Group Build: Dynafet

[Pars]

That was something I always wondered about as well--why use L bracket and bend the transistors? I was thinking about getting aluminum about 3/8" thick x 1" or maybe a bit taller, mounting that to the heatsink and have the output devices stand straight up with the board running under the 3/8" piece. Possibly drilling and tapping holes into the bottom edge for the board to mount to?

[digger945]

I also thought about that for some time. The only consideration in my mind was that it was yet another physical bond that the heat would have to cross and, like with the angle, it would be hot on one side and cooler on the other and maybe cause it to bow and lose complete contact with the large HS.

My thinking may be off here, I'm not an engineer, just thought that maybe bolting directly to the HS might avoid alot of fuss. I have since purchased a drill press slide to make the holes in a nice neat row.

On the pic above I hand lapped each angle on a piece of plate glass with 440 grit until I mated the two parts with a bit of oil and they took a fair amount of force to seperate.

It probably isn't common to see the words "hurry up" and "dynamite" in the same sentence, but I had to decide on a method of attachment and really wanted to get this finished so we can compare it to a B22 at the upcoming meet. I really think it will kick some beta arse.

[j4cbo]

The board is 4.95" wide by 4.2" (same size as the Dynahi, maybe a little longer), with all the output devices in the exact same place as the Dynahi as well.

I'm going through and updating the board with extra holes for the output resistors and a rearranged input filter now.

[kevin gilmore]

The L bracket heatsinks are very easy to install, thermally efficient, and make

any repairs necessary in the future very easy. I've seen this done literally 100's

of times. Way easier than trying to get a screwdriver in sideways.

[digger945]

I think I'll take your word for it.

It does make building and installing the two halves of the amp simpler once everything is bolted up, for me anyways using the parmetal case.

[Smeggy]

Do we know a minimum heat sink spec per board yet or is that still in flux?

[DigiPete]

Hopefully the mounting holes for the boards are in the same geometry as the Dynahi. I would like to pull out my Dynahi boardsD and plug the DynaFets right in

[Pars]

The board is 4.95" wide by 4.2" (same size as the Dynahi, maybe a little longer), with all the output devices in the exact same place as the Dynahi as well.

I'm going through and updating the board with extra holes for the output resistors and a rearranged input filter now.

Hopefully the mounting holes for the boards are in the same geometry as the Dynahi. I would like to pull out my Dynahi boardsD and plug the DynaFets right in

See above. Jacob sez they are... The boards may physically be a little deeper, but that shouldn't matter.

[DigiPete]

Going the low dollar approach, just replace the boards in my existing Dynamite.

[Smeggy]

Going the low dollar approach, just replace the boards in my existing Dynamite.

I approve of the way you think!

[luvdunhill]

Jacob:

pesky me again... so, I was thinking... would it be possible to add another set of pads, similar to the photos below to support the TO-3P Hitachi laterals? They share the same pinout and it seems to me that they could be accommodated pretty easily.. the deal is you would obviously not populate all 8 spots, since you'd want to run them harder, so this would deal with the fact that things would look kinda silly with 8 TO-3P devices next to each other. I see this was talked about a bit at headwize and I don't think the daughter-board idea is worth the effort, but adding 2 more pads for half the devices might be more reasonable. What do you think? This would be outside the scope of the group buy, but perhaps a nice addition for someone to play with down the line (i.e. speaker amp applications).

[Smeggy]

What's the expected power output of the Dyna?

[Icarium]

I think this depends on how you bias it.

[luvdunhill]

What's the expected power output of the Dyna?

I think we went over this in the thread somewhere, have a look for the exact output per device. It of course depends on the number of output devices and how it is biased, so you could scale it back (or up) quite a bit. Let me ask it this way, how much would power would you like?

Edited March 12, 2009 by luvdunhill

[Smeggy]

Hmm, prolly default or scale it back a bit as the worst it's going to encounter is a K1000 and some 'stat transformer boxes.

[luvdunhill]

Hmm, prolly default or scale it back a bit as the worst it's going to encounter is a K1000 and some 'stat transformer boxes.

