transistor matching - Page 2

AKSA · 29th April 2007, 10:57 AM

Andrew,

A lab supply is fine!

You want to ensure you are measuring at a reasonably constant and repeatable Vce. I designed a small jig using diff pairs and a current source for NPN and for PNP.

It uses a 12V battery, and a voltage reference to fix about 6V on the bases of a LTP. Diff pair action is used to set the DUT to Vce of 6V, and base bias is injected via a known series resistor so that a simple measurement can be used to figure hfe at the current set by the CCS, which is 60mA, the same as the quiescent of the amp which uses these transistors.

Of 100 C5200/A1943 from the same batch, I can usually extract about thirty pairs within 5% hfe and 0.3% Vbe. This gives around 3% matching of quiescent on the output pairs.

Test speed is important. The test dissipates 0.06 x 6 = 0.36W on each device, so they warm appreciably in a few seconds. The hfe and Vbe both change during this time, so you quickly learn to read off the values after one second, no more, so you do need a fast sampling DMM. I use an entry level Fluke.

Cheers,

Hugh

AndrewT · 29th April 2007, 11:38 AM

Hi Aksa,
would a fixed time delay after starting the measurment be equally acceptable? Say 6 or 8 or 10 seconds to allow the junction to heat to about the same temperature for each DUT. This might mimic the operational condition more closely, even for To92 devices as well as the power devices you originally referred to.

BTW,
I already copied your original reply to notepad and saved it for posterity (me

).

Nordic · 29th April 2007, 11:59 AM

suppose you could attach it to one of those aluminium resistors... and then have adjustable thermal point.... by dissipateing x W with a varable voltage source...?

AKSA · 29th April 2007, 12:20 PM

Andrew,

Time is significant. There is a huge difference in time to completion if you need to wait 8 seconds before taking the measurements. If you take your measurement at 1s, it's much quicker, and time is of the essence. I can bound through 100 devices in about forty minutes. Double, or triple that for an 8s wait.

Furthermore, you are comparing from ambient temps. A day of 10C is very different to 20C or 30C. But if all are compared on the same day, at the same ambient temperature, then all is well. It is comparison, after all.

Cheers,

Hugh

derekyu · 30th April 2007, 05:28 PM

Is this a way to check out pairs of transistor?
Transistors Matching

fotios · 1st May 2007, 09:03 PM

I use the following circuit to measure the beta (hfe) of TO-3 output transistors. I believe moreover that mainly in them it needs the matching. With this arrangement, using 2 voltmeters you can read directly in VM2 the beta of DUT (Device Under Test). The trick is found in VR1. You calibrate it in order that the VM1 shows 10mV. Then, those mVolts it shows the VM2, are exactly the beta of DUT. They need the components that appear in photos and a voltage source of 5V / 1A. Also, a heatshink as this in photo (l=15cm) in order to not heated easily the transistor and you have enough time to make the appropriate calibrations. And a TO-3 socket to screw and not to solder the transistor for the same reason as above. The beta of a transistor or hfe (D.C. forward current gain) type MJ or MJL of ON semi it varies between 15 and 75. I use this small circuit for 10 years and it is absolute effective.

AndrewT · 2nd May 2007, 08:35 AM

Hi,
Fotios circuit is excellent.

I disagree on the operation.
Fotios method uses a fixed base current and variable collector/emitter current to measure hFE.

I think one should adjust VR1 to give a fixed value of collector current and then read off the base current (V of base resistor).
Then one can compare hFE at a selected but fixed value of collector current.

BTW, adding a third voltmeter to measure Vbe completes the full set of DC measurements.

Now, if someone could show us how to get our PCs to do the measuring/logging for us. A manual sweep over a sensible range of Vbe would be the icing on the cake.
That would be a nice project.

fotios · 2nd May 2007, 10:29 AM

Quote:

Originally posted by AndrewT
Hi,
Fotios circuit is excellent.

I disagree on the operation.
Fotios method uses a fixed base current and variable collector/emitter current to measure hFE.

I think one should adjust VR1 to give a fixed value of collector current and then read off the base current (V of base resistor).
Then one can compare hFE at a selected but fixed value of collector current.

BTW, adding a third voltmeter to measure Vbe completes the full set of DC measurements.

Now, if someone could show us how to get our PCs to do the measuring/logging for us. A manual sweep over a sensible range of Vbe would be the icing on the cake.
That would be a nice project.

I suppose that, the collector current remains constant for enough time due to large heatshink used to mount the transistor under test and the low current flowing through collector between 20 to 80 mA for the MJ or MJL type transistors - ONsemi datasheet - for a given base current of 1mA (Ib=Vr1/R1=10mV/10Ù=1mA)
Regards
Fotios

jackinnj · 2nd May 2007, 04:32 PM

Quote:

Originally posted by AndrewT

Now, if someone could show us how to get our PCs to do the measuring/logging for us. A manual sweep over a sensible range of Vbe would be the icing on the cake.
That would be a nice project.

If you care to spend $39 on a USB adapter I've already written the program in Visual Basic -- but you will have to wire up the DAC and ADC etc -- or you can get a Measurement Computing data acquisition card and plug-n-play.

AndrewT · 2nd May 2007, 04:33 PM

Hi Fotios,
you have misunderstood where I am going.

Take a ClassA or ClassAB output stage.
Set up the quiescent current. Lets suppose it's 100mA per device and you have 2 parallel devices in each complementary half.

Now, I would match the devices to have the same Vbe when passing 100mA of collector current and try to get a close match for hFE at the same 100mA of bias current.
It would be even better if the hFE also closely matched at a higher collector current maybe somewhere between 500mA and 1A.

