kgst

[chiguy]

I used metal screws, nuts, and washers like you did but I put the insulating washer on the other side, in between the screw head and 10M90S instead of between the heatsink and nut.  This way, there is no metal in contact with the transistor tab.  Also, are you using the longer 532-7721-3PPS insulating washers?  http://www.mouser.com/ProductDetail/Aavid-Thermalloy/7721-3PPSG/?qs=%2fha2pyFaduh6C0mWin5INivGTZ1BbwH0%2f0f33I4BSQcPNTszO8PP9g%3d%3d

 

I also used thermal paste between the 10M90S and ceramic pad and between the heatsink and ceramic pad.  http://www.mcmaster.com/#3883k24/=uxy3cm

Edited December 9, 2014 by chiguy

[chiguy]

Here's a picture of what I described above.  http://www.aavid.com/product-group/accessories/mounting-kits

[Laowei]

Thanks chiguy. I had the parts bass ackwards. It "worked" but not reliably. Will fix the other CCS bushings before they go too.

The photos above were just thrown together to show the assembled parts. In the amp I used Arctic Silver thermal grease as you noted.

[chiguy]

I'm sure you know but be sure to use the non conductive Arctic Silver if you're using it.  

Edited December 9, 2014 by chiguy

[insanity]

When you use the insulating washers on the front side (which is how it should be done), you will probably have to enlarge the mounting hole of the transistor tab with a corresponding drill. It is easy, just make sure that you clean of the metal debris thoroughly.

 

As mentioned you have to use the right washers that are long enough to go all the way from the transistor tab into the ceramic insulator. This way, the whole metal screw will be insulated and does not add extra capacitance (which is good too).

 

With this mounting method I have successfully build the KGSSHV and KGST and never had any arcing problems. If you do it right, there is IMHO no need for PEEK screws. 

Edited December 9, 2014 by insanity

[justin]

also make sure that there are no burrs left on the hole in the heatsink. i have seen them come from the factory that way.

[spritzer]

The holes on the 10M90's are a bit too small so you can crack the insulator when mounting.     ...shit that came out dirtier than planned...    Anyway, if there is any damage to anything then throw it away. 

[Laowei]

Great info! Thanks all.

When I was first assembling, the bushings wouldn't fit either the heatsink or 10M90S. So I wrongly chose the heatsink to drill out.

[insanity]

I have a question concerning the matching of 6s4a tubes for the KGST. In the datasheet 250V is stated as the max average plate voltage, where as 550V is the maximum dc plate voltage. The KGST runs at +-350V (=700). So what is the plate voltage in the KGST? 350? 700?

I would like to measure my tubes as close to real operating conditions as possible. To what plate and grid voltage should i set my tester?

[kevin gilmore]

plate voltage with respect to the cathode is about 340 volts

[insanity]

Ok I just measured my tubes with the following parameters Vplate = 338V, Vgrid -18.4 (arbitrary, just set the first tube to around 10mA).

 

With around 20 tubes, I got values ranging from 5mA to 12mA. Since I do not have much experience with matching tubes, I wanted to ask what you would call well matched for the KGST. How close should a pair of tubes be? 5%? 10% 20%?

 

Would you only pair tubes of the same brand/manufacturer (see below)? If I put together pairs for the KGST, should I make pairs with 4 tubes as close to eachother as possible or is it fine to make 2 pairs such as 5mA 5.4mA and then maybe 10mA and 10.5mA?

 

Furthermore I noticed I have some black tubes, some darkgrey ones and also some light grey ones. I thought only GE and RCA (if I remember correctly from reading) manufactured 6s4a tubes. Does anyone know if this is true?

 

Thank you so much for your assistance.

[jose]

I replaced R 1,5K by 1,21K and now I have an offset of 0-1 volt.

 

What is the max offset value on an electrostatic amp? 

[johnwmclean]

What is the max offset value on an electrostatic amp?

I read somewhere the Stax amps aim for a max 10v offset.

[jose]

I read somewhere the Stax amps aim for a max 10v offset.

 

Thanks John. Now I'm more quiet.  

[spritzer]

Stax shoot for +/-20V.  If you want to roll Single Power style then shoot for -110V. 

[ervstil]

Hi!

 

I can see that most of you already come a long way with your builds. This turned out to be a slow starting project for me as I´m have really limited time and founds.

 

I have built many amps of all varieties but never played around with tubes. Driving electrostatic is the first application where tubes makes at least some sense to me. One of the main features is that they glows in the dark, so I figure why hide them?

 

I start to think that i will configure it as a classic tube amplifier with exposed tubes mounted on a top plate. I will also mount the ccs transistors on an external heatsink. An added bonus with this configuration is that I can build a cheap case mostly from exiting material.

