Follow-Up to Simple Transistor Matching Circuit
Вставка
- Опубліковано 5 жов 2024
- After receiving many great comments and ideas on the last video, I thought I'd do this quick follow-up. In addition to trying some of the viewer's ideas, I try one last experiment. Let me know what you think!
Thanks Tony - whether right or wrong, useless to some or not, at the end of the day we are all learning together.
I like your idea, thanks for the video.
Bipolar transistors Vbe (and silicon diodes too) have -2mV / C temperature coefficient. This effect is so consistent that it can be used to build a thermometer. When you first install the second transistor, your hand has increased its temperature a few degrees C, and they need to thermally stabilize to get a good match. To get better than 1-2 mV of matching, you should tie them together mechanically, typically this is done flat-side to flat-side with a tiny clamp, ideally with thermal compound. Same with any circuit that needs Vbe matching.
In any IC or matched pair of transistors, extreme care is taken to put them close to each other to make sure they thermally match.
Good luck!
Dave Erickson
hi there, greetings from england. why not remove switch and put diodes in parrallel, regards bob
Thank you for your time. The goal was achieved, assembling a circuit that allow us to compare two transistors. It was made clear that this wasn't meant to be a transistor gain evaluation, so no false expectation. What could be a nice add-on is the integration of an op-amp that would simply turn a led on or off if the output voltage is within a preset range or value. Doing so would make the circuit even more portable as a DMM would be required only for the calibration.
Thanks about the diodes. I thought there was something I was missing with them back to back. Thanks for the videos !
Put the 2 diodes in parallel and you won't need the switch.
Well, anti-parallel, but we get what you mean.
Just getting into the hobby myself but it was very interesting to watch and listen to what you had to say and 73 is back to you too Merry Christmas
Good approach. I'll take your idea and will work in something similar. Thanks for sharing!
You could have gotten the performance that you wanted by placing the diodes in reverse-parallel. No need to make another video about it, and your switch setup will work, but the switch, besides increasing cost, introduces another thing for the operator to have to remember.
When running it on AC like at the end of the video, the switch scheme may not be working as you want.
Great. Just a thought: would it not be essential to compare the results from this tester to the cheap tester and your gold standard trace test? That way we know if the results are valid, and if they offer an advantage to the standard cheap testers we all have lying around.
This needs more traction UA-cam
Please do my guy here a solid, promote these forms of racism against Transistors, instead of the current agenda.
Thanks.
I’m in the middle of designing my own assessment tools too. This definitely made that bag of tricks.
Thank you
I would be very interested in board to build this, and personally I would rather use my oscilloscope to see what’s going on with the signal, thanks for posting this one!
Thanks for sharing. Good stuff dude.
Either just connect the diodes in antiparallel and forget the new switch, or simply leave out the 3.9K, the diodes and the new switch. None of them have any significant effect with 0.1mA going through the transistor. If you're going for 1mA test current, then the self-heating from having 12V across the transistor will now be 12mW which will raise the junction by about 5°C, so can start to introduce noticeable errors in your Vbe matching. In this case, by all means use the 3.9K resistor and two antiparallel diodes to clamp the collectors so that only about 1.3V appears across the transistors.
Very nice and helpful idea. Can I do this with a 9V battery as well?
Tony I really enjoy watching your videos but on this one I am not sure your circuit is working as well as you think it is. I have a problem with the back to back diodes feeding the Base of the transistors. One will be forward biased but the other one will be revers bias so there will not be any current flowing into the base of the transistors. This could be why the meter was incrementing as some parasitic capacitance is charging. A better mod might be to replace the diods with a pot between the + and - voltage with the wiper feeding the base of the transistors. This way you could not only see if the transistors match at one bias point but at many bias points. This will test the linearity match as well as gain match. Just my 2 cents
Thanks for fixing the diode problem. I am still curious how a pot on the base performs under the same conditions. Good Idea to use a sine wave and a scope. Sometimes using a meter it would not be obvious if there was a oscillation going on.
For an NPN, the diode clamps the collectors to about +0.65V. The bit you may be missing is that the bases of both transistors are connected to ground, so are fixed at 0V and therefore the emitters are at -0.65V. The base currents are drawn from the 0V supply. The diodes are not feeding anything. They simply pass excess current to ground and work as 0.65V shunt regulators, or "clamps". The meter was changing because every time the transistors are handled, they warm up by several degrees Celsius, and the Vbe has a temperature coefficient of -2mV/°C. You are seeing the last transistor handled cooling down.
