i'm on a roll... the kgsshv

[mwl168]

I did an experiment earlier today - changed the bias resistors on my off-board KGSSHV to run at 15mA per output devices. It has been running continuously for 5+ hours and it appears to be working fine. I am running 415VDC rails PS. 

As far as I can tell by using an infrared thermo gun, my puny heatsinks are measuring around 50 C or so. It's a bit difficult to get good measurement of the case temperature of the 2SC4686A because how they are mounted to the heatsinks but my best guess is around 60 C, again, based on the infrared thermo gun reading. That's pretty close to the 7W dissipation limit at 60 C as stated on the 2SC4686A datasheet. For long term peace of mind, I will be looking for a chassis with ample heatsink for my HV.     

[Skooby]

I just built a KGSSHV with Birgir's rev 1.0 boards.  Everything runs fine, just that I have to use a 4A slow-blow fuse on the main AC.  Is this a tag bit too high, any issue to be concerned with?  I've tried 3.15A and 3.5A slo-blo but none worked.  I used two 4700uF caps on the +/-15V rails instead of 3300uF caps (but can't imagine these created the problem).  Thanks for any pointer.

[spritzer]

The fuse size is always about the transformer and its inrush current.  4A could be normal. 

[luvdunhill]

Fuse rating is simple. Total VA rating / line voltage * 133%. If that doesn't work, then make your PS not suck another way.

[Skooby]

Thanks!  A/C relay or soft start as Birgir suggested would work too.  But I'll stick with the 4A fuse, way cheaper solution than other options.       

[mwl168]

FWIW, I use a 200VA transformer and run my HV at 15mA output current. I use NTC thermistor on the prmiary of the transformer and use 3.15A fuse. It has been working fine for a few months. 

I have not gotten around to try smaller fuse yet so it provides proper protection in the event of faulty operation.

[DHTGUY]

I have finally plucked up courage to turn on one channel --- no smoke, but the LED in  the middle near the SA1486 is very bright, others much less, so, and after power on for ca. 30 seconds I  could smell hot components (but perhaps that is normal). Any suggestions? Just go for it? I have the global feedback jumper, no server connected, and I have adjusted each trimpot to roughly the middle.

 

I can hear a relatively quiet b-b-b-b sort of noise for 10 seconds or so, which then goes away - but apparently no signal (I have about 100mV 1Khz test tone on the input) and measured in through to the input on the amp board. 

 

I would just it all to go up in smoke at this stage.....

[mwl168]

A bit more information will help others to offer their assistance.

What version of boards are you building? What voltage PSU are you running? What's the offset and balance at cold start and are they responding to adjustment? 

A few good resolution photos of your boards will be very helpful. 

Edited February 6, 2016 by mwl168

[spritzer]

Post some pics of the boards as then stuff is easier to spot. 

A super bright LED is not normal...

[DHTGUY]

I've got +450 and -450 on the output at the stax socket on the one board I have connected. They are rev 0.63 amp boards. The PSU is 450v and running fine (rev 1.0). All LEDs are lit, but the one between the two MPSw06 (I am using 2SC2240 in there) is a lot brighter. The servo is not connected, the jumpers are on global feedback. The PSU is not drawing much current ( less than 1A). I got most parts from Mouser, I am using the SA1968

I am not using the transistors on the silk screen (right), but the ones on the left in the list below - I had an excel list in which these were given as equivalent. 

2SC2705 / MPSW42

2SC2240 / MPSW06

2SA970 / MPSW56

2Sc1815 / MPSW06

I got the 2SA1486 from little diode in the UK. I previously posted pictures, and changed a number of resistors to half watt. 

I'll try the other board to see if that is OK or not, and post some better images back & front when I have removed the board. 

 

 

[JoaMat]

You can’t just replace the mpsw with the 2sa/2sc components that way, can you? I think their pin layout is different.

[mwl168]

JoaMat is correct - MPSW06 pinout is different than the 2SC1815.

