Print Page | Close Window

Transistor Matching

Printed From: Graham Slee Hifi System Components
Category: DIY AUDIO
Forum Name: DIY Audio questions and answers
Forum Description: www.diy-audio-kits.com ... superb audio kits for experienced constructors
URL: https://www.hifisystemcomponents.com/forum/forum_posts.asp?TID=4233
Printed Date: 04 Dec 2022 at 12:17pm
Software Version: Web Wiz Forums 12.01 - http://www.webwizforums.com


Topic: Transistor Matching
Posted By: peterb
Subject: Transistor Matching
Date Posted: 02 Jun 2017 at 10:34pm
My Power amp uses a pair of input differential MPSA42 transistors and they determine the amps quiescent current. 
I have a fairly large difference (60 to 110 mA) between my right and left amps Q' current and so I am looking at trying to match a pair to see if that will bring the higher current down. Looked in to try to buy some some but no luck.

Then I came across one of the cheapo Chinese universal testers, this one in particular.
http://www.ebay.co.uk/itm/192072316090" rel="nofollow - http://w w http://www.ebay.co.uk/itm/192072316090" rel="nofollow - w.ebay.co.uk/itm/192072316090

I know, I know, it is cheap and probably has lots of shortcomings but from everything I have read it does make a reasonable job of measuring, amongst other things, hfe.

My question is: Will a hfe matching with this 'tool' serve my purpose?
Assuming I try to keep the temperatures as close as possible to constant.

Thanks




-------------
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2



Replies:
Posted By: BAK
Date Posted: 03 Jun 2017 at 1:19am
That tester looks like the bare minimum. Many DVMs (digital volt meters) today have a transistor hfe test and diode test. I would prefer to use a DVM as I can get more reliable comparison tests.
 There are many low-cost DVMs today at less than 2x the cost of your example.

The idea here is to get a relative comparison. The test does not need to be very accurate, just repeatable by using the same tester under the same conditions. Meaning, you will get the same test result repeatedly with the same device... ie, a result of 101 will be 101 again and again, not 102 or 99. An inexpensive DVM can do this.
If you need a transistor with an hfe of 40 to 50, the DVM can help you get a reading within 5% or less, but the readings will be the same % off consistently. The same will be the results of resistance measurements. The repeatability is performed by using the same meter. 

To get a closely matched pair of transistors, you need to compare their hfe AND their AC input resistance, little . Little r is the resistance "seen" across the base to emitter junction in the forward biased direction. To use this for transistor matching, the technician does not need to know the actual little r value, just make a relative measurement for comparison.
 I have done matching this way many times. Yes it is tedious work. 
1 example: I have done this to match substitute transistors in a 1000 watt, 70 volt power amp. (Same transistor part # as originals but were not supplied by the OEM, so 1/10th the cost or less.) It had 10 matched TO3 transistors in its output stage. If they were not matched, the power amp would "fry" the outputs. I always got the "hard-to-fathom" electronic repairs.

Little r can be tested with DC resistance measurements for comparison testing.
Using the higher resistance ranges of a DVM (2K to 20M, you'll have to find any high R range that will give a reading) and the "diode" test range (both forward biased and reverse biased) and record your test results in a chart for the samples. Keep the samples physically arranged in the same order as your chart... Sample 1, Sample 2, Sample 3, etc ...
 The tests are:
1. hfe
2. "high R" (base to emitter = positive on base) (between 500 ohms and 20Meg)
3. "high R" (emitter to base = negative on base) ( should be >1Meg)
4. "diode R" (base to emitter = positive on base) (between 100 ohms and 10K)
5. "diode R" (emitter to base = negative on base) (should be > max reading)

The diode test on a DVM can be used on BJT transistors without harming them as they are a junction of 2 diodes where one junction controls the current in the other junction. How much the base to emitter junction controls the other junction is governed by the AC input resistance, little r.

You will need to buy 10 to 50 transistors of the same run or batch to get enough samples for your tests. The more samples you have, the closer you can get a match.

To avoid this tedious testing,  you can use a substitute transistor part # that can be purchased in "matched pairs" or contact a transistor manufacturer for 2 matched pairs of the original part # ( always get yourself a spare pair). 
Is the original transistor a MPSA42 ? 

Datasheet @  http://www.mouser.com/ds/2/149/MPSA42-889941.pdf" rel="nofollow - MPSA42-889941.pdf


-------------
Bruce
AT-14SA, Pickering XV-15, Hana EL, Technics SL-1600MK2, Reflex M or C, Lautus, Majestic DAC, Technics SH-8055 spectrum analyzer, Eminence Beta8A custom cabs; Using Proprius
Enjoy Life your way!


Posted By: peterb
Date Posted: 04 Jun 2017 at 9:58am
Thanks Bruce for your detailed reply.
I will spend some time digesting it but a couple of points/answers.
Yes they are MPSA42s in the circuit but I haven't found any matched pairs on offer.
I understand that matching is a relative rather than an absolute measurement, it is why I thought my little Chinese tester would do the business.
I have an old, simple, Fluke DMM which unfortunately does not offer any semiconductor tests, perhaps I will have a look around to see what I can find that does do them.



-------------
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2


Posted By: Graham Slee
Date Posted: 04 Jun 2017 at 11:40am
Quiescent or standing current in the output push-pull isn't set by matching transistors in the LTP - long tailed pair - you call input differential.

It is set by the bias transistor which appears as part of the voltage amplifier load. There is usually a trimmer nearby which sets the voltage across the bias transistor, which turns the output stage more or less "on".

Often in manufacture this trimmer is adjusted for minimal distortion and often the measured quiescent current differs once this is set.

By mucking about you'll probably screw-up the distortion on one channel. Best left alone. If it ain't broke why fix it?

