Chroma 61503 AC Source: 2SK1517 and 2SK975 Replacement

[TimNJ]

Hi everyone,

At work, we use (or should I say, used to use) a Chroma 61503 AC source to run certain tests for AC-DC switchmode converters. It can do arbitrary waveform generation: AC, AC+DC, and DC.

Long story short: I was attempting to reform a capacitor in DC mode, and the Chroma went *bang*.

More detail: In effort to reform the bulk cap, I  brought the voltage up slowly from 0 to 400V over the course of about an hour. (This was done with ~100K in series.) Then I shut off the source, and turned it back on a few seconds later without current limiting resistor  @400V DC and it really did not like that. The capacitor's leakage current had dropped to a few uA, so I figured it was okay... (Perhaps the cap had discharged enough in that time period to effectively put a short across the Chroma output.) As the Chroma is typically a very robust unit, I am surprised short-circuit or current limiting allowed it to self-destruct. 

Anyhoo...With my heart-pounding, I opened up the unit, which now smelled of "magic smoke". I identified that 4 out of 16 output transistors (4 sets of 4 in parallel) were blown to bits. It appears to be in some sort of bi-polar H-bridge configuration. Each set of 4 transistors is driven by its own gate drive transformer, which gives me hope that the failure is isolated to the one bank of transistors. (I hope.) The gate drive networks are also toast.

-----

The power MOSFETs are Hitachi/Renesas 2SK1517, TO-247, about year 2000 vintage, now obsolete. http://pdf.datasheetcatalog.com/datasheet/HitachiSemiconductor/mXyzvssw.pdf

The datasheet does not have gate charge parameters, only capacitance, which I think was typical for the day. I've read it's hard to make 1 to 1 comparisons with capacitance (Ciss, Coss, Crss)...so I'm asking for some feedback. I also noticed that 2SK1517 has zener clamp diodes on the gate. I don't know much about gate drive transformer circuits, but am wondering if there's a higher likelihood of spikes that might damage the gate.

2SK1517 Basic Parameters:

VDSmax: 450V
VGSmax: 30V
Id (peak): 80A
Vgs: 2V (min) - 3V (max)
RDS-ON: 0.25R (max)
Ciss: 3050pF
Coss: 940pF
Crss: 140pF
Trr (body diode): 120ns (typ)

So far, the closest "match" I can find is Vishay SiHG22N60EF. Although I design SMPS by day, I am not sure exactly the critical parameters for this topology. Usually lower capacitance is okay, but then switching speed might be too fast...or something of the like.

Vishay SiHG22N60EF Basic Parameters:

VDSmax: 600V
VGSmax: 30V
Id (peak): 46A
Vgs: 2V (min) - 4V (max)
RDS-ON: 0.182R (max)
Ciss: 1423pF
Coss: 73pF
Crss: 5pF
Trr (body diode): 113ns (typ)

-----

The gate drive MOSFETs are Hitachi/Renesas 2SK975, TO-92 (tall style), also obsolete: https://alltransistors.com/pdfview.php?doc=2sk975.pdf&dire=_renesas

The closest I can find is Diodes/Zetex ZVN4306AV

Hitachi 2SK975 Basic Parameters:

VDSmax: 60V
VGSmax: 20V
Vgs: 1V (min) - 2V (max)
RDS-ON: 0.40R (max) @ 10V
Ciss: 140pF
Coss: 70pF
Crss: 20pF

ZVN4306A Basic Parameters:

VDSmax: 60V
VGSmax: 20V
Vgs: 1.3V (min) - 3V (max)
RDS-ON: 0.33R (max) @ 10V
Ciss: 350pF
Coss: 140pF
Crss: 30pF

The capacitance numbers are off a bit (for both power and gate drive MOSFETs). I'm wondering if anyone can give me a "feeling" about whether they might be okay as replacements.

Thanks a lot.

Tim

[Jay_Diddy_B]

Hi TimNJ,

See the attached datasheet.
The Qg curves are on page 6.





The substitute MOSFETs look okay to me.

Jay_Diddy_B
 2sk1517.pdf (52.2 kB - downloaded 90 times.)

[TimNJ]

Hi TimNJ,

See the attached datasheet.
The Qg curves are on page 6.

(Attachment Link)



The substitute MOSFETs look okay to me.

Jay_Diddy_B
(Attachment Link)

Thanks a lot JDB.

I've looked at this exact curve (for different MOSFETs) about 10 times this year; You'd think I'd remember.

Surprisingly, basically nothing else available with low RDSON, low Vgs(th) in TO-92 these days.

I'll look around a little more for the power MOSFET. Since this looks like some sort of H-bridge-ish type thing, maybe the body diode characteristics are more important. I think the body diode is usually used as a "catch diode" to siphon off the inductive flyback current.

Thanks again!

