KGSSHV Carbon Build Thread

[Blueman2]

Not sure I would want to run this when the LV is not working correctly.  I would wait to get that nailed before even powering the Carbon boards.

For the LV trafo, it is a 2x16, right?  For me, it DID matter which wires you put together for the CT connection on the board.  If it is right, the other 2 will measure 35V or so.  If it is wrong, they will measure only 17.  Try that.

I will say I like the Molex connectors.  Never seen that before and it has some real advantages.  

EDIT: Yes, I just checked my amp, and I had Color1 to one 15V, Color2 to other 15V, and Color1/2 together on CT.  It looks like you have it different?

Edited January 15, 2018 by Blueman2

[sorenb]

1 hour ago, bui501 said:

Problem 1:  The low voltage outputs from the GRHV boards goes up to only +/-10V even at 220VAC input on the transformer (17.8VAC unloaded measurement from the transformer's output, low voltage winding).

Check your CT from the transformer  is really going to CT at the board.

1 hour ago, bui501 said:

Problem 2:  I hooked up the power supply boards to the carbon boards anyways and started to play with the current adjustment pots.  At the test point FURTHEST away from the servo 2 jumpers, I can get the current to 20mA (started out at 18mA).  However, at the test point/pot closest to the servo 2 jumpers, the current started out "normal" at close to 20mA, but after 10 or 20 seconds, it rose to 36mA -- I cannot/cannot get the current to go much lower at that test point.  This is the same for BOTH amp boards, so I'm assuming it's a problem with the PSU boards.

Sounds like thermal runaway for the 10m90; the dissipate as much heat as the SiC and thus needs to be properly heatsinked.
Cannot see from your pics how you have attached the 10m90 to the sink ...use an alu oxi pad, thermal past on both sides of the alu oxi pad, and a screw to tighten )
If not the above check the CSS using 2xbattery9V or similar, and see if you can adjust the current - if not, the DN2540 has probably gone south

[wink]

Looks like the 10m90's aren't screwed down to the heatsink yet.

Best to wait until the alu pads arrive and all is fixed to the sink before even thinking of powering up.....

[bui501]

8 hours ago, sorenb said:

Check your CT from the transformer  is really going to CT at the board.

Sounds like thermal runaway for the 10m90; the dissipate as much heat as the SiC and thus needs to be properly heatsinked.
Cannot see from your pics how you have attached the 10m90 to the sink ...use an alu oxi pad, thermal past on both sides of the alu oxi pad, and a screw to tighten )
If not the above check the CSS using 2xbattery9V or similar, and see if you can adjust the current - if not, the DN2540 has probably gone south

Damn it!  I think you've nailed it, Sorenb and Wink...  The 10m90s and SiC's are not properly attached to the heat sinks yet -- I don't have the thermal paste and all the aluminum oxide pads yet as they're still on their way from Mouser.  I didn't think the 10m90s and the SiCs would get hot quickly enough to cause a problem if I powered up everything for "just" long enough to adjust the current...  The metal tab on one of the SiCs got hot enough to actually melt!!!

Thanks guys for your help.  I'll try again after the pads and paste arrive and report back.  AARRRGGHHH....  This waiting for parts  is killing me.

[bui501]

18 hours ago, Blueman2 said:

Not sure I would want to run this when the LV is not working correctly.  I would wait to get that nailed before even powering the Carbon boards.

For the LV trafo, it is a 2x16, right?  For me, it DID matter which wires you put together for the CT connection on the board.  If it is right, the other 2 will measure 35V or so.  If it is wrong, they will measure only 17.  Try that.

I will say I like the Molex connectors.  Never seen that before and it has some real advantages.  

EDIT: Yes, I just checked my amp, and I had Color1 to one 15V, Color2 to other 15V, and Color1/2 together on CT.  It looks like you have it different?

Thanks, Blueman2.  I'm double checking the CT connection from the transformer tonight.  