By close I have seen other builders aim for +-10% and even +-5% on hFE. But the more important parameter is a very close match on Vbe at the chosen quiescent current (output bias).

Your circuit can easily do this. It's the way you choose to operate it that varies.

29th April 2007, 10:57 AM	#11
AKSA diyAudio Member Join Date: Sep 2001 Location: Melbourne, Australia	Andrew, A lab supply is fine! You want to ensure you are measuring at a reasonably constant and repeatable Vce. I designed a small jig using diff pairs and a current source for NPN and for PNP. It uses a 12V battery, and a voltage reference to fix about 6V on the bases of a LTP. Diff pair action is used to set the DUT to Vce of 6V, and base bias is injected via a known series resistor so that a simple measurement can be used to figure hfe at the current set by the CCS, which is 60mA, the same as the quiescent of the amp which uses these transistors. Of 100 C5200/A1943 from the same batch, I can usually extract about thirty pairs within 5% hfe and 0.3% Vbe. This gives around 3% matching of quiescent on the output pairs. Test speed is important. The test dissipates 0.06 x 6 = 0.36W on each device, so they warm appreciably in a few seconds. The hfe and Vbe both change during this time, so you quickly learn to read off the values after one second, no more, so you do need a fast sampling DMM. I use an entry level Fluke. Cheers, Hugh __________________ Aspen Amplifiers P/L (Australia) www.aksaonline.com

29th April 2007, 12:20 PM	#14
AKSA diyAudio Member Join Date: Sep 2001 Location: Melbourne, Australia	Andrew, Time is significant. There is a huge difference in time to completion if you need to wait 8 seconds before taking the measurements. If you take your measurement at 1s, it's much quicker, and time is of the essence. I can bound through 100 devices in about forty minutes. Double, or triple that for an 8s wait. Furthermore, you are comparing from ambient temps. A day of 10C is very different to 20C or 30C. But if all are compared on the same day, at the same ambient temperature, then all is well. It is comparison, after all. Cheers, Hugh __________________ Aspen Amplifiers P/L (Australia) www.aksaonline.com

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Transistor matching	graham-r	Parts	1	6th May 2009 07:21 PM
How to matching PNP with NPN Transistor	senee	Solid State	16	7th March 2005 11:12 PM
transistor matching?	millwood	Solid State	8	20th December 2003 04:41 PM
transistor matching	weinstro	Parts	28	6th July 2003 02:52 PM

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

29th April 2007, 11:38 AM	#12
AndrewT diyAudio Member Join Date: Jul 2004 Location: Scottish Borders	Hi Aksa, would a fixed time delay after starting the measurment be equally acceptable? Say 6 or 8 or 10 seconds to allow the junction to heat to about the same temperature for each DUT. This might mimic the operational condition more closely, even for To92 devices as well as the power devices you originally referred to. BTW, I already copied your original reply to notepad and saved it for posterity (me ).

29th April 2007, 11:59 AM	#13
Nordic diyAudio Member Join Date: Sep 2005	suppose you could attach it to one of those aluminium resistors... and then have adjustable thermal point.... by dissipateing x W with a varable voltage source...?

30th April 2007, 05:28 PM	#15
derekyu diyAudio Member Join Date: Jul 2005 Location: hk	Is this a way to check out pairs of transistor? Transistors Matching

1st May 2007, 09:03 PM	#16
fotios diyAudio Member Join Date: Feb 2007 Location: ΔΡΑΜΑ - North Greece	I use the following circuit to measure the beta (hfe) of TO-3 output transistors. I believe moreover that mainly in them it needs the matching. With this arrangement, using 2 voltmeters you can read directly in VM2 the beta of DUT (Device Under Test). The trick is found in VR1. You calibrate it in order that the VM1 shows 10mV. Then, those mVolts it shows the VM2, are exactly the beta of DUT. They need the components that appear in photos and a voltage source of 5V / 1A. Also, a heatshink as this in photo (l=15cm) in order to not heated easily the transistor and you have enough time to make the appropriate calibrations. And a TO-3 socket to screw and not to solder the transistor for the same reason as above. The beta of a transistor or hfe (D.C. forward current gain) type MJ or MJL of ON semi it varies between 15 and 75. I use this small circuit for 10 years and it is absolute effective.

2nd May 2007, 08:35 AM	#17
AndrewT diyAudio Member Join Date: Jul 2004 Location: Scottish Borders	Hi, Fotios circuit is excellent. I disagree on the operation. Fotios method uses a fixed base current and variable collector/emitter current to measure hFE. I think one should adjust VR1 to give a fixed value of collector current and then read off the base current (V of base resistor). Then one can compare hFE at a selected but fixed value of collector current. BTW, adding a third voltmeter to measure Vbe completes the full set of DC measurements. Now, if someone could show us how to get our PCs to do the measuring/logging for us. A manual sweep over a sensible range of Vbe would be the icing on the cake. That would be a nice project.

2nd May 2007, 04:33 PM	#20
AndrewT diyAudio Member Join Date: Jul 2004 Location: Scottish Borders	Hi Fotios, you have misunderstood where I am going. Take a ClassA or ClassAB output stage. Set up the quiescent current. Lets suppose it's 100mA per device and you have 2 parallel devices in each complementary half. Now, I would match the devices to have the same Vbe when passing 100mA of collector current and try to get a close match for hFE at the same 100mA of bias current. It would be even better if the hFE also closely matched at a higher collector current maybe somewhere between 500mA and 1A. By close I have seen other builders aim for +-10% and even +-5% on hFE. But the more important parameter is a very close match on Vbe at the chosen quiescent current (output bias). Your circuit can easily do this. It's the way you choose to operate it that varies.