Transistor HFE observations.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Now that i have my M3 transistor analyzer kit built. i decided to run through some transistors and see what i get.

On hand i have several sets of MJ15024/25 transistors and i was curious to see what i would get.

Set #1 is whats left from the Classe DR-15 amp i am rebuilding. these should have been matched at the factory.

Set #2 is a set of 5 i recieved as samples from OnSemi, these are from the same lot, but not matched in any way.

Set #3 is the new set of 4 from Classe that is supposed to be factory matched.

The results

Set #1
NPN - Hfe- 26, 28, 31
PNP -Hfe- 54, 55, 60

Set #2
NPN - Hfe- 32, 33, 35, 35, 36,
PNP - Hfe- 71, 72, 75, 76, 81

Set #3
NPN - Hfe- 7, 8, 8, 12
PNP - Hfe- 89, 89, 90, 93

Very interesting numbers!

The NPN's of set 3 i checked twice to make sure the numbers were right! Those all seem very very low! and interestingly, the PNP seem high as compared to the others.

Set#2 The samples from onsemi are all very close. not bad at all. much better results then i would have thought.

So now i am very curious about those low numbers. Hfe is listed as hFE = 15 (Min) @ IC = 8 Adc and hFE = 5 (Min) @ IC = 16 Adc

so those are all below listed minimum. Granted i was not testing at that high of a current. I need to read my manual of this transistor tester a bit more to find out exactly how the Hfe is measured. but just in comparison to the other devices those numbers seems interesting.


Zc
 
M3 hFE Testing

Hi ZC,

Seeing your post, I fired up my M3 and pulled out some similar MJ15022/23 devices that I had previously tested with an older DMM/hFE tester. Keep in mind the last test I ran was several years ago and was a different temperature, humidity, etc. I also tested some other device in my stash and compared them to the datasheets to see how my spare devices and the M3 analyzer measured up.

Here are my numbers:

MJ15022s:
Previous hFE test: 56, 60, 55, 47, 74
New M3 hFE Test: 58, 60, 54, 49, 72

MJ15023s:
Previous hFE test: 91, 114, 10, 96, 134
New M3 hFE Test: 82, 101, 19, 85, 118

2SA1302:
91, 87, 90 (spec'd at 35 to 160)

2SC3281:
53, 70, 55 (spec'd at 35 to 160)

TIP35C:
46, 10, 41 (spec'd at 15 to 75)

TIP36C:
241, 136, 246 (also spec'd at 15 to 75, but the datasheet curves show hFE up towards 175)

I'm pretty sure transistor manufacturers run sample lots and pick statistical numbers that represent mean values. So, depending on date codes, sample lots, production-run volumes, and other factors, it seems hFE values can run +/- 200% or more.

From what the M3 literature said, and from what a couple builders commented on, the M3 tests at very low current levels and then uses some math to extrapolate the current gain. It's not perfect, but for the features and the price, plus from the measured accuracy I checked, it's very close. Besides, when I match devices, I make sure to match hFE as well as Vbe to ensure close tolerances. And I think as long as the M3 is predictable, it's the relative closeness of the hFE matching along with the absolute value that matter.

Paul
 
Wavebourn said:
Are you sure collectors and emitters of NPN in the set #3 are not swapped? :xeye:


HI Wavebourn,

One of the coolest things about the M3 tester, is that the leads can be connected in any fashion, the unit automatically figures out which pin is connected to qhich lead and displays it on the screen!

But, just to make sure, i connected the leads the same way each time.
 
Zero Cool said:
Now that i have my M3 transistor analyzer kit built. i decided to run through some transistors and see what i get.

On hand i have several sets of MJ15024/25 transistors and i was curious to see what i would get.

.....................

So now i am very curious about those low numbers.
Hfe is listed as hFE = 15 (Min) @ IC = 8 Adc
and hFE = 5 (Min) @ IC = 16 Adc

so those are all below listed minimum. Granted i was not testing at that high of a current. I need to read my manual of this transistor tester a bit more to find out exactly how the Hfe is measured. but just in comparison to the other devices those numbers seems interesting.


M3 transistor anayzer
... i do not now how that one works
... but some transistor tester is good to have
... makes 'matching' of devices a bit easier



MySelf, and others, too
still want to do some IN CIRCUIT testing of Gain, Volt Base-Emitter

Simple test could be do using only
- 12 Volt DC supply ( or one 12V CAR battery )
- some Power resistors
- a clever setup to generate 1 Ampere in TRANSISTOR Under Test
- your multimeter
- piece of paper to note results


- I often write the HFE directly onto each TO3
... or attach a little label, piece of paper with test data, onto the tested device

... so next time I build, I can quickly pick Transistors
... knowing some data


lineup :cool:
 
Member
Joined 2004
Paid Member
Hi all,

M3 analyzer : first time I hear from this one. Sorry.

I have done some evaluation of the hFE transistors too :

at Ic around 200 mA :

- MJE15030
max hFE = 114
min hFE = 60

- MJE15031
max hFE = 234
min HFE = 221

A tube has 50 transistors inside. So this is the result of 100 evaluations. These are original ONSEMI transistors.