Re. the molex connectors, I had used something similar for the initial setup when I built my kgsshv -- it really helped when disconnecting/reconnecting everything during the initial testing/setup stage, but I had to remove those molex connectors for my kgsshv because they were only rated for 250v.  The ones you see on my in-progress carbon build are rated for 500v, so I should be able to leave them on permanently.  Makes it really easy to move things around during troubleshooting even though they don't look as nice and neat as the phoenix terminals.

[chinsettawong]

 

14 hours ago, bui501 said:

Thanks guys for your help.  I'll try again after the pads and paste arrive and report back.  AARRRGGHHH....  This waiting for parts  is killing me.

I always keep many parts as spares.  If you need anything fast, let me know.

[gepardcv]

You might have fried all those transistors, so if startup does not work sinked, you should probably replace them...

[GeorgeP]

On 1/6/2018 at 6:57 PM, Blueman2 said:

I was about to pull the trigger on another set of GRHV boards for another project, but see that the C2M1000170D SIC MOSFETs are out of stock at Mouser and Digikey.  Would this part work as a replacement?  755-SCT2H12NZGC11.  Biggest difference is the much slower fall time (74ns vs the 9ns for the Cree).  Thoughts?  

Alternatively, I have a couple FQFP8N80C's on hand as well if those would work.  Again, this is for the GRHV PSU, not for the amp itself.  

 

Newark has a the SIC MOSFETs in stock.

[Blueman2]

Great news!  But I just finished soldering in a pair of SCT2H12NZGC11's into my GRHV boards about 30 minutes ago!  I will give them a try to see how they perform.  It will be a fun experiment to have a backup SKU if needed in the future.  

[bui501]

On 1/16/2018 at 11:07 PM, chinsettawong said:

 

I always keep many parts as spares.  If you need anything fast, let me know.

Thanks, Wachara! I may have to take you up on your offer. My thermal pads should be arriving in the next day or two, but it looks like the replacement  SiCs I ordered from Newark haven’t even been shipped out yet.  

On 1/17/2018 at 5:07 AM, GeorgeP said:

Newark has a the SIC MOSFETs in stock.

I placed an order with them for some SiCs several  days ago.  Hopefully, their online stock status is accurate.

On 1/17/2018 at 1:12 AM, gepardcv said:

You might have fried all those transistors, so if startup does not work sinked, you should probably replace them...

I think you’re right.  The one with the melted metal tab is probably toast.  I just didn’t expect the temperature to get SO high SO quickly. :-(

Edited January 17, 2018 by bui501

[JimL]

On 1/15/2018 at 6:20 PM, bui501 said:

 I didn't think the 10m90s and the SiCs would get hot quickly enough to cause a problem if I powered up everything for "just" long enough to adjust the current...  The metal tab on one of the SiCs got hot enough to actually melt!!!

So, here's where a little math beforehand would have helped out. Now, the 10M90S is specified as having a thermal resistance of 80K/W junction to ambient (K = Kelvin, so 0 degrees Kelvin is absolute zero, each degree Kelvin = 1 degree centigrade), and the C2M10000170D is specified as having a thermal resistance of 40K/W junction to ambient. As these devices are rated to run below 150 degrees centigrade, what this means is that the 10M90S without heatsink is rated for less than 2 watts, and the C2M is rated for less than 4 watts.  The 10M90S split the PS voltage between them (for as long as they were functioning), so 400 volts each at 20 mA = 8 watts.  Running a transistor at 2 and 4 times their rated power dissipation will blow them in short order. If you've ever seen the actual size of the transistor inside its case you will understand why they can get fried quickly by being overstressed.

 

Therefore, you need to replace ALL the C2Ms,all the constant current source 10M90S, and all the DN2540s as well, unless you're absolutely confident that the other devices protected them like a fuse. Which I doubt. Murphy's Law: a transistor protected by a fuse will blow first, protecting the fuse.

Edited January 18, 2018 by JimL

[bui501]

Thanks, Jim. I’ve got all three transistors on order for replacement.  I was so impatient to see the boards all powered up, now I have to wait longer for replacement parts.

