New transistor source

I have tested successfully low Cob KTA1381Y (hFE 148-180) and KTC3503Y (hFE 115-118), free samples from KEC.
Jacques Antoine, if I may ask: how did you get those? When I click on "Request samples" for any of them, it takes me to a generic contacts page... Did you email sales, tech support, a European distributor, other? As I have made pretty accurate Spice models of some of the originals (see here), I'd love to do the same with these equivalents so we can compare. Many thanks in advance!
 
Anyone know of a good source for matched, small-signal PNP transistors? ADI seems to have discontinued all their offerings. I tried twice to get some from Linear Integrated Systems, but found them to have terrible output resistance, around 10K Ohms, making them unsuitable for what I was trying to use them for, current mirror loading of a balanced stage.
 
One way to get matched pairs is to DIY (link to recent Forum thread). In my experience, you get 5-6 matched pairs per hour of work.

You might consider the Diodes Inc / Zetex part numbers ZTX757 and ZTX957 , PNP bipolar transistors. The '757 have higher measured output resistance but lower measured beta (90 amps per amp). The '957 have not quite as high measured output resistance, but higher measured beta (180 amps per amp)

edit- I also should add that a much easier way to get 10X higher output resistance, is to install matched emitter degeneration resistors in both halves of the current mirror. If you select these resistors so their voltage drop (Iemitter * Rdegen) is about 12 * (kT/q) ..... then output resistance goes up 10 fold. And, oh by the way, sensitivity to transistor mismatch falls by about 10x too. If you previously needed 1mV matching, with degeneration resistors you now need ~ 10mV matching. At room temperature, which is T=300K, that voltage drop works out to about 312 millivolts across each degeneration resistor (or so). Check it with circuit simulation, don't take my word for it.

_
 

Attachments

  • PCB_topview.png
    PCB_topview.png
    78.9 KB · Views: 143
  • operation_2.jpg
    operation_2.jpg
    342.1 KB · Views: 136
  • ZTX757.PNG
    ZTX757.PNG
    39.8 KB · Views: 137
  • ZTX957.PNG
    ZTX957.PNG
    53 KB · Views: 119
Last edited:
One way to get matched pairs is to DIY (link to recent Forum thread). In my experience, you get 5-6 matched pairs per hour of work.

You might consider the Diodes Inc / Zetex part numbers ZTX757 and ZTX957 , PNP bipolar transistors. The '757 have higher measured output resistance but lower measured beta (90 amps per amp). The '957 have not quite as high measured output resistance, but higher measured beta (180 amps per amp)

edit- I also should add that a much easier way to get 10X higher output resistance, is to install matched emitter degeneration resistors in both halves of the current mirror. If you select these resistors so their voltage drop (Iemitter * Rdegen) is about 12 * (kT/q) ..... then output resistance goes up 10 fold. And, oh by the way, sensitivity to transistor mismatch falls by about 10x too. If you previously needed 1mV matching, with degeneration resistors you now need ~ 10mV matching. At room temperature, which is T=300K, that voltage drop works out to about 312 millivolts across each degeneration resistor (or so). Check it with circuit simulation, don't take my word for it.

_
Thanks for the reply.
Have you been able to match the performance of like an SSM2222(IIRC the part number for the ADI dual PNP) with hand matching? I've done this before but have always felt this to be not as good as the monolithic solution due to the great thermo-coupling one gets. Even if I grease two discrete transistors together it's not as good as two transistors right next to each other on the same piece of silicon.
And yes, there's always the degeneration option. And yeah, Ro is better, matching is better, and noise is typically better. But this comes at the cost of gain and voltage. In a current mirror the gain doesn't matter, but I still have to feed the next PNP stage. At the levels you're talking about(~300Ω) and a 1mA bias, there may still be enough voltage to drive the next stage without degenerating it. Something to think about.
There may also be some fancy design tricks I don't know about how to get around this.
 
DIY transistor matching is only performed by people who are both highly motivated and desperate. If a usable dual transistor is available which approximately meets the requirements, most people will pay the (high) price and delight in the fact that they need not measure anything; the manufacturer has already done that. Using validated test fixtures, calibrated test equipment, and verified test procedures carried out by professionals.

Only when a dual is not available, or when it's likely that DIY matched individual parts will outperform the available duals, will builders bother with DIY matching. When you do this, the burden of measuring and guaranteeing performance specifications, falls upon you the builder. "Sure, I matched for parameter X, but how good a match did I happen to also get on parameter Y?" -- nobody else can satisfy your curiosity, you can only measure it yourself.

