Design of a transistor curve tracer

[rogerk8]

Hi!

I am trying to design a transistor curve tracer.

I have made some progress but there is still some problems.

I am step-sweeping Ib/Ug BOTH back and forth but somehow these traces are not the same, it seams like when I trace one way (as from low to high Ib current) I get a certain type of knee behaviour but when I trace the other way I get another type of knee behaviour and I don't understand this at all.

The Uc/Ud-sweeping is made by an ordinary triangle wave, ideally this should constitute a saw-tooth signal but I am sweeping it back and forth instead because cutting it off could mean "shadow-signals" which I think is worse.

I am sending you some pictures.

Best regards, Roger
PS
I am not that good at electronics but it is good to be back here again even if I think it was more than a year ago since i visitited you last, my whole life revolves around electronics so really I don't care if I am good at it or not, trying is more important. A Chinese friend of mine has produced 10 peaces of this PCB, the OP:amps down to the left has however been wrongfully configured by me (and only me) because plus is up which I however think is a CAD program flaw beacause OP-minus is ALWAYS up, you may read any textbook about standard OP configurations.

 

[(*steve*)]

The difference in the forward and reverse trace may be caused by inductance in your wiring.

Why not use a sawtooth and blank the trace during flyback?

 

[73's de Edd]

THAT's the way TV, is done . . . . . ( NOT that they would even WANT to be having a backwards / reverse scan display.)

 

[(*steve*)]

Well... My curve tracer actually does do a forward and reverse scan of each trace.

This allows you to see hysteresis effects (amongst other things).

There is an adjustment done during alignment to null out differences caused by wiring inductance.

Since this affects both the forward and reverse trace, nulling it out makes the curve tracer more accurate.

 

[Audioguru]

It does not matter if the positive input or the negative input of an opamp "is up" but the feedback usually is negative feedback. You wrongly show three opamps with positive feedback that does not work.

I have worked in electronics all my life and have never seen or used a curve tracer.

 

[rogerk8]

It does not matter if the positive input or the negative input of an opamp "is up" but the feedback usually is negative feedback. You wrongly show three opamps with positive feedback that does not work.

I have worked in electronics all my life and have never seen or used a curve tracer.

Hi Audioguru!

It matters to a human being like me because I expect plus to be downwards, that is the way I have learned to use OP:amps so when Eagle put out an OP the "wrong" way on the drawing board it is easy to connect it the wrong way, this is however a kind of bad excuse for not doing it right, anyway I have solved it in the manner of an ameteur by using an adaptor that switches +/-, I am posting a picture of this adaptor that by the way does not work so well due to intermittent falure while pushing on the capsule.

Bets regards, Roger
PS
I find it very nice writing in english again even though I'm not so good at it.

 

[rogerk8]

The difference in the forward and reverse trace may be caused by inductance in your wiring.

Why not use a sawtooth and blank the trace during flyback?

Hi Steve!

I have rather long wires now in test mode but normally the signals goes in a PCB, do you mean that the inductance can be too high anyway?

But why does the inductance differ due to sweep direction?

Isn't an inductans independent of current direction?

I have another theory, the Uc/Ud triangular ramp differ when it reaches zero, perhaps it begins positively at a point a bit from zero and ends "negatively" another bit from zero and this perhaps means that DUT is reversed biased somewhat, I am reaching her am I not?

Finally, I do not like the sawtooth approach because I tested it once and got "ghost signals", the only way I think around that is to complicate everything and use the Z-input for blanking, this while it "should" suffice with back-and-forth as I see it.

On the other hand, I do have problems now

Best regards, Roger
PS
Attaching the stimuli

 

[rogerk8]

Well... My curve tracer actually does do a forward and reverse scan of each trace.

This allows you to see hysteresis effects (amongst other things).

There is an adjustment done during alignment to null out differences caused by wiring inductance.

Since this affects both the forward and reverse trace, nulling it out makes the curve tracer more accurate.

Hi Steve!

This is another good argument for not using "single" sweep direction, I didn't think of that however.

This gives me an idea of perhaps even be capable of sweeping transformer cores, I really like transformers and have designed a few mainly for tube amplifiers but it is always interesting to see the hysteresis of a transformer core.

I still do not understand the assymetrical inductance problem, though.

Best regards, Roger

 

[Audioguru]

It matters to a human being like me because I expect plus to be downwards

Google Images shows hundreds of opamp. All inverting have the Positive input on the bottom and all non-inverting have the Negative input on the bottom like this:

 

[rogerk8]

Google Images shows hundreds of opamp. All inverting have the Positive input on the bottom and all non-inverting have the Negative input on the bottom like this:

Hi Audioguru!