Besides the math and patience, a bit of common sense could have come in handy:  if you guys are all bolting huge heat sinks onto these transistors, it probably should have been a clue that I probably shouldn’t have fired them up without any heatsink. 

[wink]

Yep, three principle to keep in mind.

Knowledge - The facts of the matter.

Wisdon - Knowing how to put the knowledge into practice.

Prudence - Doing it.

[Blueman2]

On 1/6/2018 at 3:57 PM, Blueman2 said:

I was about to pull the trigger on another set of GRHV boards for another project, but see that the C2M1000170D SIC MOSFETs are out of stock at Mouser and Digikey.  Would this part work as a replacement?  755-SCT2H12NZGC11.  Biggest difference is the much slower fall time (74ns vs the 9ns for the Cree).  Thoughts?  

 

On 1/7/2018 at 9:19 AM, Blueman2 said:

Here is the parts comparison:

 

 

TLDR Update: The replacement part does indeed work just fine!  I have it running in the power supplies of my Blue Hawaii without issue.  Comments below were before I figured out some other issues I as having.    

Well, things did not blow up, but I have run into a problem using the part.  I tested both PSUs and they tested fine unloaded.  +404 and -410.  I then loaded with 15K 20W resistor on each supply, and the voltages held perfectly.  I then hooked up to a new Blue Hawaii BJT I am building, and each side of the BH worked, though voltages did drop a bit from the +404 and -410.  The B- was the most impacted, dropping to -400 but still working.  I then hooked up both sides of the Blue Hawaii and voltages on both B+ and B- dropped by several hundred volts.  

My conclusion is that the new part is for some reason not able to keep the current levels I need.  What surprises me is that  PSUs could drive the 15K resistor without any issues at all.  I guess the Blue Hawaii BJT demands a heck of a lot more power than the 10 Watts demanded by the 15K resistors.  I will try loading both resistors on B- rail (so 7500 ohms) to give that rail a 20W test.  But at this point, it appears my gamble did not pay off in using the alternative part.   (This was a wrong assumption.  Apparently the part is just fine.  Read about 6 posts down)

Was worth a try I guess!!  

 

EDIT:  Loading the 15K power resistors in parallel (doubling the draw to 20W) did drop the voltage down about 2 volts on each rail (to +402 and -408).  So that should not happen, right?  I do not have any lower value high wattage resistors to test with, but clearly something is wrong with both PSUs, and probably the part I used SiC.   Only other possible issue could be the transformer.  I will double check unregulated voltage to see if that is staying stable....

EDIT2: Well, maybe it is the trafo??  Trafo is giving 312 VAC on each side.  Unregulated (after rectifier), I am measuring 436 VDC on both sides.  When loaded, the Positive side unregulated voltage drops to 395 for a second then recovers to 407.  The Negative side unregulated (which takes the most drain when using a Blue Hawaii) drops to 390V then recovers to 393V, and then slowly drops from there.  Clearly, the trafo is not keeping up.  It is a brand new one from Antek.  I measured about 50VA HV load from side of the Blue Hawaii.  The trafo is rated for 100VA per side, so I do not appear to be exceeding the trafo rating.    I think I will pull a trafo from one of my SRX-Plus builds to see if I can confirm the issue.  Maybe I was too quick to blame the replacement SiC? 

Edited January 23, 2018 by Blueman2

[mwl168]

Not sure this is a factor but, the Blue Hawaii draws significantly higher current than the Carbon does, especially on the B- rail. If you are using the 5.1R current limiting resistors on the GRHV you will for sure have issue with the B- and possibly the B+ rail too.

[Blueman2]

2 minutes ago, mwl168 said:

Not sure this is a factor but, the Blue Hawaii draws significantly higher current than the Carbon does, especially on the B- rail. If you are using the 5.1R current limiting resistors on the GRHV you will for sure have issue with the B- and possibly the B+ rail too.

I am using the 5W 5R1 resistors.  But not sure that explains the drop in unregulated voltage I am seeing.  For the 5R1, is it the wattage or the value that needs adjusting?   

