KGSSHV Carbon Build Thread

[sorenb]

1 minute ago, purk said:

It  opens up even more last night.  The depth of the soundstage more or less matched my T2 but the width is wider.  The T2 still sound more fluid and sweeter to my ears.  I still prefer the T2 overall.

Thanks.

How does the Carbon compare to your BHSE?

[GeorgeP]

1 hour ago, spritzer said:

A lot of this amp is an example of how not to wire any electrostatic amp.  There is so much fail in there but lets start off on the big stuff, how the caps are mounted...  Never do something like that!!!  Uninsulated spades on all of the AC input wiring when they clearly should be insulated.  Input wiring runs right under the transformer when it should be kept as far away from it as is possible.  Last but not least, the output wiring.  Twisting it like that is bad idea as it adds to the output capacitance of the amp. 

what would be the concern with the caps? Isn't it kind of like point to point wiring?

[purk]

38 minutes ago, sorenb said:

Thanks.

How does the Carbon compare to your BHSE?

I posted this on page 7 here.  Basically the BHSE still has a notch up in term of the tonality (likely due to tube here) but the presentation is more real on the Carbon.  It is just so 3-D and the soundstage width is just considerably wider.  Soundstage depth is better on the carbon by just a hair.  I would say they are about on the same level but different in term of an overall sound quality.

Edited December 4, 2015 by purk

[spritzer]

46 minutes ago, GeorgeP said:

what would be the concern with the caps? Isn't it kind of like point to point wiring?

I would never ship anything secured like this.  There is an awful lot of energy in those caps so if something goes wrong it will go bad in no time.  Why not just use the small caps which are the ones people are actually supposed to use here?  

Also has anybody measured the noise from the amp?  With the compromised ground on the circuit boards and the input/output wiring it could be pretty bad. 

[gepardcv]

2 hours ago, spritzer said:

Why not just use the small caps which are the ones people are actually supposed to use here?

The board is labeled for 680uF/550V, which purk used here. Looks like these caps came from the GB. What small caps do you mean?

[Pars]

Not that he was a reference on best practices, but Mike Elliot/Counterpoint used to do this kind of shit all the time on mods with 500V BlackGates and Cerafines, only they were just adhered to the board with clear silicone RTV. At least these look like they are secured with tie wraps. I don't really see a problem with that. With this case and amp, I wouldn't want the caps sticking out the top for aesthetic reasons.

 

The input wiring and ac wiring with uninsulated spades should definitely be addressed though IMO.

[spritzer]

1 hour ago, gepardcv said:

The board is labeled for 680uF/550V, which purk used here. Looks like these caps came from the GB. What small caps do you mean?

I would never use more than 330uf in this circuit which is what it was designed to use.  Two 680uf in parallel gives 340uf.  

[GeorgeP]

26 minutes ago, spritzer said:

I would never use more than 330uf in this circuit which is what it was designed to use.  Two 680uf in parallel gives 340uf.  

Just going by memory, but I thought Kevin suggested the 680uf 550v caps when he was designing the circuit and doing the board layout for the group buy.

 

[blubliss]

Now, now spritzer, be nice.  No need for the douchey tone.  Steve is an excellent DIY builder and don't worry, we won't step on your business. (and Steve does have two operating T2s under his belt)

Of course, we will make the suggested improvements (except the cap thing), thank you. 

Always room for improvement (: 

 

Edited December 5, 2015 by blubliss

[gepardcv]

2 hours ago, GeorgeP said:

Just going by memory, but I thought Kevin suggested the 680uf 550v caps when he was designing the circuit and doing the board layout for the group buy.

 

Are you thinking of this exchange?

With all our love of overkill, it certainly sounds like encouragement to use the 680uF caps. The group buy came along soon afterwards. 

[spritzer]

4 hours ago, blubliss said:

Now, now spritzer, be nice.  No need for the douchey tone.  Steve is an excellent DIY builder and don't worry, we won't step on your business. (and Steve does have two operating T2s under his belt)

Of course, we will make the suggested improvements (except the cap thing), thank you. 

Always room for improvement (: 

 

This is a place for people to learn and calling this "a perfectly executed build" needs to be called out.  Trust me on this, what others had to say about this build was far harsher. 

I for one couldn't care less if you are selling Carbons but if you are building our designs then do so properly.  That has always been the deal and the simple fact here is that this build is severely compromised. 

For the record, here is the original SiC PSU from February or so and what is the cap value? 

400V version for the BHSE but the same circuit.  Yeah some people wanted bigger and for a 3U chassis it isn't a problem but if the caps don't fit in the bloody box, use smaller caps. This PSU was also intended for the Circlotron which uses much more current, hence larger caps. 