Read through the Burning Amp "BJT Simple Matcher" presentation .pdf, it's only ~ 15 slides, and ponder whether you are motivated enough and desperate enough to try it yourself.
 
  • Like
Reactions: chermann
Inventory update for Profusion for those needing transistors for audio amplifiers. The KTA1381 / KTC3503 are the best VAS overall for new projects, however ,other types listed here might be a closer match when servicing some amplifiers.


KTA1381 / KTC3503 300V FAST VAS STAGE TO126 Replacement for the EOL ON devices, really fast high voltage VAS
These sold out on Pre orders, much larger inventory on the way however.

KTA1837 / KTC4793 230V VAS / DRIVER TO220 Replacement for toshiba 2SA1837 / 2SC4793 , good VAS and driver , in stock.

KTA1700 / KTC2800 160V VAS / DRIVER TO126 Replacement for 2SA1220A / 2SC2690A , GOOD VAS for small amplifiers, and driver. In stock.

KTB2640G / KTD1640G 150V 130W FAST DARLINGTON OUTPUT TO3P N Replacement for Sanken 2SB1647 / 2SD2560, improved SOA easily attains 100W into 4 ohms with +/- 45 V PSU and proper design. In stock.
This is a really nice output device for very simple hifi amps.

KTA1943A / KTC5200A 230V 150W WIDE SOA FAST OUTPUT TO3P L (TO264) This is an enhanced version of Toshiba devices, almost identical to 2SA1302 / 2SC3281 in every way , but higher voltage rating. In stock.
 
I have these for the original 2SD1047 / 2SB817. Based on measurements of one pair in my Rotel RA-820AX with hFE = 57 / 67, respectively. Note that the DCA75 analyzer I have only goes to ~10 mA, but I also matched to the datasheets beta droop, fT, Cob, Re, temperature dependency and quasi-saturation:

.model 2SD1047 NPN (
+IS = 4.31p NF = 1.0 BF = 100 VAF = 600
+ISE = 170p NE = 1.7 IKF = 10 NK = 0.64
+RC = 10m RE = 90m RB = 6.0 RBM = 0.6 IRB = 1.45m
+ISC = 1.5n NR = 1.0 BR = 8.6 VAR = 50 NC = 2.0
+CJC = 810p VJC = 0.5 MJC = 0.41 TF = 10.8n VTF = 10
+CJE = 100p VJE = 0.5 MJE = 0.5 ITF = 50 XTF = 12.8
+RCO = 0.5 VO = 35 GAMMA = 10n QUASIMOD = 1
+EG = 1.1 TR = 300n XTB = 0.75 )

.model 2SB817 PNP (
+IS = 4.7p NF = 1.0 BF = 109 VAF = 120
+ISE = 28p NE = 1.5 IKF = 40 NK = 0.95
+RC = 10m RE = 0.12 RB = 6.0 RBM = 0.6 IRB = 960u
+ISC = 180p NR = 1.0 BR = 14.8 VAR = 30 NC = 1.5
+CJC = 1.1n VJC = 0.5 MJC = 0.41 TF = 9.4n VTF = 10
+CJE = 100p VJE = 0.711 MJE = 0.304 ITF = 50 XTF = 21.5
+RCO = 0.5 VO = 25 GAMMA = 10n QUASIMOD = 1
+EG = 1.02 TR = 300n XTB = 1.22 )
 
  • Like
Reactions: minek123
I purchased 60 each of the KEC 1381/3503 and 20 each 1837/4793and KTB1369-Y/KTD2061-Y I could not resist testing them to see how well their HFE range matched and i can say they measured well within their spec and numbers part to part was closer than most of the parts I have bought over the years from Many sources. The 1837/4793 was not marked with Y hfe designation but they all measured within that spec the lowest was 150 highest was 170 across PNP/NPN the Koreans seem to have their process control very well structured and it shows in the Parts that i purchased. This is 5th time i have bought parts from Profusion and I would not hesitate to say they are Legit place to buy parts and they ship fast I used the 9 dollar international shipping and had my parts in a week from payment not bad at all it sure beats paying 35 dollars for 3 day international shipping. The parts that surprised me was KTB1369/KTD2061 they measured in 150 range for a 2 amp driver with specs similar to sanken 2sc4883/2SA1859 these look perfect when you want a single driver for 3 or more pairs of output devices and fills in the blank space when you want a faster driver than the MJE150XX from Onsemi. Anyone on the fence because they are worried about quality from Profusion i would in my experience say that they are a Trusted Source