[mwl168]

The value needs adjusting not the wattage. Don't remember for sure but try 3R 5W or so. In a pinch you can parallel 2 5.1R to get 2.55R but it's a good idea to replace it with a value that's more appropriate for your application. Going by memory I think the B- on the BH draws around 140mA.

[Blueman2]

BTW, I am measuring 0.68V across my B- 5R1 resistor.  So about 133ma, about what you predicted for B- rail.  The voltage then drops slowly over time across both the current limiting resistor and the B- output itself.  I am still of the mind that either the trafo is not keeping up or the SiC chip I am using is not working in the same manner that the original C2M was designed.  And given the dropping unregulated voltage, I am thinking trafo issue.  (see below; current limiting resistor is more likely cause, or at least, a contributing factor)

Edited January 20, 2018 by Blueman2

[mypasswordis]

Which Antek did you get? 0.68V means it is hitting the current limit so you would have to lower the resistance a bit regardless. 

[Blueman2]

Ah, I was not understanding how the current limiting resistor works.  Looking at the schematic, does it give a reference voltage for the 2n3904 and that feeds back to the C2M? Doh!  

Sorry, mwl168, I was being dense.  Now I understand the importance of that resistor!!

Out of curiosity, how do I calculate the current limit imposed by the R11 (the current limiting resistor) in the circuit?   What limit does the existing 5R1 impose?  

 

EDIT: Partial success!!  I changed the current limiting resistors to 2R5 (paralleled 2 5R1s as mwl168 mentioned) and things did get better.  Voltages no longer drop off over time.  But bad news is they are still both around 390V, and fluxuate a bit as household devices such as refrigerators turn on/off.  But stays in a range of about 390 +/-3v.  Enough to allow me to get offsets down to under a volt and confirm the outputs to 20ma.  

As for the low voltage, I think I will now replace the trafo with one I have in one of SRX builds and see if that helps.  

Thanks mwl168 and mypasswordis!!!!  Both for your guidance and your patience as I learn along the way....

 

EDIT2: Oh, and I could not resist giving it a try.......WE HAVE MUSIC!!!

 

And sorry for polluting the Carbon thread with what turns out to be Blue Hawaii BJT issues.  There was more discussion on the GRHV boards here, which is where I was guessing my issues where.  I am moving the discussion back to the Blue Hawaii BJT build thread.

EDIT3:  Just for future reference, it appears that the SCT2H12NZGC11 does indeed work as a replacement for the C2M1000170D, so we appear to now have a backup when those go out of stock as they seem to do pretty often.  

Edited January 21, 2018 by Blueman2

[Blueman2]

I currently have the Op27a based servo (Servo 1) in use on my Carbon.  Works great.  I just got hold of a couple 4N25's for the Optocoupler (Servo 2) and was going to plug them into the board.  Is there any advantage to the optocoupler servo over the op-amp servo?  Just curious why both were put on the board.  For now, I am leaving Servo 1 enabled and Servo 2 disabled.  

[sorenb]

27 minutes ago, Blueman2 said:

I currently have the Op27a based servo (Servo 1) in use on my Carbon.  Works great.  I just got hold of a couple 4N25's for the Optocoupler (Servo 2) and was going to plug them into the board.  Is there any advantage to the optocoupler servo over the op-amp servo?  Just curious why both were put on the board.  For now, I am leaving Servo 1 enabled and Servo 2 disabled.  

in any case the OpAmp is needed. Servo1 acts on the input, and Servo2 on the output stage rather. You need  some ~ +15V offset to use the Servo2.

[Blueman2]

Thanks.  Sorry for my confusion on the OpAmp.  But is there any advantage to engaging Servo2 rather than Servo1?  As I recall, only one should be enabled at a time.    

[sorenb]

1 minute ago, Blueman2 said:

Thanks.  Sorry for my confusion on the OpAmp.  But is there any advantage to engaging Servo2 rather than Servo1?  As I recall, only one should be enabled at a time.    

only one at a time. Both will keep the amp offset ~ zero, and both are altering the operation point of the particular stage.

[JoaMat]

Nothing wrong with Servo 2 - the only servo you need.