i'm on a roll... the kgsshv

[livewire]

Correction: XLR pin1=ground

[Craig Sawyers]

I have a question for the gurus. What is the best way to ground this thing?

I want to avoid ground loops and due to the high voltage on the supply rails,

If you are running balanced inputs, just ground the amp to the chassis. If you are running single ended, there is a risk of hum loops. So what you do is hard ground the mains green/yellow to chassis - which you have to do in any case for safety. Then you ground the amp to the chassis via a 3W resistor in the range 1k to 5k (value not too critical), preferably non-inductive metal oxide, and keep the leads short. Professional audio gear, where roadies have to wire truck loads of gear together without hum, tends to have a three position switch - grounded, ungrounded (ie floating), or ground lift (the resistor).

[spritzer]

The reason why all the mounting holes are isolated is to control the ground paths and stop any loops from emerging. High voltage or not, ground is just that but the aim here is to control how it is connected. I for one star ground everything so the amp boards, chassis earth connection and input ground are all connected to the PSU board at one spot.

[Craig Sawyers]

The reason why all the mounting holes are isolated is to control the ground paths and stop any loops from emerging. High voltage or not, ground is just that but the aim here is to control how it is connected. I for one star ground everything so the amp boards, chassis earth connection and input ground are all connected to the PSU board at one spot.

I forgot to say - not connecting the amp ground to chassis safety ground (either hard grounding as Spritzer does, or ground lifting via a resistor as I prefer) is seriously bad in an electrostatic amp that can kill. Always connect the chassis to the green/yellow safety incoming mains in some way that cannot possibly come adrift; star washers to bare metal are manditory, as is removing paint or anodising where the chassis component bolt together to ensure the chassis is bonded to safety ground. Most commercial chassis (including all the Hammond ones I've used) omit this point entirely and just anodise or powder coat each plate so they are nicely insulated from one another, leaving it up to the user to take necessary measures to prevent death.

[livewire]

Thank you Craig and Spritzer so much for clarifying the above details. I will proceed from there.

And thanks to Marc and the others for your input on the inrush.

I will leave those things alone for now.

Lastly, many thanks to KG for bestowing all of your fantastic designs upon the world.

Happy New Years to you all !!!

[rille]

I have a set of PCBs here and have still a question about the transformer. I read earlier in the thread 475 volt and 39VA each winding. Is this still up to date? The board is marked with 485V. I guess this is for 500V DC? How much voltage difference does the regulator need?

[kevin gilmore]

Whether you run the output at 450v or 500v, you still need about 625 vdc unregulated for the bias regulator.

So 485 vac (center tap is better but not required) is the target input voltage.

[rille]

Thanks for the quick answer. Is 40VA per winding sufficient?

P.S. A info for the guys from outside the US. Littlediode also has the IXCP10M90S. The price is

[kevin gilmore]

i spec'd 100 ma per high voltage winding, just in case you want to turn the bias up a bit.

[Lil' Knight]

Here's the design number from SumR if anyone needs it:

The design # RC0120 064 1.

+ 2x 485v/100mA

+ 36VCT/500mA

With Primary 2 x 115V.

A/ Encapsulated into toroid container: US $ 96/e

B/ Without container, but with core band for shielding: US $ 75/e.

[rille]

Is there an alternative for the IXTP01N100D?

I guess the drain to source voltage is the most important point? What about the STF5NK100Z or STP8NK100Z from ST? They have also 1000V and 30V gate to source voltage. And are easier to get outside the US.

[luvdunhill]

are those DMOS devices?

[rille]

I'm not sure. Here are the datasheets: http://www.st.com/stonline/books/pdf/docs/10850.pdf and http://www.st.com/stonline/books/pdf/docs/11781.pdf.

[kevin gilmore]

Neither of those are going to work. You need a 1000 volt or more depletion mode mosfet that

likes to work at a couple of milliamps.

Use the 2sa1968 with +/-450 volt supplies, or use the ixys parts.

Same thing with the required ixys parts in the power supply.

[wink]

The 561R10TCCV50 5pF 1KV caps finally arrived from RS Components.

When I opened the package the plastic pack was from alliedelec part number 507-0763.

Just looked www.alliedelec.com up. They have 848 in stock.

Add to

compare

Vishay Specialty Capacitors

CAPACITOR, CERAMIC, PRECISION, DISC, RADIAL 1000V, C0G, SIZE C, 5PF

Mfr's Part #: 561R10TCCV50

Allied Stk #: 507-0763

In Stock: 848 On Order: 4000 1+ $0.4225+ $0.377100+ $0.315250+ $0.291Each

[livewire]

Ok, I'm still haunted about physically grounding the power supply DC ground plane to chassis (AC) ground.