[mairo]

Not exactly what you are after, but I have several Chroma DC supplies (62000P series) and few years ago one of them blew up. When I opened it the visual fault was in the AC-DC power supply board (they have very good PCB modular design on theirs equipment) but the damage was too big for a board level repair. I got in contact with the local distributor to see if they can source me a spare board, which they did after consulting with Chroma, I think I payed ~$700 for the board (mine was a RevA board and they send me RevE board , but they confirmed it was going to be OK). The model you are trying to repair I think is still in production, so an option is to get the whole board, or try ask them what transistors they hev moved to if the one failed in your board is no longer made (they do not have the best customer service when it comes to 'free' support ).

[TimNJ]

Not exactly what you are after, but I have several Chroma DC supplies (62000P series) and few years ago one of them blew up. When I opened it the visual fault was in the AC-DC power supply board (they have very good PCB modular design on theirs equipment) but the damage was too big for a board level repair. I got in contact with the local distributor to see if they can source me a spare board, which they did after consulting with Chroma, I think I payed ~$700 for the board (mine was a RevA board and they send me RevE board , but they confirmed it was going to be OK). The model you are trying to repair I think is still in production, so an option is to get the whole board, or try ask them what transistors they hev moved to if the one failed in your board is no longer made (they do not have the best customer service when it comes to 'free' support ).

Thanks a lot. Yes, I think buying a spare board will be option #2 if it blows a second time after replacing these parts...as long as the second power-on doesn't make a bigger mess of things. 

As far as I know, Chroma is pretty okay with spares/repairs, so I don't foresee it being a $20K issue (knock on wood).

I'm not optimistic they'd give me the new transistor P/N..although TiN (xDevs) just did a teardown of 65104 on Sunday on Youtube. Maybe I can ask him. I don't know the vintage of his unit, but it looked a bit newer than our 2002 model. Side note, very odd that TiN did a 61504 teardown the very same week that I'd blow up almost an identical unit. I thought "Hmm, that's neat, I'll probably never open up ours so it's cool to see it!" Lol.

[TimNJ]

After replacing one of the four failed output transistors and the respective drive circuitry, the 65103 is back up and running.

In the unlikely scenario that someone else has this problem and finds this thread, I used:

Vishay SiHG22N60EF

...and instead of the Diodes Inc part listed above

ON Semi FQU13N06LTU-WS

...as the pinout of the Diodes ZVN4306A did not match 2SK975. When looking at the front (flat-side) of the TO-92 package, 2SK975 is source-drain-gate. But, ZVN4306A is source-gate-drain. So, I opted for a TO-251 MOSFET which had the same pitch as the holes in the PCB. Leads slightly too large, but fit fine. TO-251 MOSFETs are almost always gate-drain-source, so you just have to rotate it 180 degrees.

The only concern I have about the replacement is Vishay SiHG22N60EF's peak (pulsed) drain current rating, which is only 58% of the original parts rating. Usually, I don't worry about the peak current rating too much, as any converter I've worked on never really approaches it. In this case, I do not know the topology, so I do not know what the current through it looks like. However, my gut says it will be okay.

Finally, I should note that I replaced every single part in the gate drive circuit. There was also a Rifa metalized paper film cap (Y2) that read OK when I pulled it out of circuit, but I noted some scorching around the base near one lead. Plus, these caps don't have a great long term track record.  I replaced it with a metalized polypropylene Y2.

[TimNJ]

Well, it has blown again after about an hour or two of use. Same MOSFETs and their gate drive resistors, although failure not as spectacular. It failed with almost no load on it. (We were running a no-load standby power test.)

Hard to say if it's something in the surrounding area that's not quite right or if the replacement transistors are not close enough. No schematic so I'm not even sure what half of the parts are for. There is a massive snubber that we hypothesized might not be doing its job...but R and C values are correct after measured.

Another hypothesis is that since the replacement transistors (both power and gate drives) do not have integrated zener clamps on the gates, whereas the originals do, that the gates may have experienced over-voltage. The latent failure seems to indicate that something was getting worn out slowly...perhaps voltage over-stress on the gate.

This might be beyond repair at this point. I've asked Chroma if they'd sell us a board. No reply yet.

[Haenk]

Isn't using the original parts the better way?

ebay has two sellers for the 2SK1517, UK & China. Even if the chinese parts (5 pcs. for 5 USD, including shipping!) are refurbished, they are probably way better than using some random replacement parts.
Several sellers available for the 2SK975, too (including USA).

Might be worth a try before spending hundreds on a new PCB.

[TimNJ]

Isn't using the original parts the better way?

ebay has two sellers for the 2SK1517, UK & China. Even if the chinese parts (5 pcs. for 5 USD, including shipping!) are refurbished, they are probably way better than using some random replacement parts.
Several sellers available for the 2SK975, too (including USA).

Might be worth a try before spending hundreds on a new PCB.

At this point, I also agree it would be wise to use the originals. As I've said, I'm not sure what the critical parameters for this particular design...It may be more highly tuned to a particular MOSFET than I thought.

Chroma quoted us ~$500 for a replacement board, which I think is fairly reasonable considering there are 4 main boards in this machine, and the machine costs $20,000 new. Usually, I'd go for a second repair attempt, but if it fails again, and we have to pull it out of the rack again, the engineering hours start to pile up. -