"modern" STK Darlington power packs, idle/bias problems

I've ordered STK0080IIs from three different suppliers over the last year or so, and all of them fail to produce any idle current (when installed in a Yamaha CR-1040), and thus have bad crossover distortion. Now, I know these aren't NOS Sanyo parts, but I 'm not sure that they are intentionally sub-standard counterfeits, either.

When I finally got around to busting one open, what I see doesn't look entirely unreasonable: two stout-looking power devices with metal heat spreaders (atop the metal overall backing), and the rest done up with very ordinary-looking SMD transistors and resistors. The idle bias circuit is a single transistor (two-stage Darlington, apparently) with a fixed resistor network. If I jumper a 5K pot from base to emitter (paralleling the existing 220 ohm), I can easily set idle current as desired.

So, I'm wondering:

Has anyone gotten STK0080IIs recently that DO idle correctly?

Does anyone have any particular insight into why the manufacturer of these SMD-based parts would produce parts that don't idle? Are they just clueless about what's required of an idle bias circuit?

If you've gone ahead and installed non-idling parts into equipment, have they been reliable otherwise? (I'm concerned by how small the driver stage transistors are, though I reckon they get good heat extraction.)

Assuming no properly working parts are available, I'm thinking about using the SMD-based ones, but hole-sawing a port into the plastic cover for installation and adjustment of an idle-setting pot. I think this will be relatively practical. I'm not thrilled about making a hole in the package, but just because of esthetics - I don't think there is any operational need for the package to be sealed. Any thoughts?

Thanks,

chazix

I don't think there's any need for the package to be sealed either; the pot idea should be fine. Just leave the heat spreader alone of course.

They could indeed by counterfeit parts. STK chips are notorious for this...irregardless of the branding.

If the Yammie has no external provisions for setting bias, then yeah the chip should indeed have a fixed bias resistor or a pot as you mention. Something is amiss here.

The CR-1040 only has a setting for DC offset AFAIK, at least that is what you are measuring across TPL to ground and TPR to ground.

chazix said:

Any thoughts?

Click to expand...

You go boy!

I just reach into the oscilloscope and grab the waveform at the top and bottom. I stretch it out, as if it's made of rubber. The crossover distortion always goes away.

Fred

> I just reach into the oscilloscope and grab the waveform...

Yabut, wouldn't it be simpler to just listen only to square waves? Now, where'd I put those Vangelis recordings...

I guess I'll go ahead and make it a "someday" project to acquire some SMD pots for retro-fitting an idle adjustment into these packs. Will report back (eventually) on how it goes.

Wonder how the original Sanyos had their idle current determined? I think they were "thick film" construction, which I believe can have laser-trimmed resistors - but I'm not sure that would have been practical in the consumer electronics parts market.

chazix

...and the verdict is...

Guilty on all counts - the STKs that won't idle are apparently also unable to survive at full load. I reckon they are deliberate counterfeits.

When I tacked in a bias adjustment pot to one of these ersatz STKs and installed it in a CR1040, I could get it to idle, though the adjustment seemed overly sensitive to both power supply voltage and temperature. I decided to ignore that for the moment and press on with some power testing. The part lasted for maybe two minutes of operation at the onset of clipping into 8 ohms. I suspect the little SMD driver transistors don't have anywhere near the SOA they need. It's also possible that the poor temperature compensation allowed the part to get into thermal runaway.

chazix

Not guilty, but audacious.

I looked at the STK0080II schematic...

If the current in the diode (D2 on schematic), in series with the 220 Ohm (R3 on schematic)//your trim pot, exceeded his maximum rating, then suddenly the bias will jump beyound reasonable limits.

This was a $600/hour experiment. Sorry for you!

The cost of the parts was already "sunk money", and I consider the parts unfit for real-life use even if their only fault is failure to make idle current, so the experiments are free, sort of.

I'll double-check when I have the schematic and one of the suspect parts in front of me, but I'm pretty sure the diode you refer to isn't actually in the suspect parts - another way the manufacturer "cheaped out". But I don't know of any reason why there would be enough current in that part of the circuit to blow a diode, whether or not the diode is present (and also whether or not my added trimpot is present).

Thanks for giving it some thought!

chazix

Interesting. Would you care to share where you bought these parts?

Where to buy from?

