Precision transistor gain measurement circuit

Started by jgarnold, October 07, 2010, 08:11:51 AM

Previous topic - Next topic

jgarnold


I recently purchased the DuoYi DY294 transistor tester from a dealer in Hong Kong. This model is sold by many dealers on Ebay for about $50 shipped. I bought it mainly to test transistor gains, but it can do many other types of tests including reverse breakdown voltage of diodes and the voltage capacity of electrolytic caps. The DY294 has three gain measurement settings that supply different base currents to the transistor (10uA, 1mA, and 10mA). Unfortunately, I quickly discovered that this is a poorly designed piece of equipment. First, the power supply is completely unregulated internally (whether you use four AA batteries or an external 6V supply). Second, the three transistor base currents are created by simply passing the unregulated supply voltage across one of three resistors (600,  6K, or  600K ohms). This would probably be accurate enough if the supply was regulated, but it's not. The displayed gain is completely dependent on whether the batteries are new or if they are weak. Can you imagine if your multimeter displayed different values for resistors depending on the battery voltage?

So, I decided to create my own precision transistor gain measurement device... It is completely independent of the battery voltage, is accurate to 1% or less, and is unaffected by temperature. The best thing about this circuit is that it's simple and costs almost nothing to build.



The circuit uses an LM334 adjustable current source IC. With just two resistors and a transistor (or diode), it can produce a temperature stable current source from 1uA to 10mA. I selected components to produce a 10uA current source. The 2N3904 is used to compensate for the LM334's positive temperature coefficient (see the LM334 datasheet for full details). You can also use a diode here, but I found that using a transistor as a diode (by connecting the base and collector together) provided the best accuracy. I heated the circuit with a heatgun and it continued to produce 10.0uA regardless of temperature. I put the circuit into the freezer for 30 minutes and it still produced 10.0uA. I varied the supply voltage from 2 to 20 volts and it still produced 10.0uA. If you connect your multimeter to the circuit in mA current mode, the displayed reading can be converted to gain by simply ignoring the decimal point. If your multimeter displays 2.72mA, then the transistor gain is 272. Easy!

I also added a 820 ohm resistor to the circuit to limit the amount of current through the transistor. This not only prevents damage to the transistor, but also protects the fuse in your multimeter. I blew the fuse in my Fluke 87 meter when I accidentally shorted it across the battery. Cost of a new fuse... $15. Ooops!  820 ohms with a 9V battery is good enough to measure gains as high as 1000-1200.

-Jeff

waltk

Jeff,

Nice! Thanks for sharing that.

I bought the same tester, but have always been suspicious about the tranistor gain measurements from it (wondering why they weren't always repeatable).  So I built a regulated PS for it with an LM338 and a waltwart. 

I think I'll build your circuit just to have a dedicated gain tester - and maybe add a switch to disconnect the base of the DUT for measuring GE transistor leakage.

(I also recently blew the 400ma fuse in my Fluke 87 III - after almost 20 years of continuous use.  Who would have thought that fuse is so hard to find and pricey.)

jgarnold

>and maybe add a switch to disconnect the base of the DUT for measuring GE transistor leakage.

I added your switch idea to the schematic.

-Jeff

R.G.

As you may have read in my other posts, be careful of demanding too much accuracy. What you're trying to measure accurately changes based on how long it's been since your finger touched the transistor, and other effects.

Knowing the instantaneous sea level at any time is almost superfluous. There will be waves.
R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.

PRR

> ...power supply is completely unregulated ... transistor base currents are created by simply passing the unregulated supply voltage across ...resistors ....

Gee, it sounds a LOT like the typical Beta-dependent FX pedal circuit.
  • SUPPORTER

jgarnold


The entire point of this circuit was to learn about constant current sources and to make an accurate piece of measurement equipment. I don't see anything wrong with that, especially since it's cheap and easy to build. However, I just realized that I designed the circuit for NPN transistors only. So, I flipped the entire circuit around to work with PNP, but I'm not sure that I did it right. Is the following schematic correct? Is there an easy way to add an NPN/PNP switch to the circuit to handle both?