HaHa.. I saw those two back to back diodes in the first video and thought, "Well this guy definitely knows more than me", because I just didn't get it.
At least now I know I'm not crazy.
If you use a 12V rms ac signal, you see NPN transistors turning on as the voltage goes positive on the collectors and you're measuring exactly the same as with a positive dc voltage input.
When the voltage goes negative on the collectors, you're seeing the emitter-base junction break down at about -5V for the 2N2222, which is almost matched with different samples of the same device (hence the small spikes). However, it is more like -6V for a BC547, so you see differences. If you think there's some value in matching Vebo for a pair of transistors, then there's value in your experiment. Sadly, I'm completely at a loss to think of any possible value in having a pair of transistors whose reverse E-B junction breakdown is matched. YMMV.
Isn't using the oscilloscope stepping away from the idea of a cheaper way of transistor matching?
Indeed, using a scope increases the overall price tag but I'm pretty sure there are little DIY scopes, like the DSO138 for instance, that would be plenty sufficient for this kind of purpose and some others as well. The DSO138 is cheaper than a DCA Peak transistor tester by far.
HI tony
Interesting project I like the idea of using the scope we could make the PCB using the tonner transfer method I made a pioneer SX 525 power amp pcb using this method as an experiment theirs an idea for a video for you to do watching with big intrest on the comparator
Thanks
Colin
You could include aTEXTOOL socket for holding the transistors.
this a belive in.keep on
I made the PCB layout today I'll upload the video as soon is ready mine is different cause I decide to split the test for NPN and other for PNP "no toggle switch" I need to add the trim pot I forgot xD
I’d like one of the circuit boards
when testing with AC you should replace the switch and the 2 diodes to 2 anti-parallel diodes
You really do not need 12V either, a 5V should work fine too.
How are you going to get 1 mA thru the transistor with only .6v going t the collector?
This isn't even enough voltage to forward bias both junctions of the transistor.
At best, you are only biasing the collector to base.
Assuming an NPN, you have the collector at +0.65V, the base at 0V, the emitter at -0.65V. That's Vce = 1.3V and the transistor will work fine @ Ic=1mA.
There's yur problem, one of the 2 resistors is backwards. =P
Voltage is moving cuz your finger heat!!! ~2mV/K.
Also if you need a matching transistor just buy HFA3046, HFA3096, HFA3127, HFA3128 chips They are matched transistors and thermally track too. You can not get discrete transistors in separate packages to track thermally.
Have you checked pricing of your suggested chips to replace some penny parts ?
Besides that it is hard to integrate this weird form factor into any given antique preamp design as a replacement part - 12 $ a piece without shipping for one single chip - which probably picks up a lot of noise too because you need long wires for left and right - sorry, no way !
And i don't want to even ask about signal separation in such a case.
We're talking about old hifi repair here, not about some fancy newly designed board, where you can easily use such modern stuff of course.
At 3:26 you say that a "back" biased diode will drop 0.7V, a "reverse" biased diode will not drop any volts.
A "forward" biased diode will drop 0.7V.
Your switch is correct, you are "forward" biasing, that is in NPN mode, Collector is positive and you have the anode of the diode connected to collector rail.
Put your DMM on the collector rail and show us.
What ARE you actually doing? There isn't enough being compared to tell much of anything.
Didn't you even read the comment on the previous vid you made, and don't you correct your mistake?!
YOU ARE MATCHING Vbe NOT HFE!!!!!!
I understand turn on HFE and increments to total of HFE is critical for matching Transistors and Mosfets. Especially using Mosfets in parallel in Power Loads. But I was wondering if one would adjust time base and voltage, and look closer at the towers widths between the two transistors or Mosfets. Could there be seen a HFE difference..!?! Now this would be using the AC 12-0-12 and placing cursors in the center of towers and measure out side of towers of each side in increments. Its turning on, and its turning off. Not to start argument, but just wonder.
@@overload3996 Sure, but you would need calipers, now wouldn't you?
@@airgead5391 Funny, I was wonder if slew rate in tower build time opposed to HFE. If HFE has different values , then on time slew rate ,I think would change also. But I use PICO, so I can over lay differences. And has more power and ability then other low dollar scopes.
@@overload3996 I don;t know exactly what you are asking but slew rate is dependent on the whole topology.
@@overload3996 what are you talking about? I think you are thinking of raise time and not slew rate! Slew rate is the change in voltage in 1 microsecond!