I don't want to hijack this thread, but I would like to know where to get good parts. I recently purchased (From B&D) all required tranny's to rebuild the amplifier boards in my Sansui Eight's. I was sent "ST" parts manufactured in Malaysia, and all others just say Korea on them along with the part number. My guess is I should just junk them and chalk it up to a lesson learned. The bigger question still remains, can good parts be found? Anybody have luck with Fairchild parts? Perhaps it's time I just make the effort and match all needed parts to On-Semi's? The original transistors, if interested are:
C735
A562
2SC708A
2SA537A
2SC627
A640

Thanks in advance for any input.
Bill

chazix said:

It's also possible that the poor temperature compensation allowed the part to get into thermal runaway.

chazix

Click to expand...

Did you put the cover back for your test?

FWIW I have never received a fake version of any part from B&D.

> Would you care to share where you bought these parts?

I remember for sure that CEI and B&D were two of the places I've gotten bad STK0080IIs - not sure which was the other supplier. I hasten to mention that I don't put much blame on these shops - I'm sure they are dealing in good faith.

chazix

> FWIW I have never received a fake version of any part from B&D.

Yup, it was Merrylander's and others' recommendations of B&D that got me to make one more try after I aleady suspected that bad parts were everywhere. Merrylander, do you recall when was the last time you got a good STK0080II in particular? My B&D order was within the last few months.

Oh - I just found the other place I've gotten a bad 0080II - Audio Lab. Once again, I don't want to discourage anyone from using this source (other than for this exact part), and I believe them to be straight shooters.

chazix

The last time I did a CR-840 or 1040 was close to a year ago. Audio Lab never let me down either, if you are certain they were fakes I am sure both sites would appreciate a heads-up.

chazix said:

... but I'm pretty sure the diode you refer to isn't actually in the suspect parts - another way the manufacturer "cheaped out". But I don't know of any reason why there would be enough current in that part of the circuit to blow a diode, whether or not the diode is present (and also whether or not my added trimpot is present).

chazix

Click to expand...

I don't know if you looked at this datasheet for the STK0080ii

http://www.datasheetarchive.com/dow...m/pdf-datasheets/Datasheets-10/DSA-183473.pdf

The bias stage is NOT a Vbe multiplier. In fact D2 and R2 are important parts of this bias stage. A current flows through D2, R2 and R3. When the voltage developed across R3 is approximately 1.2V, Tr1 and Tr2 start to conduct and divert the current from pin 0. Tr1 and Tr2 are acting as a kind of limiter in this circuit. The bias voltage from collector to emitter of Tr2 is given by the total base-emitter voltage drop of the darlington connected Tr1 and Tr2, plus the diode voltage in D2 and the voltage across R2.

When you shunted R3 with your trim pot, more current was flowing through D2 and R2. There was more voltage across R2 and more bias voltage to the driver transistors.

Is it possible that the equivalent resistance of (R3 // your trim pot) was so low that the current in D2 and R2 was too high for D2 or R2? I don't know.

It is possible that D2, Tr1 and Tr2 were too cold vs. the power stage, if the cover was not put back in place. In this situation, the thermal tracking was bad.

Hmm, good/bad news: I think I have to retract the claim that I got a bad part from B&D - I can't find any record of it now, which makes me think I mixed them up with another supplier. (I double-checked CEI and AudioLab, no mixups there.)

The good-news part is that there's still hope of getting good parts from a mainstream supplier. 'll take one more stab at this with a part ordered from B&D.

The bad-news part is how 'shamed of myself I am for the misinformation, but I hope this will fix it. Will report back on the B&D-sourced part.

Oh, and I'll also let CEI and AudioLab know that I believe they got stuck with counterfeits.

chazix

> I don't know if you looked at this datasheet for the STK0080ii

Hi, ecluser,

Yes, I've seen that, and I think I understand your points. Everything you say is logical, but I am still inclined to think that SOA violation (probably TR4/TR7) was the true cause of the part failure.

I've still got at least one more part that I believe is counterfeit, but that I have not tampered with. I might as well install that part and repeat the full-power testing. If the un-tampered-with part survives, it would be evidence for your theories about what went wrong earlier - and then maybe I can pursue a better way to modify the part so that it idles properly. If the part fails, I'll stick with the SOA theory, and not spend any more time trying to make use of suspect parts.

chazix

They find their profit conterfeiting 25¢ transistors, imagine an expensive power pack...

You are in a better position than me to judge the SOA of your STK

Is there some components still good in your guinea-pig? This may help to find the cause of failure.