-Jeff



tubelectron

Hi jgarnold & All's,

1/ about your circuit : I'd be tempted to say that if you have reversed all the polarities, then it's enough to test PNP. The switching operation may then require a rotary multi sections switch... But I am not a solid-state ace, my specialty is tubes...

2/ what you have discovered about the transistor meter DUOYI DY294 is worth to be noted about its unregulated PSU (somewhat surprising but...). I also have a transistor meter DUOYI DY294 and was akeen to think it was reliable when comparing with other transistor testers, but PROBABLY because the batteries where new and that I often used it with an external regulated PSU. So I will definetly use it with regulated PSU for more accurate readings : thanks +++.

3/ the transistor meter DUOYI DY294 - despite its PSU lacks - is a fair instrument, but it is unable to make reliable measurements on Ge transistors, because the leaks (mainly the Ice0) are disregarded in the measurement process - like on the majority of transistor testers.

So I also decided I needed a more accurate transistor bench test to measure NPN and PNP low and medium power transistors, but it is not a "simple portable direct instant reading" device... Specs : Ib = 0-100µA / Hfe = 0-100000 / Icmax = 1A/ real measurements of Ib, Ic, Ice0, Icb0, Ices / voltage = 3-12VDC (or more). It requires a regulated variable PSU and 2 DMMs able to measure µADC. The schematic appears on the instrument itself, and, as you can see, it is a very basic circuit. Here it measures an OC72 Ge transistor presenting an Ice0 of 139-140µA, so a rather good specimen :




A+!

I apologize for my approximative english writing and understanding !
http://guilhemamplification.jimdofree.com/

R.G.

Quote from: jgarnold on October 09, 2010, 01:11:49 AM
The entire point of this circuit was to learn about constant current sources and to make an accurate piece of measurement equipment. I don't see anything wrong with that, especially since it's cheap and easy to build.
There is nothing wrong with that at all. Anything you want to do for fun is OK. I was noting that relying on precision measurements of something that moves and drifts is a problem if you were relying on that, there are several pitfalls.

Fun with current sources? GREAT!
R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.

Earthscum

#8
I just got an order in the other night, and started matching P and N transistors right off the bat... (13 matches out of 20 PNP's and 30 NPN's isn't bad!).

The DMM I use for tranny gain is a cheapie from HF. I started with an empty ice cube tray, threw any trans back in the bag that measured under 170, and started sorting. When I went back and double-checked the pairs, they all started measuring 5-10 higher than when I started! BUT, they all increased together. I figured out it was the light I had on them... when I started, they had been in a bag in the basement. After about half hour under the light, they all warmed up pretty good (about 7 deg F air temp from dark basement to the living room).

So I resorted a couple, and found some more matches. I went through after I was all done and they were all still within .1-.2% that I had originally matched them at (200 and 201 would be .1% match, right?).

Anyways, I'm just matching them so when I use 10k, PNP, NPN to ground, and bases tied together, I get the .6V diode drop at the Collector of the NPN. With just a bit of double-checking, the cheapie DMM I use seems to be REALLY consistent for my purposes.
Give a man Fuzz, and he'll jam for a day... teach a man how to make a Fuzz and he'll never jam again!

http://www.facebook.com/Earthscum

R.G.

Quote from: Earthscum on October 09, 2010, 12:19:39 PM
Anyways, I'm just matching them so when I use 10k, PNP, NPN to ground, and bases tied together, I get the .6V diode drop at the Collector of the NPN. With just a bit of double-checking, the cheapie DMM I use seems to be REALLY consistent for my purposes.
This is an interesting point that I got hammered on by the instructor in our instrumentation class. Accuracy, precision, and repeatability are simiilar concepts, related in application but definitely not the same thing. Accuracy is "how close to the reality does it measure?" Precision is "how finely does it read?" and repeatability is "how close to the same reading of the same thing does it do on repeated trials?"

