Hi, I’ve got a really weird phenomenon with some transistors and a square 1.5MHz signal occurring with my transistors on a breadboard; I’ve tried to understand it so many times but I really have no idea what’s going on at this point.

Basically, I’ve got a 1.5MHz square wave @ about 5vpp forward biasing a transistor through a 390 Ohm resistor that is tied to Vcc. I’m only biasing the transistor through one of its junctions, and have the 3rd lead just floating. What’s really weird is when I probe the signal at the node between the transistor and resistor with my O-Scope, I get a 400mVpp signal that is kinda square and spiked mixed together. However this doesn’t happen when I replace the transistor with just an ordinary Si diode; with the diode, the resultant transient signal is the same width as the diodes voltage drop, and it maintains the same shape as the signal feeding it. Yet every single transistor I have produces this really strange dampened signal which is really small, less than the Vbe/Vbc voltage drop even with it forward biased. And I just cannot explain what’s happening!?

Anyone have any ideas or thoughts on this strange occurrence? Any feedback is much appreciated.

 

Before we get started, how much experience do you have working with oscilloscopes?
It is possible that spikes and ripples that you are observing are artifacts and not real.

Are your oscilloscope probes set to 10x attenuation as they should be?
Are your oscilloscope probes properly adjusted for frequency compensation?

 

I’ve been using my oscilloscope for close to 3 or 4 years now and have been studying and experimenting with component level circuit design for almost 10, ever since I was about 14.

My probes are well calibrated, but my desk is a huge mess, and the probe cables have some loops in them which I know is certainly not best practice…but if anything I think the breadboard or transistors could be to blame (although I’ve tried swapping the transistors: 2N3904, 2N3906, S9018 more than 10 times now) so maybe that’s not the problem.

but it’s also very possible I may have damaged my probes a few years ago from several high voltage measurements exceeding 1kv, and maybe they’re damaged now, but I’ve been able to use them up to even 230MHz (though it’s certainly finicky at those frequencies especially being so close to my scope’s bandwidth) and it’s been working fine here and there.

 

you did not answer point about calibration. did you try calibrating probes? or at least did you try swapping them around to see if the spikes move to another channel. you can connect both probes to same point and the display should be identical.

 

How are you connecting the circuit together?
How long are the circuit leads?

 

you did not answer point about calibration. did you try calibrating probes? or at least did you try swapping them around to see if the spikes move to another channel. you can connect both probes to same point and the display should be identical.

 

If you touch your scope probe on the positive and negative points on your breadboard that are connected to the 5 volt power supply what does the display show ? (Anything other than 0 volts on the negative and + 5 volts on the positive. )

Les.

 

Here’s the diagram of the entire circuit, if it would be more helpful to take a picture of it on the breadboard the I can do that too…

The circuit uses an inductive oscillator i designed myself a while back, which is amplified and buffered to produce the input signal for the circuit under test (the input is the blue trace; whereas the green trace is the test point)

 

@IsaacSutt

yes, i saw that but maybe i did not express my self clearly. so let me try again: what is the basis for the statement that probes ARE calibrated? is there an evidence of recent test or verification to confirm that? i was offering suggestion to verify that. can you get the probes out of calibration in both directions (capacitive and inductive characteristic)? or if connecting both probes to a changing signal (such as test output on the scope), do they really show the same waveform? if not problem is with probe or with scope.

 

also your circuit uses inductors but there are no bypass capacitors on supply. that is an obvious issue.

 

Below is an LTspice simulation at 100kHz:
There is a large delay of a couple microseconds until the transistor-base emitter voltage reaches it's steady-state value (red trace), likely due to the base-emitter capacitance and the storage time of the charge carriers being removed (switching diodes are optimized to minimize that reverse-recovery storage time).

When you have an apparent anomaly in your measurements, it is useful to vary the test parameters, which can often help zero in on the source of the anomaly.

 

Anyone have any ideas or thoughts on this strange occurrence?

I duplicated your test but with only a 150KHz square wave because that is the highest frequency it can deliver.
The level and quality of the signal output is highly dependent on the signal level and the supply voltage.


2N3904 at 5 volts, Output appx 100mv p-p.



1N4148 at 5 volts, Output appx 100mv p-p



2N3904 at 2.5 volts, Output appx 2v p-p



1N4148 at 2.5 volts, Output appx 2v p-p

 

Diodes and bipolar transistors are fabricated with P-N junctions. Yes.
The similarity ends there.
Each junction has different doping concentrations, depth, depletion region, capacitance, and characteristics.
Don't expect that a diode and transistor to exhibit the same behaviour.