It follows in the next post.

rephil
 
Member
Joined 2004
Paid Member
Hi all,

I have done also the evaluation of 40 transistors each of the TIP35C and TIP36C at Ic around 100 mA :

- TIP35C :
max hFE = 219
min HFe = 96

- TIP36C :
max hFe = 143
min hFE = 108

These are original ST (SGS Thomson) transistors, more than one tube each.

I have no problem with these to get well matched pnp-npn devices.

It follows in the next post.

rephil
 
Member
Joined 2004
Paid Member
Hi all,

I have now a question : are we (diy buyers) able to get npn-pnp transistors that are reasonably good matched ?

I think that we overlook old trannies as newer are up to date. I have heard that very good amps were done before the new transistors were on the market.

The only way I have to get newer devices is in getting them with someone else that has access to very good transistors. A big buy could be proficuous to all of us : cheaper and surely good transistors that it is possible to match ...

Best regards and Merry Christmas to all of you !:santa2:

rephil
 
Hi all

I do not know the M3 tester, but I've been measuring the gains of transistors for amps for many years with a trusty multimeter and a set of resistors. Matching within 25% is usually possible from a few devices, but I try to match to 10% if I can.

You need to check the gain of the transistors at a range of currents. In particular old trannys have current gains which look like a roller coaster ride. For something like a 50W amp using 2N3055/MJ2955's, the min gain of 20 at 4A is OK for 8 ohm load. But the gain could be 100 at ~200 mA. I try to match over as wide a range from the operating current range if possible (eg 50 mA-4A).

Newer trannys with "flat" gain characteristics are much more ideal and should improve distortion levels at least a factor of 4x - and these usually have better NPN-PNP matching too

Cheers
John
 
rephil said:
Hi all,

M3 analyzer : first time I hear from this one. Sorry.

I have done some evaluation of the hFE transistors too :

at Ic around 200 mA :

- MJE15030
max hFE = 114
min hFE = 60

- MJE15031
max hFE = 234
min HFE = 221

A tube has 50 transistors inside. So this is the result of 100 evaluations. These are original ONSEMI transistors.

It follows in the next post.

rephil


Yes, i am familiar with that problem, thanks to optimized processing techniques the mje15031 has only 5% variations within one batch.
I also have older mje15030/31, not from onsemi sample program, simply bought at some shop. With these the hfe's are all around, no problem to match. It seems that the mje15030/31 can no longer be recommended because of that issue. (mje15034/35 seem to be matchable)
It's sad to hear that you have 50 pieces of these, with no possibility to match.

Mike
 
AndrewT said:
Hi,
contact Classe and offer to return them for a full refund of your costs for these out of spec devices.

So do you feel these numbers are out of spec?

Set #3
NPN - Hfe- 7, 8, 8, 12
PNP - Hfe- 89, 89, 90, 93


I could use 4 devices out of this set that are the closest:

Set #2
NPN - Hfe- 32, 33, 35, 35, 36,
PNP - Hfe- 71, 72, 75, 76, 81


Again, this testing was only done with my M3 tester at very low currents and not at the bias current they are going to be used at.

My concern was that the NPN's from Classse (set#3) seems very very low as compared to the set i removed from the amp and the sample set i got from Onsemi.

If these numbers are ok, i will use them as i would have to imagine that Classe has matched them to whatever level they feel is neccisary. and in checking them with my little tester the matching between devices seems within reason.

Set#3 NPN's have a spread of 5 points. (Classe)
Set#2 NPN's have a spread of 4 points. (OnSemi)
Set#1 NPN's have a spread of 5 points of the devices that are left. (originals)


So the spread between devices isnt much different.

But set#2 is listed as half the gain of the devices that were removed from the amp.

Is that of concern?


Zc
 
Hi,
set 2, be prepared to discard 32 & 81.
But, before you do set up a psu with about 12Vdc available.
Hard wire in a couple of resistors and a pot and measure your actual hFE and Vbe @ the real Iq you intend running.

The devices can be matched for Vbe readings or for hFE DC gain. You do not have enough samples to try matching both characteristics.
 
AndrewT said:

But, before you do set up a psu with about 12Vdc available.
Hard wire in a couple of resistors and a pot
and measure your actual hFE and Vbe
@ the real Iq you intend running.
.


this what i would do to.
Actually i have a special device for this

with

- 1 heatsink, with quick arrangement to get transistor into good contact
you can for example use some clips screwed onto sink
that will hold DEVICE body in good contact, with a bit of force

- 3 homemade quick connectors, with wires to put at transistor pins
without soldering


It has been discussed in forum before, how big such a heatsink could be.
Shouldnt be too big - takes too long to warm up to get somewhat same temp for all tested devices.
Shouldnt be too small - you cant test for higher currents at a reasonable temp.


It is easy to see from most any DATASHEET
that there is a relation between Transistor temperature and gain.
As well as between current level and transistor gain.

12 Volt will allow you to test for a bit more currents,
than if using higher voltage, like the volt intended in your circuit.

