RMI-FC100, a single stage audio power amplifier

I am still a bit puzzled about matching output transistors.

I have a bunch of NJL0281 and NJL0302. When I try to initially just select them roughly by Hfe with DMM, I find out that PNPs are in the range of 66-95, and NPNs from 120-140. How do I even begin with results like this? Should I ignore this and go to the next step, assuming that my chinese dmm is off?

Mihai mentioned that he matches them with 100ma Ic and 20V Vce, by Vbe and Hfe. How about the rest of the parameters? Is Ic fixed with current source? How's base connected? Some more details would be highly appreciated, since I don't have a PS with current limiter and would have to build something from scratch with LM317.
 
Hah, someone please unconfuse me. What is this table showing then:

http://www.diyaudio.com/forums/soli...tage-audio-power-amplifier-8.html#post1351876

"The matching was made at 100mA Ic and 20V Vce. The following transistors was selected from a lot of 15 peers

NJL0281 [npn] NJL0302 [pnp]
-----------------------------------------------
hfe Vbe[mV] hfe Vbe[mV]
-----------------------------------------------
138 593 133 600
136 592 133 600
134 592 133 600
-----------------------------------------------
132 594 129 600
128 594 129 600
129 594 128 600"
 
the column headers are hFE (the DC current gain), Vbe ( the voltage between base and emitter when the transistor is passing current), hFE, Vbe
The numerals are slightly displaced from the headers but I can clearly see which column of data belongs under which header.

Now to the data
138 = the hFE when the transistor is tested at that Ta and that Tc which existed during the testing (both are probably changing as the test progresses.
593 = mVbe when the transistor is passing the test current specified earlier (100mA) when the Vce = 20Vce. These first two figures are for the NPN of the pair.
133 and 600 are the respective figures for the PNP of the pair.
 
unconfuse

Did you also read the last line ?
Those values were measured with the output devices mounted, aka connected to the emitter resistors !

For close hFe numbers of PNP/NPN, you'd need MJW0302-MJW0281, factory matched within 10%.
Unfortunately for you, Onsemi cancelled their production within two years after their first release. (one reason why folks as me monitor various audio fora)
10% matched NJW1302/3281 are still in active production, can be ordered at DK etc, 0302/0281 can be get at OnSemi's NOS outlet.
Neither the 0302/0281 and 1302/3281 are ThermalTraks though.

To obtain close matching hFe numbers for complementaries, you're stuck with the classic way : order a motherload of them.
(old fart trick to get close-matching Japanese devices was using a higher grade PNP than the NPN, e.g. O+Y grade or Y+P)
 
Last edited:
My point is this - posters above said:

"you can't match N with P , at least not without significant number of transistors on disposal"

Then I see this:

"The matching was made at 100mA Ic and 20V Vce. The following transistors was selected from a lot of 15 peers

NJL0281 [npn] NJL0302 [pnp]
-----------------------------------------------
hfe Vbe[mV] hfe Vbe[mV]
-----------------------------------------------
138 593 133 600
136 592 133 600
134 592 133 600"

Looks like Mihai not only matched P with N, but also from a lot of 15 peers. Unless peers means "loads of them", I am still :confused:
 
History time :

Mihai Rauta ordered a MJW sample kit from OnSemi, somewhere in '06, those are factory matched within 10%.
Then he got the Thermaltraks from OnSemi, the 15 complementary pairs he mentioned, could be those were early production samples too.(with closer hFe numbers)

You measured them with a DMM, which is pretty meaningless.
DMM measuring of hFe somewhat works for little critters, TO-92 and the likes, not for power devices.
Afair, hFe of the Thermaltraks is 75 minimum, so the 66 you measured should already have told you that the DMM routine is not OK.

A TO-264 you have to mount to a heatsink, with a Ic in the order of 100mA and up going through it.
To make it even more realistic/accurate you'd have to mount all of them to the same heatsink simultaneously, wait for thermal equilibrium, then measure.

Even when you measure their hFe properly at 100mA Ic, that Mihai managed to obtain 6 pairs within 8% (138/128 is ~1.078) from 15 pairs total doesn't imply that you will. (production hFe spread can vary from 0 to 100%)
 
Roender was brilliantly lucky or brilliantly profligate to get his pairs that closely matched.

I just paid for premium N & P quad matched sets, dearer than N matched pairs and dearer than stock pairs and the supplied items have Vgs that are 120mVgs different. What a waste of my money.

I have still to test them as pairs simultaneously to see whether the N pairs match up and whether the P pairs match up.
I am very doubtful.

If the pair matching is as bad as I suspect I will reveal who sold them.
If the matching even closely approaches Roender's results I will publicly thank the seller for his product and efforts.
 
Last edited:
Guys, I busted my ba**s setting up Hfe/Vbe measuring jig and got - nothing:) Results are pretty consistent with DMM values, meaning they only shift for about 20. I tested few NJL0281D that measured Hfe=78 in DMM, and they all turned up as Hfe=98-99, Vbe=616-617, Ib=1mA, Ic=98-99mA on a heatsink at T=23C.

I did not test p ones, but I can probably expect similar results, meaning they will be with a Hfe at least 20 higher than highest npn.

So, back to beginning. I have 20 NJL0281 with Hfe from 66-90, and 20 NJL0302 with Hfe from 120-140 (DMM values). Is this batch usable at all in any combination?
 
As I remember Andrew you bought lots of LSK170 as FETs for this amp?
Yes, lot's of lsk and now some 2sk and 2sj.
After many long days matching them I made them available to UK Members (easy posting and cheap posting).
But UK Members were hardly interested. My orders came from Europe and the North Americas.
Shame that the vast majority of these carefully matched pairs got wasted in B1 and DCB1, but who am I to argue, the customer is always right.
A few will remember that I tried to dissuade them, some times successfully to not waste their money on matched pairs. B1 and DCB1 only requires selected Idss devices. Matched over a range of Id is a total waste.

For this project matched pairs are very useful.
Let's see if any wake up to the difference between matched pairs and the usual offerings on this Forum that are miss-described as matched pairs when all they are, are selected Idss pairs or quads or whatever.
 
Hi AndrewT Do you have any matched pairs left for this amp (2sk170bl) . regards john


Yes, lot's of lsk and now some 2sk and 2sj.
After many long days matching them I made them available to UK Members (easy posting and cheap posting).
But UK Members were hardly interested. My orders came from Europe and the North Americas.
Shame that the vast majority of these carefully matched pairs got wasted in B1 and DCB1, but who am I to argue, the customer is always right.
A few will remember that I tried to dissuade them, some times successfully to not waste their money on matched pairs. B1 and DCB1 only requires selected Idss devices. Matched over a range of Id is a total waste.

For this project matched pairs are very useful.
Let's see if any wake up to the difference between matched pairs and the usual offerings on this Forum that are miss-described as matched pairs when all they are, are selected Idss pairs or quads or whatever.