If it doesn't sound right buy Proprius Wink


-------------
Not simple enough for Google-Bot to understand...


Posted By: peterb
Date Posted: 04 Jun 2017 at 12:30pm
Thanks Graham for your input. Yes it was the LTP I was considering matching.
I will revisit the measurements I made and check my facts
On the positive side there is no audible distortion on either channel so leaving well alone is a good option.


-------------
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2


Posted By: Graham Slee
Date Posted: 04 Jun 2017 at 12:48pm
The problem with high voltage transistors is their current gain is low to start with. An hFE of 25 at 1mA to me is a joke. So I guess the bipolar LTP and VA is driving MOSFETs?

If an amp designer selects a high hFE transistor there is less chance of mismatch (if there is any) because he can set a much lower design hFE and so it wouldn't matter about matching because the true hFEs are going to be considerably higher. Here the 2N5551 offers an improvement, but the user must ensure it will be OK in all other characteristics.


-------------
Not simple enough for Google-Bot to understand...


Posted By: BAK
Date Posted: 04 Jun 2017 at 3:19pm
Originally posted by peterb peterb wrote:

My Power amp uses a pair of input differential MPSA42 transistors and they determine the amps quiescent current. 
Originally posted by Graham Slee Graham Slee wrote:

Quiescent or standing current in the output push-pull isn't set by matching transistors in the LTP - long tailed pair - you call input differential.

It is set by the bias transistor which appears as part of the voltage amplifier load. There is usually a trimmer nearby which sets the voltage across the bias transistor, which turns the output stage more or less "on".
 Thank you Graham. I obviously missed Peter's mention of the "pair of input differential MPSA42 transistors". 
 Peter, Graham is spot on about there being another transistor stage after the differential input voltage amplifier stage that converts the voltage to a current drive for the output transistor stage. Many amplifiers employ a quiesent current control trimmer, some don't. This stage, in my aged memory, was the stage I thought you were referencing. The "driver stage" is where the matching of transistors may help only when there is no Q' current adjustment trimmer.

 I would look out for crystalized, dull grey, or cracked solder joints anywhere in the current driver and output stages. Also, old electrolytic capacitors can become leaky drawing extra current. These failing solder joints and leaky caps can change the Q' current... as well as old carbon resistors increasing in value. All the parts in an amplifier affect the operation of all other parts in actual operation.
 After 10 to 20 years age, all of these factors can become a problem, but re-soldering can also cause more problems if not done very carefully. 

Originally posted by Graham Slee Graham Slee wrote:


By mucking about you'll probably screw-up the distortion on one channel. Best left alone. If it ain't broke why fix it?
So true, Graham!


-------------
Bruce
AT-14SA, Pickering XV-15, Hana EL, Technics SL-1600MK2, Reflex M or C, Lautus, Majestic DAC, Technics SH-8055 spectrum analyzer, Eminence Beta8A custom cabs; Using Proprius
Enjoy Life your way!


Posted By: peterb
Date Posted: 05 Jun 2017 at 1:39pm
Ok, admission of guilt! Don't know if it is senility or the red wine but I my facts were wrong in my first post. 
It is not the quiescent current that was different left and right but the output dc voltages. Big difference!
The q current is indeed set by a trimmer and is to spec, the dc on the outputs are however 60 and 110 mV.
The amp was built and working 18 months ago and I put on my memorised to do list finding the reason for this difference. I thought it could be due to different Vbe of the LTP transistors, and that matching a pair would be the answer.
How 18 months changed that to current in my brain, I don't know why. Apologises everybody.


-------------
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2


Posted By: Graham Slee
Date Posted: 06 Jun 2017 at 11:55am
60mV develops 0.45mW into 8 Ohms. 110mV develops 1.5mW into 8 Ohms. I don't think it of any real consequence.


-------------
Not simple enough for Google-Bot to understand...


Posted By: peterb
Date Posted: 06 Jun 2017 at 7:47pm
Yes true, I guess I am just being anal to want them in balance.

-------------
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2


Posted By: Graham Slee
Date Posted: 06 Jun 2017 at 9:23pm
It is probably the difference between the VA amp hFEs on each channel that is loading one LPT more than the other. I would have expected another trimmer to null the offset?


-------------
Not simple enough for Google-Bot to understand...


Posted By: peterb
Date Posted: 06 Jun 2017 at 10:09pm
Sadly no other trimmers.



-------------
Peter
---------------------------------------
Dual 505-1, Cyrus CD T, DIY 80W MosFet amp and PreAmp, 2xKEF 103.2


Posted By: Graham Slee
Date Posted: 07 Jun 2017 at 8:42am
T6,7 being a darlington configuration won't be loading the LTP much at all.

Correcting what I said earlier, there is no bias transistor because this is a MOSFET output stage which has a negative temp. co. so it won't go into thermal runaway like a BJT output stage. The BJT uses a bias transistor thermally connected to the output devices (sometimes in a rather particular way) such that temp. co.'s cancel, preventing thermal runaway. This design simply uses the resistance of a trimmer but with zenner protection in case it turns the output devices both on.

-------------
Not simple enough for Google-Bot to understand...


Posted By: JardCrocker
Date Posted: 31 Oct 2017 at 4:24pm
Hi...i am a new user here. As per my knowledge test does not need to be very accurate, just repeatable by using the same tester under the same conditions. You will get the same test result repeatedly with the same device.< style="display: none;" id="__symantecPKIClientMessenger" -install-updates-user-configuration="true" -sups-flavor-configuration="true" -extension-version="0.5.0.161">

http://www.7pcb.com - turnkey pcb assembly



Print Page | Close Window

Forum Software by Web Wiz Forums® version 12.01 - http://www.webwizforums.com
Copyright ©2001-2018 Web Wiz Ltd. - https://www.webwiz.net