KGSSHV Carbon Build Thread

[mwl168]

On 8/22/2016 at 2:37 PM, insanity said:

Ok then I think I have a bigger problem. I took out an LV board and tested it out of circuit without a load and it measured way above the 15v it should have... 

This sucks... Looks like a major investigation... 

Sorry for being thick - but why would a short on the B- header take out the LV regulator?

[insanity]

18 minutes ago, mwl168 said:

Sorry for being thick - but why would a short on the B- header take out the LV regulator?

maybe there was a voltage spike on the ground because the trace the B- shorted to is going to gnd over a 25k resistor. 

As a matter of fact both of them dont work anymore. I have already taken one apart and tested both lt1021. one of the two lt1021 is fried. the other one works. 

Can anyone give me a tip on an easy way to test the opa134?

Edited August 24, 2016 by insanity

[insanity]

During the repair process of my carbon I also checked on the other channel how it looks under the terminal block. There was no short, but it looked like it would have failed as well within some time. 

I must therefore recommend for everyone using the v0.5 boards not to use such terminal blocks as I have, even if they are rated 400v. I they are mounted on the reverse side, it shoud be fine. But on the top side with that trace going right below the angled metal pin of the terminal block failure is programmed. FYI these are Phoenix terminal blocks from mouser and not some cheapo ebay stuff. 

After all I think I might be able to repair the amp board. 

The consequence of the -400 shorting to that trace caused a failure in both servos. The 1n914 were all shorted leading to a full short between +15 and -15. The 12V protection zener of the servo was shorted. On one board, the 4n25 measured a bit strange, so I will replace it. I don't know how to check the opamps. Will probably just replace them not to risk anything. 

Furthermore the GRLV both suffered severe damage. I didn't bother to check which parts were dead, so I just replaced all the active parts and they seem to work again. There were no overheated parts or so, because I switched of immediately after the failure. I guess it was a voltage spike thing. What I know is that on one of the GRLVs the 1n4007 was also shorted, which is rated 1000v 1A, so here it is probably a current issue. 

[sorenb]

37 minutes ago, insanity said:

There was no short, but it looked like it would have failed as well within some time. 

How did it look? did you by any chance take a photo?

[insanity]

Took a picture for you ;-)

This is no coincidence. And I cannot clean it off. 

On the broken board, I plan to bridge the trace with a wire and put some kapton tape below the terminal blocks. 

Edited August 27, 2016 by insanity

[kevin gilmore]

has to be the connector you used getting too close to that trace.

I can move the trace a bit, in fact I just did, but better to pick a different connector.

(not posted yet)

[johnwmclean]

Insanity did that board also have flux leakage under the terminal block?

[insanity]

there was also some flux, yes. Being it my soldering or not, since its not possible to check after soldering the terminal if there are any residues, it is unwise to continue building with this risk. 

 

Concerning better teminal blocks. I have done an extensive search for more suitable terminal blocks. As we already know, there are none rated more than 400V with a pitch of 5mm. 

During the search I found a lot of terminal blocks like this: 

They are usually only rated 250V oder 300V. Would you risk using those? Sofar the only other options seems to solder the cables directly to the board, but that is very unpractical. 

[sorenb]

As I respect the findings of Insanity, I took my Carbon apart. It is different from most here, as I have used boards from Spritzer that features a ground plane (like the board from the GB Carbon V7). The ground plane should just make things even worse.
Anyways, good news is that I found NOTHING, nothing at all ...I have had this amp running almost 24x7 for about a year so I guess any error should have occured anyways..

What I noticed is that Insanity hasn't really been shy with solder, I on the other hand tend to be somewhat cheap (in general). I suppose that a really heated leg of a terminal block and generous on the solder would 'grow' the leg on the opposite side and thus might damage the isolation ...as this is what I suspect has happen from seeing the latest picture

As more isolated terminal blocks might be a good idea, it might not have prevented this in the first place ...

[spritzer]

As Soren states, my boards are very different as I never route this close to the connectors.  The ground plane has a set clearance so that is never an issue once that is set correctly. 

One thing does spring to mind and that is the quality of the boards.  PCB's are not created equal and what is fine for LV stuff can fail at HV.  Just look at the lil'knight boards from back in the day.  Utter garbage and a source of a lot of headaches. 

[insanity]

those are the boards from the last group buy. i dont think these are bad quality.

[sorenb]

1 hour ago, spritzer said:

As Soren states, my boards are very different as I never route this close to the connectors.  The ground plane has a set clearance so that is never an issue once that is set correctly. 

