines.

[spritzer]

...Yeah, I remember listening to his beta22 at full bore with his Pearl into the Fostex...

The Fostex sure needs serious power so don't try something like a HD600 in Marc's amp.

[Smeggy]

The beta 3-board here has a gain of 8 and I have to max the volume to get reasonable listening levels from my Isodynamics and orthodynamics so a little more wouldn't go amiss for me with the Dyna.

[n_maher]

You and Ray Samuels need to meet Smeggy.

[Smeggy]

You haven't met my Wharfedales... 10WPC is their rated input

[luvdunhill]

You haven't met my Wharfedales... 10WPC is their rated input

Smeggy, what does your source output?

[Smeggy]

I have no idea, Pico DAC output.

[digger945]

what, matching hfe for every transistor to the same value on all four boards isn't unreasonable, is it?

By the way, this amp needs some more compensation at gain of 4... at gain of 7 I didn't need any compensation.

What gain do other typical four channel amps use, like the dynamight and beta22?

Kevin claimed success with 4 on his proto FET.

Ti used 7 on his Dynahi, but recommended 10 or 11 and increasing the feedback cap from 33p to 47p.

I read on HW that Digipete had no satisfaction with his Dmite until set at 11.

I have used 7,9,10 and now 11(yes, bal. gain of 22) and like it at 11. It picks up on nuances and details better for me. If I had a good source I'de probably need a diaper on to listen.

Have you been playing around with a proto pcb?