The implication of "precision instrument" is that it is accurate, precise, and repeatable. That's not the case, although instrument makers try to get all three of them. Something can be consistent by being wrong the same way every time. And frankly, that's probably more useful for matching transistors than high precision, which now that I write it down, is what I was trying to say. For matching, it's more important to be repeatable and consistent that precise or accurate.
R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.

John Lyons

Well, I thought it was a nice contribution.
Yeah, it's dealing with an imperfect devise under test but whachagonnado?
Thanks for the schematic and effort Jeff!
Basic Audio Pedals
www.basicaudio.net/

R.G.

Quote from: John Lyons on October 09, 2010, 08:18:49 PM
Well, I thought it was a nice contribution.
So did I. My concern was that someone not get disappointed at getting precision instead of accuracy.

QuoteYeah, it's dealing with an imperfect devise under test but whachagonnado?
Indeed - about all you can do in an imperfect world is try to understand where the levers are and what they really do. It is indeed all about learning about what works and what doesn't, and using imperfect - but better! - results as you can.

Very little learning is ever truly wasted.

R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.

cthulhudarren

#12
I was looking for a circuit to test transistor gains (Si mostly). Was there an actual schematic posted, I can't see anything?  I can't quite get it from just reading the paragraph describing it.  Thanks in advance!

Ripdivot

I would like to see the schematic as well...

LucifersTrip

if you reply with quote you can see the original link...which is dead now a couple years later

for now, the best I could find is a cached thumbnail



edit:
but don't forget
http://www.geofex.com/Article_Folders/ffselect.htm
always think outside the box

cthulhudarren

Quote from: LucifersTrip on April 05, 2012, 02:18:18 AM
if you reply with quote you can see the original link...which is dead now a couple years later

for now, the best I could find is a cached thumbnail



edit:
but don't forget
http://www.geofex.com/Article_Folders/ffselect.htm

A thumbnail is better than nothing! Thanks!   Of course I have and treasure the geofx article.

cthulhudarren

Quote from: cthulhudarren on April 05, 2012, 08:46:15 AM
Quote from: LucifersTrip on April 05, 2012, 02:18:18 AM
if you reply with quote you can see the original link...which is dead now a couple years later

for now, the best I could find is a cached thumbnail



edit:
but don't forget
http://www.geofex.com/Article_Folders/ffselect.htm

A thumbnail is better than nothing! Thanks!   Of course I have and treasure the geofx article.

I've been preparing to build this circuit with an NPN/PNP switch, but it occurred to me that in addition to the battery needing to be reversed, doesn't the regulator circuit need to be reversed also because the LM334 has V+ and V- sides?

cthulhudarren

Quote from: cthulhudarren on April 12, 2012, 11:15:22 AM
Quote from: cthulhudarren on April 05, 2012, 08:46:15 AM
Quote from: LucifersTrip on April 05, 2012, 02:18:18 AM
if you reply with quote you can see the original link...which is dead now a couple years later

for now, the best I could find is a cached thumbnail



edit:
but don't forget
http://www.geofex.com/Article_Folders/ffselect.htm

A thumbnail is better than nothing! Thanks!   Of course I have and treasure the geofx article.

I've been preparing to build this circuit with an NPN/PNP switch, but it occurred to me that in addition to the battery needing to be reversed, doesn't the regulator circuit need to be reversed also because the LM334 has V+ and V- sides?

In case anyone else cares: I've been experimenting with bread-boarding this. It appears that the original poster was incorrect about the 2n3904, the tempco is -2.2mV/C as compared to the 1N457 that is recommended that is -2.5mV/C. The tempco of whatever diode is used should be 11 times the tempco of the LM334 in order for the ratio of R1 and R2 to be 10.

Also he has R1 = 13.8k for 10uA, which needs to be 13.4k in order for the circuit to output the desired current. I also had to drop L1 to 110k in series with the pot to be able to get the right output current since with the tempco of the 2N3904 the ratio of R1 to R2 is less than ten.

I'm going to be wiring this up to be able to test NPN and PNP, so in addition to reversing the leads on the battery I'm also thinking I'll need to turn the current source circuit around as well as otherwise the V+ and V- on the LM334 will be not correct and the diode will be wrong. Thoughts?