luvdunhill Posted July 12, 2009 Author Report Posted July 12, 2009 higher values for the output resistors. bah, wasting all my matching efforts and raising output impedance? madness!
kevin gilmore Posted July 12, 2009 Report Posted July 12, 2009 I'm liking it just fine with 2 ohm emitter resistors. Vbe multiplier just fine with 10k resistors and 20k pot. a resistor in series with the 20k pot is a really good idea. I'm not using the input bias trimmers. As long as the voltage at the output of the opamp is less than +/-2V no need to trim. working on a custom fully machined heatsink now.
j4cbo Posted July 12, 2009 Report Posted July 12, 2009 Kevin: as it is now I've got a solder jumper to ground the middle feedback point (just before the diamond buffers). Under what circumstances is that necessary? Do you think it should be kept as a solder jumper, exposed as a 1x2 pin header, or just removed?
luvdunhill Posted July 12, 2009 Author Report Posted July 12, 2009 Kevin: as it is now I've got a solder jumper to ground the middle feedback point (just before the diamond buffers). Under what circumstances is that necessary? Do you think it should be kept as a solder jumper, exposed as a 1x2 pin header, or just removed? Kevin discussed this in posts #1001 and #1076. I did as well in a few more posts.
j4cbo Posted July 12, 2009 Report Posted July 12, 2009 I think this could be the one... http://b.j4cbo.com/temp/gbf2f-top.png http://b.j4cbo.com/temp/gbf2f-bottom.png http://b.j4cbo.com/temp/gbf2f-signal.png
DigiPete Posted July 12, 2009 Report Posted July 12, 2009 cool.... when all have weighed in and the design is final, send me a BOM and I will work to order the parts. Great work by all!!
j4cbo Posted July 12, 2009 Report Posted July 12, 2009 Kevin discussed this in posts #1001 and #1076. I did as well in a few more posts. I was referring to this:
kevin gilmore Posted July 12, 2009 Report Posted July 12, 2009 (edited) SJ2 NOT necessary. EXCEPT you have to flip pins 1 and 2 of the feedback jumper connector. The inverting input needs to be the middle pin. Edited July 12, 2009 by kevin gilmore
digger945 Posted July 12, 2009 Report Posted July 12, 2009 I am finished tinkering now. The board is biasing properly with the parts Kevin mentioned above (10k resistor, 20k pot), and has amazingly stable offset without the servo. Yea it's gonna be a little hotter than the Dynahi. The one FET board transplanted into my amp with only two outputs is a little warmer than 2 Dynahi boards with all 16 outputs. In the mean time, I will continue to tinker with my own proto board and see if there is anything I can come up with to reduce the bias and make things not so hot. On the biasing resistor values Marc wanted, I have 480 on the positive rail and 520 on the negative. Marc had 470 and 510 so basically the same thing. Should you choose to leave the board and values as-is, then 620ohm paralleled with a 10k pot will be, to me, perfect. The board Marc sent has 681 and 10k and is very workable and has good range.
luvdunhill Posted July 12, 2009 Author Report Posted July 12, 2009 digger945: cool! Can I also get the Vf for the LEDs in the buffers, so I can finish calculating the dissipation there? j4cbo: I'll sit down and print a copy of the boards out and look over them tomorrow, but on first glance things look perfect!
digger945 Posted July 13, 2009 Report Posted July 13, 2009 1.645 on the positive side, 1.655 on the negative.