 

@IsaacSutt

yes, i saw that but maybe i did not express my self clearly. so let me try again: what is the basis for the statement that probes ARE calibrated? is there an evidence of recent test or verification to confirm that? i was offering suggestion to verify that. can you get the probes out of calibration in both directions (capacitive and inductive characteristic)? or if connecting both probes to a changing signal (such as test output on the scope), do they really show the same waveform? if not problem is with probe or with scope.

I used the built-in 1KHz test signal on the oscope to adjust the probes with the little side knob using a screwdriver while probing it… I just twist until the edges appear as square as possible, with no visible spikes on either the positive or negative transitions. When I probe the same signal on both channels simultaneously, they do indeed display the same exact signal.

Also, I agree with your observation and comment that I’m not using any bypass capacitors as is normally advised on the supply to reduce ripples/noise, however I don’t believe my signal source is high enough in amplitude or frequency, nor do I believe the signal under test is small enough to really have that be a significant concern. But I might try adding some just in case, and maybe see what my scope reveals in case the supply noise is more significant than I assumed.

 

since there is no capacitance in the circuit, frequency of oscillation is depending on parasitic values of components, breadboard and actual circuit construction. i still strongly suggest adding bypass capacitors across 5VDC and testing again. without this, even inductance of supply wires becomes part of the system. also not sure what type of PSU is used. since you have the scope, can you check ripple just to be sure? if using switching regulator, even small ripple can have an effect on your oscillator

 

Below is an LTspice simulation at 100kHz:
There is a large delay of a couple microseconds until the transistor-base emitter voltage reaches it's steady-state value (red trace), likely due to the base-emitter capacitance and the storage time of the charge carriers being removed (switching diodes are optimized to minimize that reverse-recovery storage time).

When you have an apparent anomaly in your measurements, it is useful to vary the test parameters, which can often help zero in on the source of the anomaly.

View attachment 285389

I just measured the supply rail noise, and it doesn’t look too good, but adding some caps across the supply only reduces it down to like 160mVpp. I guess it’ll take some serious LC filters and RF chokes to bring it down any further idk, would be nice to know when it comes to some of my VHF projects… Anyways, I tried moving my probe’s ground lead and also shortening it using just a spring that came with it and a small 22AWG wire on a different probe and the signal now pretty much looks exactly like your LT spice simulation, but that baffles me a bit, is this something that’s completely unavoidable? Does this mean that I’ll have to completely change my design approaches to account for this? Maybe I should completely retire breadboards for anything above the KHz range now

What started this whole side quest of strange behavior was when I tried amplifying the signal again through another PNP transistor and found that all 3 leads where complete at Vcc voltage potential yet the prior transistor was still showing the same original signal and the only thing between the two stages was one 390 Ohm resistor, yet the PNP amplifier/Buffer looked like all three leads were shorted to Vcc with a little noise on each leg, and this was the case for every PNP transistor I tried, yet I checked each junction before and after with my DMM using the Diode function, and nothing was wrong with the transistors.

Yet, after all this, i still feel clueless. Who knows, I might even try that additional stage one more time and come to find that everything now works as expected, it wouldn’t be the first time I’d encounter some crazy F’N quantum shit or just unpredictable scopes and loose breadboard connections.

Yet I’ve even built my own bjt based 27 and 56MHz Transmitter & Receiver on breadboards before and got it to work from up to almost 20 ft away… and yet again I still feel so stuck from time to time when I encounter the simplest of problems but can’t solve them. Maybe breadboards should no longer take part in any RF prototyping and I should just go through many many hours of PCB fabrication for my designs come to find that it needs redesigned and shipped 5 more times before i solder everything and it works.

 

You will find even diodes are different.
Compare a 1N4001 with a small signal diode such as 1N914 or 1N4148.

 

Thanks for everyone’s input, I appreciate all the quick replies, help and feedback!

 

Maybe breadboards should no longer take part in any RF prototyping

They can be done, but the breadboard should be a Vector-type proto board with a copper ground plane, and using careful layout with as short leads as possible between parts.

 

There are other ways to prototype HF RF circuits.
Look up (1) dead bug and (2) Manhattan style circuit construction.

Also this method is very effective: