Gain split in 6n139 opto-coupler ??

Thread Starter

Nik

Joined May 20, 2006
55
6n139_2238i0.jpg Thumbnail of internal circuit.
http://www.avagotech.com/assets/downloadDocument.do?id=1005
Link to big PDF datasheet.

Test Disclaimer: I am a hobbyist, not a student.

Hi, the 6n139 split Darlington opto-isolator has a photo-diode directly connected between Q1 Base and Q1 Collector. Although Q1 Emitter internally connects to Q2 Base, Q2 is fully pinned EBC to allow gain/bandwidth trade-off.

Different suppliers quote overall current transfer ratio (CTR) of 400~~5000 %. Allowing diode-diode CTR of ~0.2%, this makes Q1xQ2 gain 2000~~25000.

Now, is this gain evenly split ?? Not 'matched', surely, but +/- 30% perhaps, not ratioed 90/10, 80/20 or 70/30 ??

Though detector diode, Q1 & Q2 are probably laid down in same process, I don't know what to think !!

I'm not trying to 'design to gain', I'm trying to set workable limits on a home-brew S/Spice model...
 

Thread Starter

Nik

Joined May 20, 2006
55
Well, I posted a query on LTSpice group, and can't get to it since it scrolled off the 'recent' page. Seems my ISP and YahooGroups don't get along, and the available link is *very* slow, sub-dial-up...

Moans aside, an apology:

OF COURSE the Darlington's gains will be different !! Makes sense to have Q1 Lo_noise + Hi_gain, while Q2 is a lower-gain 'Driver'.

Okay, the question stays the same: how are their gains etc probably ratioed ??
 

Papabravo

Joined Feb 24, 2006
21,157
Too bad abour your ISP. I think I can say without fear of contradiction that Helmut and the Yahoo Group was your best shot. I just checked the group and there are no replies.
 

Thread Starter

Nik

Joined May 20, 2006
55
Thanks for checking !!

Okay, time for Plan B: I'll buy a bunch and check them out. Should get a trend from 20~~25, I suppose...

D'uh, been a long, long time since I measured beta or Hfe beyond a 'smoked' test, so should be fun !! I'll spreadsheet my findings, report them 'for info' when I can.

Hmm, must remember to get ZIFs, too...

Thanks again,

Nik
 

hgmjr

Joined Jan 28, 2005
9,027
Greetings Nik,

I was going to suggest that you try the empirical approach but you have leapt to the conclusion on you own.

Papabravo's suggestion was the way to go but since you have encountered a lack of response thus far from the LTSpice group, your plan B seems to be reasonable.

I am not sure if you are going to be able to get directly at the beta of the first transistor. Unfortunately, you only have direct access to the first transistor's emitter and the combination of its base (with series photodetection diode) and collector. You will no doubt be able to infer its beta since you will be able to establish the beta of the second transistor in the pair.

I believe it should be relatively straightforward to measure the Current Transfer Ratio (CTR) associated with the first transistor (the one with the photodetector diode in its base circuit). That together with the beta of the second device may be all you need.

Good Luck,
hgmjr
 

Thread Starter

Nik

Joined May 20, 2006
55
Hi, thanks for the useful advice !

I reckon I can start with nominal ~~0.2% CTR for the Diode/Diode stage, measure (D+Q1) CTR, estimate Q1 gain, measure Q2 gain directly...

Then comes a sanity-check, akin to 'Saturated ?' for bipolar bias. If I get wild, super-beta results for Q1, perhaps that nominal D/D CTR was too low...

Of course, that D/D CTR embraces both LED and optical efficiencies.

Among other sites, my Googling took me to an arcane NASA report on radiation stability of optocouplers. Seems 'modern' LEDs are much more robust and predictable than older versions, and diode detector couplers are far less vulnerable to degradation than 'big base' opto-tranny designs...

May take me a while to get enough parts and data, but the numbers should be interesting !!
 
Top