How to match MJL3281A?

Status
Not open for further replies.
Jump to Latest
C

Calebay

Member
Joined 2003
#1
Hello

I need to match 2 pairs of MJL3281A NPN, for my JLH 15 amps.

But i do not know how this is done. or what i need to do it.

I have a multimeter so i can mesure Ampers. and i have a +-22volt power supply. do need more then this?

Hope someone know more about this than me...

tnx

Carsten
 
#2
can i use this?

and can i use +22 and -22volt for the supply?

How do i chose the value for Rb

Tnx
 

Attachments

  • basic transistor tester.png
    basic transistor tester.png
    2 KB · Views: 459
#4
Ok.

my multimeter does not have that funktion.
i think i read i another thread that this funktions on multimters is not very accurate.

and will is be posible to mach em whitout i go indvest? since these are the only transistors i need to match.
 
#5
Multimeters are not extremely accurate but they are consistent. So, while the actual hfe might not be right on the nose, the unit to unit numbers will be consistent. If two units come up with the same hfe then they match each other.
 
#8
Same rig.

Ic = Ib * Hfe...

So.... you can measure the collector current by the voltage accross the resistor, and you're already measuring the base current.... do some division.

EDIT:

You don't really need to measure Hfe, you just need to match them. Or match Ib/Ic's as Dr Leach suggests. Since the values are so random for given transistors, amp circuits are sort of designed to take advantage of there being an Hfe but not it's actual value.
 
#10
thx for the links.

i just tryed to mesure 10. MJL21194

40.4
40.1
47.1
38.5
43.5
40.4
44.3
43.5
41.0
40.1 mA

Mesured 30 sek.
using 12v DC over E and C.
and 0.59 volt on B? is that not to high?

do this then mean i have 3 matching pairs. 40.4,40.1 and 43.5?
 
#12
Calebay said:
I need to match 2 pairs of MJL3281A NPN, for my JLH 15 amps.

Carsten
Click to expand...


I question the necessity to match upper and lower transistors in the JLH.

I have done extensive simulation, including real life transistors and "ideal" transistors and I saw no impact from beta-mismatched transistors. You can check it out in the JLH thread.

and I never matched my BJTs for my JLH1969 and they all sound as good as the next one.
 
#13
millwood said:



I question the necessity to match upper and lower transistors in the JLH.

I have done extensive simulation, including real life transistors and "ideal" transistors and I saw no impact from beta-mismatched transistors. You can check it out in the JLH thread.
Click to expand...

Carsten

I suggest you ignore millwood's comments. It is clearly shown in Table 2 of the original 1969 article that the measured distortion deteriorates significantly as the mismatch between the current gains of the output transistors increases.

Also, I have shown in the thread referred to by millwood that the maximum available output power is reduced if the output transistors are not reasonably well matched.

Matching power transistors using a multimeter hfe test facility is a waste of time. The currents used are far too low for realistic readings. You need to measure the hfe at the collector current that will be used 'in circuit', i.e. your proposed quiescent current. Rod Elliott's transistor tester project - http://sound.westhost.com/project31.htm - can be adapted (simplified) to meet your needs. Alternatively, build your amps and measure the gain of the transistors in the working environment then swap them around to achieve the best match.

Geoff
 
#14
Geoff said:


Carsten

I suggest you ignore millwood's comments.
Geoff
Click to expand...


I think what ought to be ignored are comments whose author cannot articulate and back up. Quoting the JLH article is helpful but you and JLH did not state the other conditions under which those experiments were taken.

Not to mention that your credibility was seriously shot by those mistakes you made in that very thread.

But i wouldn't be as rude to tell others to ignore your posts, however wrong I might think they are.
 
#15
matching

Hi there,

Matching of NPN with PNP in push-pull DOES reduce distortion.
For reference, check out MJL21193/MJL21194 data sheets.
The improvement is 10 fold ie an order of magnitude. (open loop)

Don't use multimeter, they are only for small signal transistors
where the test collector current is under 3mA.

Here is an easy way to match: mind you, you will need a large number of your type of BJTs.

Get a precise constant current source IC from National Semi, the LM334Z. set up the resistor needed so that it outputs 1mA precisely. You will also need a voltage source ie power supply.
Set it to the sat voltage. Something like 6V will do. Inject that
1mA into the base, mesure the collector current and the reading
in mA is the Hfe value.

You can get free LM334Z at NSC website if you are commerical.
I don't mind sending you one for free if need be.

Just one comment: The MJL3281A/MJL1302A is out dated.
They are replaced by 2SC5200/2SA1943 which are not only more
linear but also less junction capacitance. (so that the amp is
more stable) I do sell them, matched pairs as well at very
competitive prices. If you do need them, pop me a email!

God Bless

James Yung
Praise! Audio:smash:
 
#16
matching

Sorry guys! and Carsten!

I didn't read your first mail properly. I was assuming
NPN / PNP push pull. I can't remember JLH design off head.

Anyhow, the method in my last post is still valid if you want to
measure Hfe.

However, to match NPN with NPN or PNP with PNP, there is
an alternate way: Test their Vbe. Most multimeter has diode
check function which also tells you the forward voltage. So,
just pop the 2 leads across the base and emitter and read
the readings. The range is very narrow. 0.6V - 0.8V depending
on your multimeter test current.

My sincere apology for confusion.

James Yung
Praise! Audio

PS I can match 2SC5200/2SA1943 this way as well. :smash:
 
#18
I'm inclined to second, in part, Millwood's comment that transistor matching may not be all that critical. While I normally match the small TO-92 transistors that are used in the long tailed pair and in current mirrors, I don't usually bother with the others. Granting that matching all pairs would reduce distortion, the actual measured distortion has been so close to what an earlier simulation suggested as ideal that I suspect the improvement from further matching was less than the resolution my equipment could measure.

Even when I have not bothered to match current mirrors and differential pairs, the results from the amp measured as a completed unit were more than acceptable.

I am most definately NOT from the "THD+N doesn't mater" school, but feel that after a certain number of leading zeros are achieved, additional refinement is a question of design aestheic rather than a auditory benefit. Nonetheless, I fully appreciate the satisfaction that may come to a designer/builder is sqeezing out the last drop of distortion.
 
#19
millwood said:
Not to mention that your credibility was seriously shot by those mistakes you made in that very thread.
Click to expand...

I fail completely to see how my credibility was reduced during the recent exchange of views, though if it makes you happy to think it was then please feel free so to do.

I have reviewed all the posts I made in that thread during December and there were no mistakes. There may have been some statements with which you did not agree but, there again, you did not agree with Alex, Thijs and Graham (who expressed similar views to my own) on the topic of the workings of the BJT output stage.

I will leave it to Carsten, and other readers, to decide if they prefer to believe the published results of measurements taken by a highly-regarded designer or yourself.
 
#20
Geoff said:
I fail completely to see how my credibility was reduced during the recent exchange of views,
Click to expand...

I am in no way, shape or form responsible for other's failure to comprehend. Tho. I would rather not pull this thread into a debate about why your view was wrong.

But if it makes you happy to think I was then please feel free so to do.
 
Status
Not open for further replies.
Top Bottom