This is not so important because obviously you have to be thorough but I do not agree with you because I have three books on electronics, one written by Millman (a more wide view on electronics) and two written by Walter G.Jung (OP:amp circuits only), exactly all OP:amps in Walter's books are with minus upwards, in Millman I did however find ONE that used plus upwards and that was a non-inverting circuit which I actually think i written better with plus upwards.

But my OP-guru is Walter and he never, regardless of configuration, uses plus upwards which is what I have learnt even from the school bench.

None of this really matters but I wish minus to always be on top because if it were that way I would never have made this kind of huge mistake (my board is even autorouted by a friend of mine beacause I failed which means that wires even goes beneath and inbetween IC pins which makes it really hard to correct mechanically).

Best regards, Roger
PS
I think the inverting configuration is extremely versatile, firstly you may even attenuate, secondly gain is extremely easily calculated, thirdly you can easily "inject" a bias due to the perfect summing feature without croostalk and the drawback called inverting may be "easily" fixed by hooking up yet another inverter, it is fun working with OP:amps in that aspect.

 

[(*steve*)]

But why does the inductance differ due to sweep direction?

Because inductance opposed the change in current. If current is increasing, as it will with the sweep going one way, inductance will work one way. And when current is decreasing, in the opposite sweep direction, it will work the other way.

I do not agree with you because I have three books on electronic

It doesn't matter if you decide to always show the non inverting input at the top, as long as you were it up correctly!

However, be aware that there are standard ways of drawing certain common circuits, and if these common circuits happen to have the inverting input at the top then you can't reproduce then the way people are used to them. This leads to the reader having to put more effort into reading your circuit, possibly misinterpreting it, and possibly missing errors that woukd otherwise have been noticed.

 

[rogerk8]

However, be aware that there are standard ways of drawing certain common circuits, and if these common circuits happen to have the inverting input at the top then you can't reproduce then the way people are used to them. This leads to the reader having to put more effort into reading your circuit, possibly misinterpreting it, and possibly missing errors that woukd otherwise have been noticed.

But this is why I am asking for a standard, if everybody (read every CAD program) for instance used minus on top then everything would be solved, using plus on top sometimes and minus on top the other times just makes it hard to actually design and show your design to others and what's worse, make your design work (without having to think of everything which I think is the whole point of using computers).

Anyway, this is not so important but it kind of irritates me.

I have made som progress tonight, by adjusting the level of the step voltage (Us) I have got a better I/V-diagram that however still shows some differences in the back-and-forth sweep but the difference is less, there was a kind of high peak around zero voltage before but that one is gone, yet the traces are not equal so their sweep direction gives a different result around zero.

Best regards, Roger
PS
I am attaching a picture that shows this, it also shows the sensitivity, in X-direction it is then 1V/DIV (that is a maximum of 10V mainly because I temporarily run the system on 12V and my TL084 are no rail-to-rail OP), in Y-direction sensitivity is 10mV/DIV which approximatelly means 1mA/DIV due to a collector current measureing resistor of 10 Ohms, the appromations are due to the fact that I am using VAR on both X and Y channel (to fit it all on the scope).

I am controlling the base current by using two OP-based BJT voltage followers for higher currents (want to test TO3s also), the lower side is set by a blue LED voltage reference (some 3V), the other side has the exact same DC bias but is stepwise swept, NPN for instance has 3V on the lower side but while the step volage is 1Vpp 3V->2V is swept on the upper side, this way base curent is flowing upright and into the base.

Maximum base current here is some 1V/5k=0,2mA and the highest trace means some 8 DIV*1mA, this does not compute

 

[Audioguru]

I think the inverting configuration is extremely versatile

But it has a low input impedance. A non-inverting opamp can have an extremely high input impedance.

did you understand that your schematic had the opamp feedback connected to the wrong input?

 

[(*steve*)]

everybody (read every CAD program) for instance used minus on top then everything would be solved,

Hardly.

 

[rogerk8]

But it has a low input impedance. A non-inverting opamp can have an extremely high input impedance.

did you understand that your schematic had the opamp feedback connected to the wrong input?

You are right about the non-inverting amp but I think it is more suitable for audio because when you design a system like mine which almost may be called a PID regulator (the only thing I don't use is derivation), the inverting configuration is better mainly because of it's summing and attenuating feature, the inverting configuration may also have quite high input resistors, just use a FET device (but if gain is high there is still a problem).

Yes I understood that the opamps where connected wrongly but I understood it after a serious fault seach by measureing without getting almost any signal (in one case I did get the Us signal on output but with 5Vpp in I had some 20mVpp out and moreover, it was non-inverted).

I did of course think this was strange but it wasn't until I took a close look in the schematic that I saw my mistake.

