kevin gilmore

should have no trouble with that one, 1.5kva. The semisouth parts perform much better at 100ma, but that is clearly silly. I ordered some bimosfets, and igbt's, and some very beefy npn transistors to see if i like any of them. Its very hard to beat out the 6ca7 in that application.

I can't explain it yet either. I am running it at 500 volts with a 1kv supply and 20ma current source. On a decent sized heatsink. The gate current is a factor of 10 more than i expect.

I wanted to be sure i was doing things right, so i brought the test circuit into work where i have more equipment and an easier way to test. The semisouth parts have a significant problem. The massive amount of gate current necessary. Actually measures about 150ma per part, So 300 ma per channel, so to do this an extra 50 volt power supply would be required, spec'd at 1 amp, and referenced to the -500 supply. Plus extra current on the -500 supply. Kind of what you would expect for a jfet. Going to mess with a few of the ixys parts soon.

I don't know how much power your transformer is good for, but the T2 is 200 watts for sure, and the BA has to be at least 100 watts. So you are definitely pushing it. T2 can handle -15% line just fine, but i doubt the BA can.

so i did not have the floating number, then went into messenger, then all of a sudden the floating number appeared... not sure what happened. likely the increased speed is due to the export of the database into a new database which removes all fragmentation. An export and a re-import every 6 months might be a good idea.

Andy finally got his working. A bad Rv1 in one of the batteries. I would imagine listening impressions soon. Andy's first project was rebuilding a ES1, and with a bit of help did so successfully. Andy's second project was the T2. Anyone with the time and perservance should have no trouble with this

There are clearly some things that tubes do way better than transistors. The output tube is one of them. 15 pf input capacitance. Although that really does not matter as much as you might think. An equivalent Rds of about 5k. Also really does not matter except it limits maximum voltage swing. Neither the ixys parts or the semisouth parts are perfect for the job, its the lesser of 2 evils. I plan on trying many semiconductors to see which performs the best. But the ixys parts with the 1500 ns rise times are clearly not what you would want. The slew rate and distortion characteristics are what i'm after. It might be a bit easier to replace just the front end and leave the output tubes. And then you have something similar to the BHSE which is a bunch simpler with a lot less parts. I'm looking at other alternatives and trying to find out where apex gets their complementary 1200 volt fets.

I think that justin really wants a T2. Trouble is, he does not have the 2 weeks of time to build it. Hey mikhail... you wanna help justin out??

Simulation is a great way to find out if you are in the ballpark or not. If the simulation says it won't work, then it is highly likely that in practice it reallly will NOT work, and sometimes in spectacular ways. But taking a successful simulation, producing a board, and getting the thing to perform as designed is much tougher. Especially with killovolts of AC flying around at frequencies that can sustain a continuous arc.

Yes, and by changing all the operating points, the sweetness of the sound may change. The 50 volts across the 2sk216 and 2sj79 seem to be the magic values where those parts sound the best. Lots and lots of things can be changed or simplified. And i encourage you to build one and test it.

This is not srpp. At least its not driven that way. Its the solid state version of grounded grid. If it was bipolar it would be called a common base amplifier. The capacitance while way higher than a tube would be is still not subject to the miller effect and hence its much easier to drive than you might think. The idea is to keep the sound the same which means keeping the same way the output stage works except with solid state. I likely need to add a gate resistor on the current source to keep it from oscillating. You could use the original current source with 2sc4686's with a bit more base drive current. The normally on part cannot be used as the gain stage because that messes with all the bias stuff elsewhere. The 2sj79 would not be able to drive it.

four per channel... So that's $240 for the output transistors. Going to need bigger heatsinks I wonder how many pieces of justin's new 5 inch heatsink i would have to buy... Nope, not doing this again. NOT doing this again. Someone find me a standard heatsink/chassis assembly that can dump about 150 watts of heat. I might be able to shrink the board to 12 x 12...

here is the first cut on the all sand version. i synthesized it, and everything looks good. The semisouth parts are NOT cheap... http://gilmore.chem.northwestern.edu/t2schemsand.pdf what i would still do is replace all remaining 2sc3675's with 2sc4686a and replace the fet and led's in the batteries with the precision regulator and change R34,R34 as necessary to bring the battery in the range of 640V. And likely make Q23 a darlington out of a pair of 2sc4686a. +250 volt supply modified to make +150v, and cascode with 2sc3840. Overall, less power due to no filaments, but more power in the outputs, making the thing still dish out just about as much heat.