One thing does spring to mind and that is the quality of the boards.  PCB's are not created equal and what is fine for LV stuff can fail at HV.  Just look at the lil'knight boards from back in the day.  Utter garbage and a source of a lot of headaches. 

In other words Kevin doesn't know how to layout a board? The quality of PCBNet boards a way better in comparison to the boards I got from you anyways..

[Skooby]

Not a fan of termination blocks,  Wire fray and screw down are too iffy.  I just solder 18 awg wires straight to the board.   

[spritzer]

3 hours ago, sorenb said:

In other words Kevin doesn't know how to layout a board? The quality of PCBNet boards a way better in comparison to the boards I got from you anyways..

Only an idiot would draw those conclusions and how do you know they are better quality?  I take it you have them do an insulation test on them? 

[sorenb]

7 minutes ago, spritzer said:

Only an idiot would draw those conclusions and how do you know they are better quality?  I take it you have them do an insulation test on them? 

Oh, you actually tested the Group Buy boards and didn't tell the result? at least the pads don't fall off the group buy boards ...

[johnwmclean]

The Pheonix terminal blocks do not like heat, you have to be quick soldering not to compromise their integrity.

I don’t have any issue with flux creeping through to the reverse side of the pcb when soldering, all the flux is contained within side I’m soldering, I love Kester.

[gepardcv]

My observations match @johnwmclean's. I don't really want to disassemble my Carbon to check under the terminal blocks, but I never noticed flux on the side of the PCB opposite the one which I have been soldering. Maybe I'm just in denial. It would be great to find a nice alternative to Phoenix, but none of the connector blocks with the right spacing which I found are rated for ≥400V.

[sorenb]

Did a short run of isolations tests on the pcbs I have at hand (Carbon v5,v6, circlotron driver board) , they all seems to do +1kV fine (including one from Spritzer, in all fairness)

[GeorgeP]

I think Soren and Kevin are right on this - more likely than not it is a generous amount of solder on the underside the permitted an eventual short through the solder mask. I think the flux is likely a red herring.

[MLA]

6 hours ago, sorenb said:

Did a short run of isolations tests on the pcbs I have at hand (Carbon v5,v6, circlotron driver board) , they all seems to do +1kV fine (including one from Spritzer, in all fairness)

While at it, would it be possible to test some terminal blocks as well? I've tried to figure out what the UI tests for those 300-400V ratings actually mean. I don't get it fully, but DC seems less of a problem than AC in those tests...

[insanity]

Ok I agree that if you don't have flux under the terminal block, it will probably not be an issue. Nevertheless, terminal blocks take a bit more solder to fill the holes compared to other components. 

To conclude: I think the boards are fine to use, as long as you are very careful about the soldering of the blocks. Furthermore, the ORIENTATION of the terminal block is curcial for the problem to appear. If you mount on the reverse side, there will be no problem. If you mount on the regular side, it may also matter where the openings of the terminal blocks point to (although there are traces running on both sides of the terminal blocks). In my project, the holes were pointing to the c2ms. 

One could also just put some insulating material over both traces, be it kapton tape or something else that is non conductive. 

[headinclouds]

It does look like the solder is spreading and causing the problem.  We are normally soldering with the board inverted -joint on top.  If too much solder is applied excess runs through and spreads underneath, but should tend only run on hot metal unless capillary action   The kapton tape was used to protect against shorts to traces /components above the surface, for example, in the case of big caps mounted on the PS with ground planes. If we were to put kapton tape on the surface I suspect it would not necessarily stop solder running under it (capillary action too) and causing the same problem.  I think the solution is care with the soldering but better, moving the trace as Kevin has done.

[insanity]

Red: soldering pad

Green: Traces

Blue: Suggestion for kapton tape, although depicted otherwise, I would not  put it over the red pad. Just over the green trace. 

Yellow arrows: Metal of the terminalblock coming very close to the green trace. 

FYI: I used the terminal without the additional test point, but the crucial construction point is identical to the blue prints below. 

I think the problem was that the solder/flux ran along the bent pin of the terminal block coming very close to the green trace. I think one or two layers of kapton should work because it slightly lifts the terminal block form the board and provides addition insulation between the bend pin and the trace. If place correctly, I think its very unlikely that either solder or flux would go under the kapton tape.

Edited August 28, 2016 by insanity

[headinclouds]

Yes you have to try it.  I would like to raise the terminals more but they rely on the surface to steady them.

[spritzer]

As I said, this is a solder mask issue.