As I told before in this topic,
I have found using an old 12 V CAR BATTERY has been perfect for me.
As easy supply as only quickly attach 2 wires from batt+ to batt-

No problems!



lineup
 
Member
Joined 2004
Paid Member
Hi john_ellis,

Thank you for your friendly help in defining the way matching transistors should be done. 10% of match is what I will use from now on.

What I have done :

All of the measurements that I have presented here have been done using a small curve tracer HM6042 from Hameg. It has been so possible to measure all the hFE without touching the devices with my hands. So room temperature at 20 deg. was for all these measurements available. Obviously, the junction temperature has not been maintained at 25 deg. as has been done in the ONSEMI datasheet curves

Vce was 2V and the power dissipated was thus around 200mA x 2V = 0,4 Watt. With so little power, the measured hFE were growing rapidly. So I have standardized for all the transistors the measuring procedure as the faatest repeatable I could do in order to limit the effect of the rising temperaure on the transistor.

I wanted to check the values with what the curves available in the datsheet claim.

It follows in the next post

rephil
 
Member
Joined 2004
Paid Member
Hi john_ellis,

At 200 mA, they say for of the MJE transistors : mostly around 150. I have not bought one MJE 15031 transistor around this value. The MJE are all below hFE = 114. There are even 8 tranistots that have an hFE around 60. I could not believe in what I had measured.

Your procedure :

Please john_ellis,
check that what I am writing now has the same meaning as what you have written ! Thank you.

1- you check the gain in order to get a wanted minimum gain for the sued load at the wanted power _(e.g. 50 W in 8 Ohm is Irms = SQRT (50/8) = 2,5 Arms and it is also 3,54 A peak. So you test the transistor at asy 4A and you want a hFE greater than 20 (so the drivers will have to give 4A/20 = 200 mA peak to the output transistors and with an hFE greater than 100, these drivers should not load the VAS too much (max 2mA).

It follows in the next post

rephil
 
Member
Joined 2004
Paid Member
Hi john_ellis,

2 - you check the gain at say 50 mA and look to find NPN-PNP matched at 10%. This should ensure that the amplified diode does well its job in eliminating the crossover distortion.

Is this a procedure that I should follow or did you mean that I should sample Ic (e.g. say from 50 mA to 4A) at different values and match the transistors within 10% at all these different Ic values ?

In my opinion, measuring at many amps will be difficult as the temperature could "blow" the device and surely will change the measured hFE with the speedy rise of the temperature, this even with a Vce as low as 2 V. This measurement is a difficult one, even using some heatsink as it must be accurate within some % and repeatable.

Some match the transistors also with the Vbe characteristics. What do you think of this kind of match ?

It follows in the next post

rephil
 
Member
Joined 2004
Paid Member
Hi john_ellis,

Opinion :

Yes, newer trannys with "flat" gain characteristics are much more ideal and should improve distortion levels. But are these really sold to hobbyists ? I don't believe it. Let us think e.g. as ONSEMI must do it. Here I have an industry that want to buy 100 000 matched MJE transistors. Yes, we will sell them matched. A good deal for both of us. In the production line remains to be sold the other transistors that could not be selected as matched pairs.

You put them in tubes and give them to "samll numberes orders". All is sold.

These big orders on matched devices are much less required on older trannys as the industry uses the newer because these are also (and this is not the only reason) more attractive to the potential buyers of their products.

My question is : are we sure that the old transistots are not able to produce the same best sounding amps that use the newer components with a careful selection of components that diy-hobbyists can and wil surely do with love ?

It follows in the next post.

rephil
 
Hi current matching or NOT?

Hi,
you will need specialised equipment that can set a low mark space ratio and display realistic measurements during those low mark periods. Impossible for the low resourced DIYer, unless one obtains secondhand kit from a semi testing dept.

Settle for much lower currents. Iq is a sensible starting current.
Including 100 to 200mA would also suit if Iq is very low.
I prefer getting hFE close but trying to match Vbe @ Iq. That way each semi should take near identical current when operating in parallel with it's partners and thus run at similar temperatures (if the heatsink loading is also shared equally). The emitter resistors then provide some feedback to keep the output current share between partners in close balance.
 
Member
Joined 2004
Paid Member
Hi john_ellis,

Practical :

People who sell transistors are of two kind. The local shop in your street and the professional one that generally is far located from you.

The local shop don't look at what they get. Fakes and good components are mixed. All is OK for them. All my fakes come from my local shop.

Professional : they sell to industries. So, they don't like small numbers and in general hobbyists. They (not always) will accept to sell to hobbyist if the total to be paid is high enough (say more than 300 Euro). And they want you buy a standard quantity of componenents (e.g. a tube of all the same transistors, a box (2000 pieces) of 1% 0,5W resistors all the same value, ...). The advantage is that you are sure of what you get. You get only good components, and you know their origin. They are even low in cost. Well, the total to be paid is high, but for a "one time only" buy, it is acceptable.

The drawback is that you will never use all what you have bought from them.

It is time for us all to wake up : we all do need good components and so we al really need group buys.

Best regards

rephil
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.