luvdunhill Posted July 13, 2009 Author Report Posted July 13, 2009 (edited) Jacob: The changes look fantastic! I just collected all my notes and here are the remaining suggestions that I have: 1.) On C1 can you enlarge the holes to 1mm diameter? 2.) In post #884 I suggested a change to the silk for the dual devices. It would be great if you could implement this, as this part of the assembly is very confusing. I'm sure someone will screw this up. I'd also add the 2SJ109, 2SK389, 2SA1349, 2SC3381 nomenclatures on the silk as well, per my the mockup part I included in post #884, as you have plenty of room. Trust me, when people are installing the devices, they will thank you profusly for this change. 3.) Along these lines, can we label the positive and negative power inputs? For some reason I blindly assumed the positive was on the left... Glad I double checked it in the end! 4.) Furthermore, can we add the cap ratings as well to the silk? It's hard to tell without really understanding the layout that the top two caps need to be >=63V caps and the bottom for need to be >=35V caps.. Again, I made this mistake, and I'm glad digger945 pointed it out early 5.) For the four bottom caps, I think if you separate them out a bit, you might be able to cram in slightly larger diameter caps. I guess what I'm asking, is it possible to make the bottom four caps 18mm diameter as well? 6.) I'm not sure what your DRC rules are set to, but the output pad seems very close to the FET pads. In looking at the copper, if you moved the output down a bit, you could perhaps get another fenestration to connect from the power planes to the output FETs, and that might be worth it. I'm not sure, but it might be worth looking at. 7.) Would it be possible to go back to the prototype buffer layout where none of the transistors are facing each other? I'm just concerned because these get so hot that the BJTs will start to runaway and the buffer become unstable. I liked the prototype layout where there was two rows and they were all lined up. I think you could even space the two rows apart a little more to help with cooling, and if this is possible, which it looks, I'd definitely try and do this. If you want to face some of the transistors together, I'd recommend the CCS transistors (i.e. the three pair that are not a part of the buffer) as this could potentially help the offset be even more stable. For example, I'm not sure if the BJT facing the LED is the best idea, but who knows. Maybe at minimum you could turn these two to face each other? 8.) Pars recommended a flush socket part in post #886 that may be worth looking into supporting. Looking into it, it's an easy change I believe... along these lines, I'm assuming it was too hard to enlarge the outer pads on the resistor footprint? You mentioned that you might be able to do this in post #893. I'm assuming this wasn't possible? Even the smallest amount of room would be greatly appreciated, as it's a very tight fit and many of us are hoping to use PRP resistors, which due to uneven tolerances will be even tougher to fit than the Dales. 9.) I noticed that all the holes on the top of the board for mounting the FETs and the board itself almost fit a #6 screw... maybe it's a M5 hole? Anyways, could we enlarge these a tiny bit so #6 would fit without "threading" the through-hole plating? 10.) You might want to make the screw on the center pot point to the right of the board, as that way the value will be visible when it is installed and hooked up to the heat sink, since this is a rather important value for determining the bias range. I had the hardest time trying to read this value once it was installed on my heat sink Obviously people can swap the part themselves, but the silk is a good reminder 11.) I'd line up R57-R58 with R7,R8,R10,R12 and line up R1-R4 with R59, if possible. I like resistors in nice neat rows I suppose Finally, what PCB fab are you considering? I'd sort like black solder mask, with gold plated pads, with 2 oz copper Depending on where you get the boards made up, these aren't really all that expensive. Check Phoenix Gold for example, who even offers colored solder mask and gold plated finish for free for larger orders, and at a nominal cost for smaller orders. I wasn't so impressed with the silk screen from PCBFabEx, but everything else was top notch... However, I'm not sure that their quantity pricing is very competitive, compared to some of the other places, hence the question... oh, and black boards FTW. Edited July 13, 2009 by luvdunhill
Icarium Posted July 13, 2009 Report Posted July 13, 2009 If there's a consensus I don't mind going with someone else. Just let me know, but really I don't have the experience or knowledge to like figure out the best peeps to go with myself.
luvdunhill Posted July 13, 2009 Author Report Posted July 13, 2009 If there's a consensus I don't mind going with someone else. Just let me know, but really I don't have the experience or knowledge to like figure out the best peeps to go with myself. Cool. I'll get with you and we can hash it out sometime this week
DigiPete Posted July 14, 2009 Report Posted July 14, 2009 I second the PRP suggestion, as that is what I plan for the group buy, at least for any signal path resistors (you guys will let me know which ones those are )
j4cbo Posted July 14, 2009 Report Posted July 14, 2009 http://b.j4cbo.com/temp/gbf2g-top.png http://b.j4cbo.com/temp/gbf2g-bottom.png http://b.j4cbo.com/temp/gbf2g-signal.png How's that? All the suggestions you made look good to me; if I missed something, it's probably because I forgot, so let me know.
kevin gilmore Posted July 14, 2009 Report Posted July 14, 2009 One other thing, i forgot to mention is the 5k resistors are now actually 2k on my prototype so as to match the gain between the two different feedback types.
luvdunhill Posted July 14, 2009 Author Report Posted July 14, 2009 Jacob: The silk for the 2SC3381/2SA1349 need to be swapped around. See my board pic here http://www.head-case.org/forums/247143-post899.html
DigiPete Posted July 15, 2009 Report Posted July 15, 2009 not sure if its gonna sound good... The Dynamite would be tough to beat
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now