Edited December 5, 2015 by spritzer
Image code...

[sorenb]

12 hours ago, spritzer said:

With the compromised ground on the circuit boards and the input/output wiring it could be pretty bad. 

Meaning that the GND from the PSU are connected two times at each AMP, and uses the signal GND for ref  for the HE90 voltage divider?

[sorenb]

35 minutes ago, spritzer said:

For the record, here is the original SiC PSU from February or so and what is the cap value? 

400V version for the BHSE but the same circuit.  Yeah some people wanted bigger and for a 3U chassis it isn't a problem but if the caps don't fit in the bloody box, use smaller caps. This PSU was also intended for the Circlotron which uses much more current, hence larger caps. 

That looks like your own personal edition of the "kgsshvpssicfetsmall3" having 330uF pre-reg, and 220uF post-reg - but as you like mini, you've replaced those 220uF for 100uF and shrinked it a bit to go from small to mini.

The only schematic the other persons did publish has 680uF pre- / post- reg, so how are people to know that 680uF caps is NOT what the PSU is designed for?

[kevin gilmore]

the fat power supply boards do say

"up to 680uf /550v"

i'm going to update the power supply boards today so that the power caps exactly line up when done mirror image.

currently there is 30th of difference which might be noticeable if you punch holes in the top of the case

 

[GeorgeP]

1 hour ago, sorenb said:

Meaning that the GND from the PSU are connected two times at each AMP, and uses the signal GND for ref  for the HE90 voltage divider?

I think he is referring to the fact that Kevin purposely left out the ground plane - though he said there would be no difference. Maybe Kerry can do some noise measurements when he finishes his.

[spritzer]

That is the cleaned up original of the PSU and you can see how all the other boards were based on this one.  This is a smaller version naturally as my boards are always much smaller. 

The ground plane was indeed removed because people had been having issues with parts being too close and it shorting.  The assembly needs to be perfect or you are in trouble.  For my boards there is a heavy ground plane on both sides of the board hence the swarm of via's all over the place. 

[sorenb]

4 minutes ago, GeorgeP said:

I think he is referring to the fact that Kevin purposely left out the ground plane - though he said there would be no difference. Maybe Kerry can do some noise measurements when he finishes his.

Meaning that any Carbon build will be "severely compromised" just by using the boards from the Group Buy?

[spritzer]

No but a shielded transformer is a must have and keeping all the wiring away from the transformer.  Here the input wires run directly next to the transformer. 

It must be said though that a ground plane is a better solution than running ground traces all over the boards.  Both in terms of shielding and the low potential of the ground but something had to be done with people shorting out the amps.  If we pull back the ground too much then it comes patchy as the part count goes up relative to the board size. 

[sorenb]

20 hours ago, spritzer said:

No but a shielded transformer is a must have and keeping all the wiring away from the transformer.  Here the input wires run directly next to the transformer. 

It must be said though that a ground plane is a better solution than running ground traces all over the boards.  Both in terms of shielding and the low potential of the ground but something had to be done with people shorting out the amps.  If we pull back the ground too much then it comes patchy as the part count goes up relative to the board size. 

Will there be new V5 and GR HV boards with ground plane? I guess, if people mount the components raised and stick to boards of good quality it shouldn't be a problem?

Edited December 6, 2015 by sorenb

[kevin gilmore]

if that is what people want.

what clearance do you want? I would think 30th minimum

any ground plane on the top of the amp board is going to be problematic due to all the surface mount

 

Edited December 5, 2015 by kevin gilmore

[GeorgeP]

With higher parts count and everything sharing the same ground plane, it would be questionable as to whether there would be any benefit - if I recall that was the reason Kevin stated in leaving it out. But as always the proof is in the pudding. For example the t2 has a massive ground plane, but many folks have experienced noise issues nonetheless.

[gepardcv]

If new board versions are in the cards, I'd much appreciate one without a line of angle bracket holes. That should make it possible to mount the boards vertically on a 100mm heatsink, making a nice "2.5U" enclosure. 

[spritzer]

5 hours ago, GeorgeP said:

With higher parts count and everything sharing the same ground plane, it would be questionable as to whether there would be any benefit - if I recall that was the reason Kevin stated in leaving it out. But as always the proof is in the pudding. For example the t2 has a massive ground plane, but many folks have experienced noise issues nonetheless.

Those noise issues have nothing to do with external sources but rather a flawed circuit design. 

[GeorgeP]

So we have gone from it being an excellent circuit that could not be improved upon to now a flawed circuit design? Clearly above my pay grid, but seems to be a lot of revisionism occurring of late.

[luvdunhill]

Perhaps a flawed implementation of an excellent circuit.