Actually the DAG is from my picture tube days at zenith. Yep, thats the conductive stuff we painted the insides of the bells with. DAG is black. I was thinking about the RED stuff which is definitely a great insulator. Don't remember what we used to call that stuff, although i do remember the slang term.

Excessive amounts of tobacco seem to do way more harm to high voltage electrostatic amps than dust or condensation. Especially when combined with dust, the combo is definitely conductive. Many of the original T2's toasted the power transformers which were way underrated for power. Micromass (now owned by Waters) of the UK continues to sell multi-hundred killobuck instruments and do things with transistors at 5kv that i would not do at 1kv. No conformal coatings, no rtv, no nothing. The stuff manages to live thru its expected lifetime of about 8 years. And in an industrial environment thats full of all sorts of nasties. If it bothers you, buy a can of DAG and paint it on the transistor leads after you solder them in. The stuff is good for about 25kv.

The two of you, and you both know who i'm talking about... Knock it off with sticking chassis in ovens where you cook food. Seriously... There are VOC's... Even in the water based stuff.

titanium screws here UTI-SCREW Inventory-Titanium Screws, Zirconium Screws, C276 Screws, Nickel Alloy Screws tungsten carbide screws are available, and look very cool neither requires any coating of any kind, and don't rust. The servo won't care if its on +/-12 or +/-15.

and the ones you use have fuses that are rated at 250 volts. I'll bet they arc real nice, and might even explode at 500 volts when they pop. The 600 volt fuses are huge.

The kgsshv power supply is NOT current limited. To do that the power supply board would have to be about 1.5 inches bigger in the one direction... Adding 2 x ixys parts and 2 more heatsinks. Making the board this much bigger would guarantee that the thing would never fit in one box. Maybe for a 2 box second version. Still the amount of energy stored in the output caps, subjected to a dead short results in fireworks no matter what you do. But at least the semiconductors would not have popped. I may have come up with an even better solution for the mini heatsinks. have to try it first.

kerry came up with this one... a BHSE with the servo's from a T2. http://gilmore.chem.northwestern.edu/BlueHawaii-T2-Servo-v1-2.zip opens with ltspice.

yeah... please don't do it that way. The alumina insulators really are required. The plastic #6-32's definitely work plenty fine.

I want to make the kgsshv absolutely trivial to build, and want to make sure that people know how to mount the semiconductors. For all the TO220F transistors, a hex head #4-40, a #4 plastic washer, the transistor, a thermal pad, the heatsink, then preferably a belleville washer (which i can't find stock on) and a nut. So for the 2sa1968 version, that leaves only the ixys parts and the low voltage regulators. The small heatsinks are #6-32 tapped. So you have to go with a plastic screw. For the ixys parts, thats the plastic screw, the transistor, the aluminum oxide insulator, and the heatsink For the low voltage regulators, all you need is a screw and a regular thermal pad. The #6-32 lexan screws are pretty strong it seems. unless someone knows where to get the screw tabs that are #6-32 on one end and #4-40 on the other end.

i asked this to marc and craig in PM's, but i might as well ask here. Anyone know the voltage ratings of a TO220F package? Specifically if i put a metal screw in the hole, how many volts does it take for the package to break down between the screw and the collector tab inside. Has to be well into the 1.5 to 2kv range right? I've found some new polycarbonate (lexan actually) screws that seem to be extremely strong, but still metal screws are better.

I think the designs of the servo's are such that the amp reaches and maintains the correct dc voltage very fast, certainly less than 5 minutes. On original BH's, it usually takes about 20 minutes for stability. BHSE is probably a bit quicker at 10 minutes. The tightly regulated power supplies also make a significant difference.

Rule #1 Never mess with high voltage at the end of a day when you are tired. Rule #2 Always remove the lead from the AMP terminal on the DVM as soon as you are done measuring current. Yep, i fried a meter once. It was more than 20 years ago. Half a dozen DVM's around the house, including a couple of $19.95 cheapies makes things a lot easier.