2SC3153 high voltage NPN switcher alternative

[Audiorepair]

Trying to find an easily available alternative to 2SC3153 used in a SMPS.

NPN
Vcb  900v
Vce  800v
6A,  100W
fast switching times

Seems to be more difficult than I thought it would be.


https://datasheetspdf.com/pdf-file/638800/SavantIC/2SC3153/1





 

[T3sl4co1l]

Not really *fast*, but shouldn't be hard to find something like that.  MJE18006 comes to mind, though Vceo is lower and package is TO-220.  Would have to go looking for a TO-3P or TO-247 but I'm sure something is out there.

Tim

[Audiorepair]

Not really *fast*, but shouldn't be hard to find something like that.  MJE18006 comes to mind, though Vceo is lower and package is TO-220.  Would have to go looking for a TO-3P or TO-247 but I'm sure something is out there.

Tim

That's what I thought.

[Audiorepair]

Best I could find available is 2SC5354, but this is only 5A/8A as opposed to 6A/20A for the 2SC3153.

https://www.mouser.co.uk/datasheet/2/408/2SC5354_datasheet_en_20131101-1101952.pdf


I did find this FJL6920, but it is designed for a colour monitor, and I do not have the expertise to know whether this would be ok in this SMPS switching circuit.

https://www.onsemi.com/pdf/datasheet/fjl6920-d.pdf

[T3sl4co1l]

Huh.  Well, I'd be willing to bet the current doesn't need to be so high; looks like about a 50W converter?  So a peak current of a few amperes at best.  The 20A peak rating is absolutely not required, and a 3-5A transistor would probably do.

Allowing reduced Vceo is arguable; it should be okay since the base drive is strong and Vcbo will apply.  Better to be safe than sorry?  If you can afford potentially blowing out another transistor, I'd consider that a "test it out first" option.

Beware that faster switching speeds will increase EMI.  They have good enough snubbing around there that it shouldn't go crazy, but maybe also something to check.  Likewise, try to look for similar capacitance (which, ugh, that Toshiba part doesn't even say.. thanks Toshiba).

Absolutely, FJL6920 is too big for this (physically and electrically).  Also quite a bit faster, though capacitance might dominate in that case.  (Which again isn't specified, ugh.)

Tim

[Audiorepair]

I've powered the unit up on bench supply and it takes 2.4A at 5v, and way less than an amp for the 15 volt rails supplying the audio circuits as expected.

It seems Yamaha went for the biggest baddest device out there at the time.


I'm just very wary that they did this to save having to deal with any design issues, and that derating the wrong parameter could end in tears.


This is in a vintage synthesiser where catastrophic failure is not an option.

And it's not mine.


Thanks.

[floobydust]

I looks like a cheapola two-transistor self-resonant SMPS, so liable to be finicky about the part. I think h_FE is also important yet the part is grossly oversized, I'd guess to handle start-up.
If you're limited to Mouser UK, they only have one TO-3P part I'd consider suitable 2SC5354 h_FE 15-60 at 0.5A
Otherwise you need to peruse another disti or go TO-220 which is what now dominates these kind of specs. The FJP 1,700V part is really low gain and I would not use it.
Also critical to operation is C14 1uF 50V electrolytic, I would replace it (edit: with a film cap). If the part goes low value, regulation can fail.
I might consider a dummy load or crowbar on the secondary DC rail for protection during initial power up, to see if it's regulating properly.

[T3sl4co1l]

Also consider disconnecting the output (cut traces if necessary..?) and adding a dummy load.  The hazard is then limited to PSU components, or maybe at worst, fusing some traces (but the fuse should be rated enough to avoid that already).

These kinds of circuits don't blow up the output; if switching stalls, there's simply no flyback and thus power output.  Which is true even if the device fails shorted, dumping the filter caps + mains fault current into the transformer.

There can be a runaway effect when the feedback (drive) voltage exceeds Vebo, returning bias current to the base circuit that's otherwise meant to be managed by the feedback circuit, hastening turn-on for the next pulse; the maximum output is still limited by transistor rating and transformer saturation, so a shunt regulator or crowbar can handle this just fine.  (In particular, a crowbar also (nearly) shorts out the transformer, forcing it back into a startup condition.)  Anyway, Vebo is usually pretty generous (have seen/measured 10-12V on HV BJTs before I think?) and never less than, well, whatever it says of course (7V in present cases).  So, if it works in the first place, this effect won't show up with whatever else; that is to say, it's a design issue more than a component selection issue.

Tim

[Audiorepair]

Fortunately the SMPS board is separate.

I connected a +/- 20v to the SMPS rails before the 15v regulators, and a 5v supply direct to the 5v output, thus testing the output section of the SMPS was OK.
The output section also has an oscillator circuit on the 5v rail to power the LCD backlight, this was also singing like a canary.

Once connected to the synth, it powered up, and the LCD was working and the synth complaining it had lost its memory.



So now I have just the primary side to try and get working.


Ah, and this shows the half of the Opto coupler fed back to the primary side to regulate.

[floobydust]

The opto-feedback is looking at both the 5V rail and -ve rail (R19) input to 7915.
I would just put a several amp load on the 5V rail and see what happens. Mains input spec is 30W and it sorta looks like Yamaha had no idea how much power the digital section would need.

Caution if you run the PSU with the main board disconnected, the EL backlight inverter Q4/T2 would see no load, and likely would die it's not good for it. Maybe lift a leg on L9 or Q4 to disable it.
Afterwards, make sure all PSU caps are gently discharged before ever plugging that in to the synth, or you will damage things due to stray charge.

[Audiorepair]

Thanks for all the info, it's made my position much clearer.

Bit of a puzzle on how to best proceed.
If it wasn't for the LCD backlight it might just be easier to replace with a linear supply, which would be foolproof.


And perhaps also replace EL LCD with OLED if possible.

[T3sl4co1l]

Or an off the shelf (switching or otherwise) PSU, assuming one can be found which fits.  I think +/-15V still isn't uncommon among supplies, or perhaps using a DC-DC from 5V (or two from say 12V, for +/-15 and 5).  Cut off the backlight circuit and go.

Or replace it maybe, if a suitable part can be found -- pretty long shot, granted!  Which on that note, isn't EL notorious for aging quickly -- is it quite faded?  Maybe a replacement would indeed be worth the work.

Tim