Discrete Opamp Open Design

... There are good dual bipolars still available but fewer.

Yup, most of the inexpensive duals from Rohm, NXP, Infineon, FSC, etc. are either General Purpose or Digital (with integrated resistors). There are very few, if any, low-noise BJT isolated duals with 2x 120V, 0.05A devices, similar to some of the old through-hole Sanyo devices. It's not a huge problem yet because I can always use 2x SOT323 for minimizing board area.

Are there any recommendations for SOT323 JFETs, apart from the Toshiba 2sk880 and 2sk1875?
 
I just noticed that Mouser carries the Toshiba 2sc4207-BL in SOT23-5 or similar. This is an emitter-common NPN pair which looks to be 2x 2sc2712, 2sc4116 or 2sc1815 in the same package. It's a GP pair, but known to be highly linear and with excellent DC beta. The common emitter connection limits its usability a bit, since it rules out external degeneration resistors. Still, it's pretty competitively priced in volume and usable in some LTP applications.
 
Mr. Marsh, in a 'perfect' BJT, Env is inversely proportional to collector current so Yes. It will change ALOT.

This is what JC was referring to in #3100.

Looking to see how how a real device departs from this behaviour gives us rbb' and other stuff that helps us to design good LN amps.
Duu.uuh! That should be

in a 'perfect' BJT, Env is inversely proportional to square root of collector current

JC clarifies this further in #3114

RNMarsh said:
I havent found any perfect transistor behaviour, yet.
The specified Hitachi, Rohm & Toshiba operating from 100uA - 3mA are near perfect .. at least for noise.
__________________

Mr. Marsh, thanks for the measurements of BC377.

0.6nV/rtHz is Rnv 21R6.

At 3mA, a perfect transistor has Rnv 4R2 as per JC in #3114.

So a good value for BC377 rbb' is 21.6-4.2 = 17R4 .. rather better than Toshiba 2sa970 but not quite as good as the supa Hitachi, Rohm & Toshibas.

Have you got the PNP complement to measure? Which maker? I can only find a Semelab datasheet and it doesn't have much (any?) info.
 
Disabled Account
Joined 2012
yes, i have the compliment.... used the NPN for worse case.
Give it to you if you like to know the numbers.

Also, give me the numbers for those better supa-dupa bjt's referred to and I'll test them also.


THx-RNMarsh

typo Correction -- thats BC337
 
Last edited:
OK Here are your numbers for the PNP type bjt - BC327:
At 10/100/1KHz/10KHz/100KHz

Measures -- 5nV/.7nV/.5nV/.55nV/.55nV

Same 3mA/15Vdc
Beta = 407 [using PEAK atlas DCA Pro.... 5mA]
Thanks Mr. Marsh.

Which manufacturer made these BC337 & BC327?

The supa LN devices are

NPN Rohm 2sd786, Toshiba 2sc3329, Hitachi 2sc2545/6/7
PNP Rohm 2sb737, 2sa1316, 2sa1084/5/6

Small stocks of the NPN devices are still available for sale but the PNP devices are long Unobtainium :mad:

1mA & 3mA are good currents to measure their performance. The Beta is useful too.
 
Last edited:
Disabled Account
Joined 2012
Thanks for BC3x7 measurements! What is the corresponding Rb for these numbers?

see up above --- #3144

Anything this low in noise is good enough for me... I never plan to buy an LP as long as I live. And, I have not bought one in probably 20 years. So MC pre-pre etc is of no interest to me.
but, it may have other secondary benefits to the sound worth exploring.

-RM
 
Last edited:
I updated my BC3x7 quasi-saturation models with the appropriate values:

.model BC337-40_kq NPN(
* Updated Rb 8-20-2014 by keantoken
+ IS=7.809E-14 Ibc=100f NF=0.9916 ISE=15f NE=1.4 Nk=0.55
+ BF=436.8 IKF=1.15 VAF=103.6 NR=0.991 ISC=6.66E-14 NC=1.2
+ BR=44.14 IKR=0.09 VAR=14 RB=10.8 IRB=2.00E-04 RBM=8
+ RE=0.12 RC=0.05 Rco=3 XTB=0 EG=1.11 XTI=3 CJE=80p
+ VJE=0.6657 MJE=0.3596 TF=620p XTF=1.5 VTF=2 ITF=0.5 PTF=88
+ CJC=1.306E-11 VJC=0.3647 MJC=0.3658 XCJC=0.455 TR=2.50E-08
+ CJS=0 VJS=0.75 MJS=0.333 FC=0.843 Vo=500 Gamma=4n Qco=8n
+ Vceo=45 Icrating=800m mfg=OnSemi)
.model BC327-40_kq PNP(
* Updated Rb 8-20-2014 by keantoken
+ IS=2.077E-13 Ibc=230f NF=1.005 ISE=1.4f NE=1.3 NK=0.5
+ BF=475 IKF=.65 VAF=29 NR=1.002 ISC=2.963E-13 NC=1.25
+ BR=20.92 IKR=0.104 VAR=10 RB=17.4 IRB=1.00E-05 RBM=5.3
+ RE=0.14 RC=0.32 Rco=3.75 XTB=0 EG=1.11 XTI=3 CJE=115p
+ VJE=0.9296 MJE=0.456 TF=500p XTF=3.25 VTF=2.5 ITF=0.79
+ PTF=80 CJC=2.675E-11 VJC=0.8956 MJC=0.4638 XCJC=0.459
+ TR=3.50E-08 CJS=0 VJS=0.75 MJS=0.333 FC=0.935 Vo=500
+ Gamma=10n Qco=8n Vceo=45 Icrating=800m mfg=OnSemi)
 
I thought that we were designing 'optimum' op amps here. Kgrlee is right on top with his part numbers. These parts have been available since the late '70's. The really important thing, in general, is that these special Japanese parts have low Rbb' (as low as 2-4 ohms respectively) yet have high beta, and EVEN high voltage breakdown. They ran Motorola right off the page! Now, of course, most of these parts are discontinued, and that makes it problematic for amateurs to get them.
Personally, I prefer low noise jfets, because they can be almost as quiet (ideally), easily paralleled, and have low input current. They are really ideal for an 'optimum' op amp input device.
 
OK Here are your numbers for the PNP type bjt - BC327:
At 10/100/1KHz/10KHz/100KHz

Measures -- 5nV/.7nV/.5nV/.55nV/.55nV

Same 3mA/15Vdc
Beta = 407 [using PEAK atlas DCA Pro.... 5mA]


THx-RNMarsh

Mr. Marsh, thank you.
Yes, those old geezers (BC327/337) are very decent performers even per today's standards.
Your work have inspired me to start measuring transistors in my parts bin. Found another decent performers, Zetex ZTX651/751.
 
Disabled Account
Joined 2012
Thanks Mr. Marsh.

Which manufacturer made these BC337 & BC327?

The supa LN devices are

NPN Rohm 2sd786, Toshiba 2sc3329, Hitachi 2sc2545/6/7
PNP Rohm 2sb737, 2sa1316, 2sa1084/5/6

Small stocks of the NPN devices are still available for sale but the PNP devices are long Unobtainium :mad:

1mA & 3mA are good currents to measure their performance. The Beta is useful too.

I dont have any of these. And a quick check with digikey and mouser came up with a big zero there. If I can get them, I'll measurement and give results here. Or you (or anyone) can send me one or two-- I'll return them. But dont send any that are not made any more.

THx-RNMarsh
 
Disabled Account
Joined 2012
I thought that we were designing 'optimum' op amps here.

Originally this came about by a challenge to make the best circuit with the fewest transistors.... well under .001% and drive low Z and use 8 transistors or less.

Scott took the challenge and produced a very good circuit but I never knew anyone who tried to make and measure it. Used quit a few transistors though.

THx-RNMarsh
 
I updated my BC3x7 quasi-saturation models with the appropriate values:
BC337 should have RB=17.4

Though I haven't done the sums, I think the PNP BC327 should be RB=10.8

Thanks for these models kean. Let us know how you get on with LTspice to model 1/f noise. If you can just model a 6dB/8ve noise rise, that would be good.

I looked at ZTX651/751 in the early 80's and so did Baxandall for QUAD. You had to do a lot of screening for 1/f noise and the rbb' was still some way from the Hitachi et al. Dunno if modern production is better. I have some correspondence with GG Baxandall on this from that period.
 
Last edited: