kevin gilmore

There are holes in the board to stick a screwdriver in. i may have to make them bigger. I figure the size of a hex head #4-40 should do it. About .2 inch. look here... http://www.amb.org/audio/beta22/ off board heatsinks, bottom mount. edit: like this http://gilmore.chem.northwestern.edu/kgsshvampv7.jpg now with .2 inch holes for hexhead screws.

justin pointed out that the transistors are at the top of the heatsink, but you also could just turn the board upside down. I don't have any decent thermal analysis software, but if you figure on about 18 watts of heat, the temperature differential (as measured on my T2) from the bottom of the heatsink to the top of the heatsink is only about .3 degrees C. (and the T2 is more power) Nothing to worry about. But still, i'm going to try and calculate by hand what the differential would be on a 3U size heatsink. I've seen ayre's and firstwatts with the transistors at the top... krell with top and bottom, B24 with top and bottom... I'm making all of the board files available. If people want to make multiple board runs of the multiple different versions, that is certainly OK. I of course will have at least one of each. But i can do my own chassis so for me its easy. What we need is some company like front panel express that can take standard heatsinks and drill and tap holes...

Yes, but the top and bottom bracket on the heatsink might get in the way of the transistors. Need much better real picture.

I took care of the high voltage fets instead of the 2sa1968's this way http://gilmore.chem.northwestern.edu/kgsshvampv7.jpg The ixtp parts are not isolated, you would need the ceramic insulator. From justin's measurements, the inside finger to inside finger is 4.740 and the board is now 4.570. The transistors are inset by an additional .20 so everything would fit perfect. (the ceramic insulator might be a tight fit?? will check) Its very hard to tell from their lousy dxf drawings but i think it would fit this case too http://www.modushop.biz/ecommerce/cat079_l2.php?n=1 edit: found a way to get both current sources on the board. edit: board reduced in size, that is about it.

Get me some sabretooth tiger tusks. I would be more than happy to make some stax jacks out of those. Or ancient wooly mammoth ivory. Really any kind of ivory that is at least 200 years old. You know i'm not kidding. Too bad i gave away all of my macor.

Sure they have the die. They also have the 5 inch version. Justin owns both. I will make it so that the board fits vertically on the heatsinks. I think i have to shave off an additional .25 which should be easy. Would make a very nice box with the 2 amplifier channels, one on each heatsink, the power supply board in the middle on the bottom... But i'm not doing any more chassis, someone else would have to do them. And that would include the minimum batch of about 400 lbs worth. Board reduced to 7.2 x 4.8 inches. Should fit perfect in a single bellatone chassis.

http://gilmore.chem.northwestern.edu/kgsshvvert.jpg in progress

for those of you that want to do a 2 box thing with side heatsinks, i'm almost done with the amp board for that. Very symetrical.

Plexiglass blast shield and rubber gloves... Really good high voltage teflon caps to take the DC out, so you can measure the AC, and floating 3 wire 5.5 digit nanovoltmeters.

1 x lsk389 or 2 x 2sk179 1 x lsj109 or 2 x 2sj74 same power supply for both one power supply should be enough. but 2 power supplies moar better.

Next you gonna ask for a BOM with the mouser parts labeled I have absolutely no clue. The originals were done before i ever did a bom. Likely similar to a 4 channel B22. here is the dynahi balanced parts list for one channel not including power supply QTY PART-REFS VALUE --- --------- ----- Resistors --------- 2 R1,R4 500 2 R2,R3 30k 4 R5,R15,R53,R57 10k 6 R6-R8,R13,R14,R16 5k 6 R9-R12,R23,R34 100 8 R17-R20,R32,R33, 300 R45,R46 4 R21,R22,R35,R36 3300 16 R24-R31,R37-R44 10 2 R47,R48 1M 1 R49 5000 2 R52,R56 200k Capacitors ---------- 2 C1,C3 1pf 1 C2 .47uf Integrated Circuits ------------------- 1 U1 OPAMP Transistors ----------- 5 Q1,Q11,Q12,Q23,Q24 2sa1145 5 Q2,Q13,Q14,Q25,Q26 2sc2705 2 Q3,Q4 2sc3381 2 Q5,Q6 lsj109 one of these, or 2 singles 2 Q7,Q8 2sk389 one of these or 2 singles 2 Q9,Q10 2sa1349 8 Q15-Q18,Q27-Q30 2sc3421 8 Q19-Q22,Q31-Q34 2sa1358 Diodes ------ 2 D1,D2 LED

Any diodes you can get your hands on that are 1000 volts or better, and a couple of amps or better and fit in the board will work fine. But if you are nuts, get a few different kinds and pick the ones with the lowest noise. But be wary, because actually measuring the noise differences between the diodes at the regulated output requires all sorts of trickery to get valid answers.