[congo5]

It seems that there were dual transistor boards with the extra hole .... But that one is MPSW only.

the base is in the middle not on a end... EBC vrs ECB

I have MPSW if you need a few

[Skooby]

I too have used MPSW/As instead of SC/SA (my board is rev 1.0).  

My LED between the MPSW/A's didn't light up at first but need to adjust the 2K bias pot to get 0V between +/- outputs.  May be want to adjust the 1K offset pot too.  

Things survived the first 30 sec, sounds promising.  63mm height heatsink?

[spritzer]

Yeah only the MPSW and MPSA transitors work on that board.  Is it also a reflection or are the 1486's in backwards? 

Also those thick film resistors will work but not a great idea in the feedback line.  Also all those 1/4W resistors have to go as those are the spots that either need higher power handling or more voltage handling. 

[DHTGUY]

Thanks everyone- looks like l know where to start. I would be really grateful congo5 for some MPSWs - will you pm me? I checked the 1486 - the metal side is above the hatching on the silkscreen, so I think it was a reflection. 

On 1/4w and 1/2w - I will check again, but some of the smaller resistors are rated 1/2W. They are vishay or xicon. I used thick film due to the following statement by Kevin Gilmore "

"...The standard dale 1% rn60 resistors are all spiral trimmed to value and 

are therefore slightly inductive, and the inductance changes with the value.

 

I have found that the absolute best resistors for this are the thick film on ceramic substrate. But they

are very fragile (crack easily if you overheat when you solder) and 150 times the price of regular resistors."

Why should they not be a great idea? 

Any ideas if the board is going to be ruined? 

[mypasswordis]

https://www.digikey.com/Web%20Export/Supplier%20Content/Stackpole_738/PDF/Stackpole_ThickFilmXThinFilm.pdf?redirected=1

KG was probably referring to something like the Caddock non-inductive film resistors?

Have you tested the 2sa1486es to make sure they're real? I think it's easier to just use stn9360s; they're cheaper than any real 2sa1486es and in production so no worries about fakes.

[spritzer]

Yeah Kevin was talking about the TI super expensive film resistors or the crazy pants stuff found inside high end multimeters.  Those are power resistors so not the same thing. 

 

[kevin gilmore]

http://gilmore.chem.northwestern.edu/g9f21766.jpg

 

[spritzer]

Well it is either that or start trimming ones own with a laser. 

[nopants]

if you can do that can't you make your own dac chips too

[kevin gilmore]

the stax mafia are definitely planning on a multibit dac. Problem is finding switches good enough (and fast enough) to do 24 bits, 1/2 lsb

 

[mtoc]

On ‎2016‎年‎2‎月‎8‎日 at 8:11 AM, kevin gilmore said:

the stax mafia are definitely planning on a multibit dac. Problem is finding switches good enough (and fast enough) to do 24 bits, 1/2 lsb

 

wow! have you guys find the good switches?

[astrostar59]

On 7 Feb 2016 at 1:11 AM, kevin gilmore said:

the stax mafia are definitely planning on a multibit dac. Problem is finding switches good enough (and fast enough) to do 24 bits, 1/2 lsb

 

Interesting. I was looking at this as a project using the Soekris board.  But looks like the guy has stopped making them. There are a few newbies on the block developing this right now.  A nice tube line stage would make a very nice unit indeed. TotalDAC watch out.....

[kevin gilmore]

The soekris board is a perfect example of fucked up.  Lvc595 chips where the 0 and 1 can be 60 mv different between drivers on the same chip let alone matching to the next chip. Then the crap 4v references. So when you test for missing codes and equivalent of stuck bits, major issues. Then the drift with temperature.  Super expensive stuff (MSB) that use data general series cmos switches not much better. Hand picked discrete fets with hand picked clamps are superior. And discrete. And you will be stuffing parts forever.

no tubes. The reason should be obvious

Edited April 23, 2016 by kevin gilmore