Ok, just remember power doesn't (always) equal gain. You can have a 500W amp with zero voltage gain.

[Smeggy]

As long as it can drive those I'm happy.

[j4cbo]

Here we go, better late than never:

Prototype Candidate 1

Let me know what still looks wrong...

http://b.j4cbo.com/temp/gbf2.0rc1-all.png

http://b.j4cbo.com/temp/gbf2.0rc1-no-gnd.png

http://b.j4cbo.com/temp/gbf2.0rc1-sch.png

luvdunhill: I'll look at adding those TO-3P footprints as well...

[luvdunhill]

Jacob:

Looks fantastic to me! I do need to sit down and verify the schematic though...

I have a few questions. What footprint are the large output resistors? I think it would be nice to accommodate two Digikey part P2.0W-2BK-ND per location (used in a lot of Nelson Pass designs) and if someone wants to drop a lot of cash, the Mills MRA-5.

Also, could the TO-220 regulators be omitted for the TO-92 versions? Basically these just have to regulate the Iq of the opamp at a 15V drop?

I'd increase the pitch of the four PSU bypass films. I think something with LS=5mm and 5.5Wx7.2L (i.e. Wima FKP2 10nF) would work well.

Finally, what are the dimensions of the large PSU rail caps? 5mm pitch by 10mm diameter? I just want to have a particular cap in mind for that spot before we commit on the boards, as it seems that we have room to increase the size slightly.... We should look at the standard candidates from Panasonic and Nichicon.

Very nice by the way.

[digger945]

If the lead spacing for the 15V regs is the same as the Dynahi board a TO-92 part should fit nicely with no problems, but why if a TO-92 part is all that will be needed.

I agree it would be nice to have lead spacing for standard box films, where those may be appropriate. Are you thinking just one spacing or a dual spacing?

Great looking board.

[Pars]

Looks very good Jacob. The only request I would have would be:

• on the input pads: could you add a ground pad so that a Molex KK 0.100" connector would also fit? I personally like using these over the Phoenix connectors.
  

I'll have to look again, but I would think the TO92 regulators would fit the TO220 pads alright, and I don't like normal TO92 pads anyhow (if not inline, and too small). Otherwise, I agree with the other comments. I'm curious as to how Marc plans on mounting/heatsinking a TO3?

[luvdunhill]

mounting/heatsinking a TO3?

TO3P, not TO3 Similar to a TO-247 package. Yes, the metal can TO3 would be pretty hard to do. For how to mount and heat sink, it would be similar to TO-220, see the 2 pics posted above. I don't have any immediate plans for using these output devices. The Hitachi LFETs are probably some of the more counterfeited devices out there. The Magnetec/Semelab/"whoever the heck in the UK they are now" versions are pretty cool, but cost an absolute fortune, and for matched versions, you better forget it (they were often sold as matched sets, and if you don't find them this way on the used market, you're kinda hosed). I think these versions (and their other "improved" devices) are probably the holy grail of (L)FETs available for power amp use.

[grawk]

omg, a board designed by committee

[Pars]

TO3P, not TO3 Similar to a TO-247 package.<snip>

Yes. I figured it had to be something different than the TO3s I am used to. I couldn't fathom how you were going to do those

omg, a board designed by committee

Yes, design by committee is a big no-no in my eyes. Ironic that I am one of the committee voices

[DigiPete]

Jacob:

pesky me again... so, I was thinking... would it be possible to add another set of pads, similar to the photos below to support the TO-3P Hitachi laterals? They share the same pinout and it seems to me that they could be accommodated pretty easily.. the deal is you would obviously not populate all 8 spots, since you'd want to run them harder, so this would deal with the fact that things would look kinda silly with 8 TO-3P devices next to each other. I see this was talked about a bit at headwize and I don't think the daughter-board idea is worth the effort, but adding 2 more pads for half the devices might be more reasonable. What do you think? This would be outside the scope of the group buy, but perhaps a nice addition for someone to play with down the line (i.e. speaker amp applications).

A bit unsightly...