Remember the pictures of some of mikhail's disasters with the juice can caps.. Those were 4700uf/450v.

Here is how to avoid that problem. Turn on the amp remotely with some fancy automation gidget first. Then turn on the room lights. You should see my lights dim when i turn on the krells.

Those parts are CHEAP. Besides which, it makes it easier for me, less parts to keep in stock The most expensive part is the transformer.

Well after working on this for a bit, and Ti can certainly prove that i'm wrong... The opamp servo is in fact a common mode servo. The differential stuff comes from the susy. So i think that the balanced dynahi and balanced dynafet are correct as is. But i ran into something last night i have to warn everyone about. I got to playing with one of these new all discrete I/V converters. The thing has 100mv of DC output. Why is that a bad thing, for a differential amplifier like these two, with a voltage gain of 10 (differential voltage gain of 20) that 100mv of dc turns into up to +1 volt of output at one of the terminals and -1 volt of output at the other terminal. So 2 volts DC across the headphones. While no big deal for electrostatics, and this is something the stax mafia have known for years, this is REAL BAD. Think AMB's protection circuit will protect you. Think again. His circuit adds +L and -L and +R and -R together. In this case it would be +1 volt plus -1 volt. Result is zero, so the protection circuit is happy. But there is still 2 volts of DC on the output ready to destroy your headphones. Gets even worse if you are making an amplifier with a voltage gain of 40. Think about it for a minute, what the gain has to be to be able to do 160 volts peak to peak on the output. power supply board in progress http://gilmore.chem.northwestern.edu/dynahips.jpg edit: i finally see what Ti was talking about. So i added a single 1M resistor.

Those are the same as the krell heatsinks from the ksa series, anodized black, in 9 inch lengths, and turned sideways.

Ti wanted me to add a common mode servo. Which i did to the board. Will test circuit later. I don't think its necesary, but what the hell do i know. Have the design for the power supply. The main difference between the dynahi and dynafet is whether you like the sound of bipolar as outputs vs fets as outputs. And a slight difference in the cost of all the additional parts. The safety circuit really is a great idea, but no easy way to get the output wires to a corner of the board. Will look at that later.

dynafet a bit more because of the extra parts, but the bias is adjustable. Say 20 watts per channel. Dynahi is about 18 watts per channel.

This is possible, but need to find the absolutely smallest latching DPDT relay that run on 5 or 12 volts. With the current relays, the board would be 7.4 x 2.6 inches and you would need 2 of them.

first rev of both done. someone needs to do some serious checking on these. Both boards identical size with identical mounting holes and compatible with the T2 power supply angle brackets. http://gilmore.chem.northwestern.edu/dynahibal1.zip http://gilmore.chem.northwestern.edu/dynahifetrbal.zip

dynahifet in progress http://gilmore.chem.northwestern.edu/dynahifetbal.jpg because of all the power supply stuff all over the place, this might end up as a 4 layer board.

I suppose i could yank that part out and do it as a seperate board.

They would work just fine. But i want something that is compatible with my stack of machined brackets. getting close now. http://gilmore.chem.northwestern.edu/dynahibal1.zip trying to figure out which one of the 14 or so dynafets is the one i want to start with OH.. and this

Well the patent for super symmetry has run out, so i don't have to even ask nelson for permission... (not that i think that was ever a valid patent to begin with) Current board size 9.12 x 5.22 target board size 9 x 5 People really want to use HE6's as room speakers and dump 60 watts RMS into them, this will do the job for sure. The only right way i know of to do unbalanced to balanced, and balanced to balanced at the same time. opamps, bah humbug. If you go with the opa445 as the servo, you don't need the +/-15 volt regulators. Anyone want to suggest additional power supply capacitor sizes. I have plenty of room left.