My multimeter shows a difference of 180VAC between the two. Spooky!

The meter shows zero differential for DC, so I guess that's OK.

I'm running a transformer with separate 475V secondaries, no grounded center tap.

Therefore I'm not using the center tap terminals on the power supply board.

I'm pretty certain that this is what's messing with my head.

The 180VAC component is 1/3 of the unloaded 540VAC of one secondary winding.

So, should I just go ahead and connect the two? (AC ground to DC ground)

Or maybe reconfigure the separate trafo secondaries as one with a grounded center tap?

I'm really not in the mood for seeing fireworks....

[spritzer]

You don't need the CT to create ground but something is up with that AC. How is the transformer mounted? Toroidal's are picky when it comes to that...

[kevin gilmore]

leakage current due to capacitance in the transformer...

hook power supply ground to AC neutral thru say a 1K resistor

and measure again. leakage current should be no more

than about 1 ma.

[livewire]

The sparks were flying on my kitchen table last night!

Truly a Kodak moment (or two). Thought it was the 4th of July.

Checked my calendar and I was wrong....

I came home after a long day at work - eight hours of tracing faults in a computerized

hospital communications system. I was very tired and obviously not thinking well.

I powered up the kgsshv psu and proceeded to check all of the output voltages.

Everything was right on the money.

Next, it was time to do what KG suggested to tame my perceived grounding gremlins.

(actually Craig Sawyers suggested something similar earlier.)

But with 180VAC floating around where it shouldnt be, I wasnt ready to go there, yet.

So I wired a 1k resistor between the DC ground of the psu board and 120VAC neutral

and checked the current. Yup, the Doc was right, it was sitting right at 1ma. No more problem.

Everything was now grounded as it should be.

Now is where the fun began.

As I was looking at my minor accomplishment, with the psu still under power,

the snap of a spark with the resulting flash of diying parts erupted from one the 15 volt VR's.

Not sure which one yet. I wasnt touching a thing, the sucker just blew.

I immediately powered down the psu and unhooked the transformer, ran some resistance checks,

didnt find any shorted windings, so I hooked the HV secondaries back up to the psu board

while leaving the 15VAC sencondaries unhooked. I wanted to test the +/- 450V outputs to make

sure nothing was affected there. I didnt think there would be, with the exception of the trafo,

they are two compltetely independent circiuts, just on the same pc bpoard.

So I grabbed my DMM, switched it to the DC volts range, put the probes on the -450VDC terminal block

and almost went blind from the resulting flash - crackle - pop.

Anyone care to guess what just happened? To be continued...

[livewire]

Ha! I just noticed my fruedian slip above.

That one is sig worthy. I'm such a dork sometimes...

[spritzer]

I've had something similar happen once. I thought I'd save a bit and not solder in terminal blocks in every spot on a Blue Hawaii PSU which lead to a lot of fireworks when I decided to use one of the unpopulated spots during testing. A short to the between the two supplies will do that. Now this PSU is current limited so it might have survived the ordeal...

[livewire]

Nope, it did not survive. I see an order going to Mouser sometime soon.

It was my mistake in that I didnt re-plug the ammeter hot side (red) test lead back into the voltage position

before putting the probes on the terminal block. So, I induced a direct short.

My Fluke is fried as well.

The negative DC rail output now shows -200VDC instead of -450VDC.

The positive side is unscathed, it still reads +450VDC exactly.

Spritzer, FWIW on the trafo mounting I think that all is ok.

I used the supplied large rubber washers below and above with the steel washer pancaked on top.

The unit is mounted firmly, but not too tightly. It has a little wiggle room.

I used a 2.5 inch long 1/4-20 center bolt with a nylok nut on top.

[livewire]

In the future, I plan on having two separate DMM's on my test bench.

With different case colors too. One dedicated for current checks and the other for testing voltage.

I've done this before in the distant past. (not replugging the test leads into the correct holes)

[Craig Sawyers]

It was my mistake in that I didnt re-plug the ammeter hot side (red) test lead back into the voltage position

before putting the probes on the terminal block

That is a classic. If you haven't done that at least once, you haven't lived! Depending on the Fluke, you are likely only to have fried the input protection. There are a couple standard fuses, a couple of fusible resistors (which you have to get from Fluke), four thermistors (cheap standard parts) and some transistors wired as diodes. Some or all may have gone to heaven. You can tell from the way I rattled that off without looking at a schematic that I've been where you are now.

The lesson I've learnt the hard way on several occasions is don't test stuff when tired.

[spritzer]

Yeah, going without sleep can lead to very stupid stuff. Once I wired up a time delay circuit so that it shorted the neutral and live from the wall instead of passing it through. The gigantic spark did wake me up...