I did some calculations why the output signal might be almost zero and I came to the conclusion that while a true inverter has a current that goes thru both resistors but is virtually grounded at input the amp must swing downwards when the input signal is positive and it does so because the input signal is at minus, if however the input terminal is plus there can be no current because output would want to go high and the output potential will then be higher than the input potential.

Please correct me if I'm wrong.

Best regards, Roger
PS
I find it hard to really understand opamps, I only almost understand the basic configurations.

By the way Us is not 1Vpp it is 0,5Vpp, 0,5Vpp/5k=0,1mApp and this is the total ladder, each step is then 0,01mApp. 10mV/DIV Y-wise sets highest collector current to 8 DIV*10mV/10R= 80mV/10R=8mA (1mA/DIV) which is rather low because a rough estimate is the highest base current times 200 or 0,1mA*200=20mA (where hfe for BC546B is estimated as 200). However, true hfe is a dynamic parameter and while I do not have the plot in front of me a rough guess is 0,01mA/DIV (0,1mA max should actually be devideded by 8 DIVs Y-wise), dynamic hfe is then the collector current I have inbetween two traces divided by the base current I have inbetween two traces that is approximatelly collector current inbetween a DIV divided by basecurrent inbetween a DIV that is 1mA/0,01mA=100 which might be true because I am testing at a base current of such low value as some 10uA and it is a well known fact that hfe is worse at low currents.

I am planing to trim the bottom traces today, I think I know what the problem is because as I have calculated and trimmed things there is a rest-voltage of some 0,3V while it should be zero, these 0,3V gives a 0,3/0,5=60% lowest sweep above ground so while it should read 0mA base curent, it reads 0,06mA and while every step is 0,1mA it follows that the lowest base sweep do not work so well. The problem here is that I use blue LEDs as voltage references and they are fed from a 7-18V supply but even LED's are not perfect Zener's so the reference will drift. I do not think it will drift so much (perhaps 100mV) but any drift in their regard is not so welcome, but if I trimm the stuff as to allow some drift there is no problem but I must trim on the right side so to speak because I can't have that the base current reverses polarity, I can however have that there is some offset so that zero is not zero but of course I wish that offset to be as small as possible.

 

[(*steve*)]

Applying net positive feedback does not result in a smaller output voltage. It will result in the output being at one supply rail or the other (or however close to them the op amp can swing them).

 

[rogerk8]

Applying net positive feedback does not result in a smaller output voltage. It will result in the output being at one supply rail or the other (or however close to them the op amp can swing them).

True but AC-wise this does not matter, there can be no output signal because, as I interpret you saying, the OP is saturated at either rail.

Best regards, Roger

 

[(*steve*)]

It depends. The input signal can also cause the output to alternate from one supply rail to the other.

 

[rogerk8]

It depends. The input signal can also cause the output to alternate from one supply rail to the other.

Interesting, please teach me some basic opamp theory.

If, as in my case, Rin=Rf a rather large signal is required to make the opamp saturate (in the correct configuration), using the wrong configuration offset and DC bias/input should be the key to which rail i rests in and when it rests there it cannot move because it would need a negative voltage greater than (offset + bias) to change rail.

Might this be true?

Best regards, Roger

 

[rogerk8]

I have now tried to trim my curve tracer but I failed, it doesn't get any better than this.

But I am not sad, I am kind of proud.

It turns out that the residual collector current was so small that it can be neglegted, the behaviour of BC546B is thus as you see above, the dynamic current gain at low Ib is obviously small.

I think this is a success and I will move on to integrate this device in my box-project (KTT, Knoppson Transistor Tester).

It is kind of hard to understand that I have come this far, I have three box-projects and these are this project KTT, KAP (Knoppson Analog Processor) and KGA (Knoppson Guitar Amplifier).

KAP involves two microphone amplifiers (KMA and KMA_D, where D stands for descrete), an OP-based function generator (KFG) with a wire-wrapped module enabling autosweep (KFD) capable of sweeping from 20Hz to 20kHz in one sweep and a descrete power OP using CFP's for low voltage performance.

KGA is intended for battery use only and it uses a push-pull transformer along with MOS-FETs, the reason is that I have noticed that the nice clipping of tubes is not due to tubes, it is due to the use of an output transformer. I have test data showing this and most interesting is the fact that it seams that it is not even harmonics that sounds good but odd harmonics because a tube push-pull amplifier will ideally only generate odd harmonics so I have designed KJA which is an JFET amplifier inspired by D.T.N Williamson but sadly, I use MOS-FETs for the output even though I actually own a couple of POWER LU1014D JFETs (try to find such ) and this from a tip from this nice forum, however the biasing is more simple when it comes to MOS, especially while my amplifier is of Class B.

Best regards, Roger
PS
The power MOS are made in Japan by the company Hitachi, 2SK175 is their name. Sadly these are extremely hard to find but the datasheet is awesome, I/V-diagrams included.