 

- Given that most of the heat will radiate outside the case would it be safe to use build the 400v version?

 

- I'm working on a pair of DIY headphones and right now they are really inefficient compared to my stax 202:s. Will the 400v v version give me more headroom for tough to drive headphones? Is there any other advantages at all with the higher voltage version?

 

- I there any issues of mounting the CCS and tubes off-board (added capacitance by cabling, hum from longer loops etc) except the obvious safety and practicality issues?

 

- I imagine that if the tubes are exposed to the free air their running temperature will fluctuate more within the LTP pair. It can be an issue with fets, it is the same with tubes or will the differences be swamped by the higher temperature?

 

Best regards Christian

[spritzer]

With plenty of cooling then the 400V version is feasible.  Don't worry about the amp drifting because it won't. 

 

If you use external sockets then simply wire the heaters to them instead of to the board.  Should clear up any issues with doing this. 

[nikongod]

-110V-(-110V)=0V

Edited December 17, 2014 by nikongod

[ervstil]

With plenty of cooling then the 400V version is feasible.  Don't worry about the amp drifting because it won't. 

 

If you use external sockets then simply wire the heaters to them instead of to the board.  Should clear up any issues with doing this. 

Thanks spritzer for the fast reply.

 

Will the 400v version give me more usable voltage swing? Will it make any real world difference for inefficient headphones? 

 

BR Christian

[spritzer]

It's a bit more voltage swing on top of the overkill we have already so not a whole lot of difference. 

[kevin gilmore]

-110V-(-110V)=0V

 

actually 400V - (-110V) == 510v of bias (which is close)

but the negative voltage swing is -110v to -400v (less the voltage of the tube) which

means real nasty and early clipping

[Kerry]

I finally got around to the power supply for my mini-kgst.  I was able to get the +/-400V with delay, filaments, +/-15V and a +5V supply all on one board (5.9" x 6.25").

 

Here's what I came up with...

 

 

 

Here's the 3D models from DipTrace.  I did all of the non-standard 3D component models myself.  I'm getting good at it.

 

 

 

I ordered the wrong size caps for the HV supply .  I'm going to put in another order and should have it up and running shortly.

 

 

 

 

 

 

 

I redesigned the timer circuit because I'm using it with a solid state relay.  There was a short ON burst in the older one that would turn on the relay for a small fraction of a second.  This one keeps the output low until activation.  I've been testing it for about a month and it seems to work well and is stable.  It was a combination of reading the datasheet for the LM555, modeling some ideas in LTSpice and then testing on the bench.

 

 

 

I'm usually careful, but made a mistake on placement of the charging capacitor.  I had to move it, but got luck and it was a fairly simple surgery.

 

 

 

 

 

 

I'm looking forward to getting the caps for the HV and testing it out the rest of the board.  All in all I'm happy.  Most parts are fairly accessible and it seems like I'll be able to get to them if need be.

 

This board will mount under the top panel of the amp in the back.  The transformer will sit on the top above the top panel directly above the supply.  I'm using the Antek 100VA 350V transformer and steel case.

 

More to come

[Laowei]

Impressive design and build Kerry.

[spritzer]

I like it even more in the flesh. 

[nopants]

I finally got around to the power supply for my mini-kgst.  I was able to get the +/-400V with delay, filaments, +/-15V and a +5V supply all on one board (5.9" x 6.25").

 

Here's what I came up with...

 

 

 

 

Here's the 3D models from DipTrace.  I did all of the non-standard 3D component models myself.  I'm getting good at it.

 

 

 

 

I ordered the wrong size caps for the HV supply .  I'm going to put in another order and should have it up and running shortly.

 

 

 

 

 

 

 

 

 

 

I redesigned the timer circuit because I'm using it with a solid state relay.  There was a short ON burst in the older one that would turn on the relay for a small fraction of a second.  This one keeps the output low until activation.  I've been testing it for about a month and it seems to work well and is stable.  It was a combination of reading the datasheet for the LM555, modeling some ideas in LTSpice and then testing on the bench.

 

 

 

 

I'm usually careful, but made a mistake on placement of the charging capacitor.  I had to move it, but got luck and it was a fairly simple surgery.

 

 

 

 

 

 

 

 

I'm looking forward to getting the caps for the HV and testing it out the rest of the board.  All in all I'm happy.  Most parts are fairly accessible and it seems like I'll be able to get to them if need be.

 

This board will mount under the top panel of the amp in the back.  The transformer will sit on the top above the top panel directly above the supply.  I'm using the Antek 100VA 350V transformer and steel case.

 

More to come

 

Those HV supplies are tiny, especially since you're using the fancier version. Nice work

Edited December 21, 2014 by nopants