Current clamp oscillating at 50MHz

[treez]

Hi
We have a DALI power bus comprising a 15mA current clamp in a 15V rail. (as in attached schematic)
As a test, I turned on the transistor, Q4.

When I clipped a TA041 differential probe across the 82R resistor, I saw a 50MHz sinusoidal oscillation across this resistor, of about 250mV pk2pk. (I was using a 50MHz scope). The current in the 82R was on average less than 15mA, …about 11mA on average.

When I unclipped the Diff Probe, and instead attached a normal scope probe aross the 82R, the oscillation was gone, and I saw a perfect 15mA of current flowing in the 82R resistor as it should. (1.24V across the 82R resistor).

I may be being pedantic here, but I want to solve this oscillation. (even though it only happened when the Diff Probe was connected). Do you agree that a good way to damp out this oscillation is to put an RC series circuit from Collector to base of Q3? ( Values about 10n, 10R)

I’ll provide more info later when I can get back to the lab if it's ok.


NXJ1S1215 datasheet
https://power.murata.com/data/power/ncl/kdc_nxj1.pdf

TLV431 datasheet
https://www.diodes.com/assets/Datasheets/TLV431.pdf

TA041 Differential probe
**broken link removed**

 

[betwixt]

If you look at all the application schematics in the TLV431 you will see a capacitor across it's anode and cathode. It contains a high gain amplifier which can be unstable if not sufficiently damped, in fact if you search for TL341 (equivalent) device applications you will find it can be used as an audio amplifier to drive a small loudspeaker!

I suspect it really is oscillating in your design but adding the capacitance of your scope probes to ground at each end is enough to stop it. I don't think the probe bandwidth is related to the frequency you see, most likely it is a coincidence they are both numerically the same. Track layout and device capacitance is likely to be deciding the frequency.

Adding a small fixed capacitor across the device (10nF - 100nF) shouldn't affect your circuits performance.

Brian.

 

[treez]

Thanks, i see load capacitance across anode to cathode.
Regarding stability, there is parts of the datasheet where certain values of load capacitance are not recomended.
We were also wondering about Collector to base RC?

 

[FvM]

50 MHz is far beyond TLV431 unity gain frequency. The oscillation can't exist without the transistor gain. I presume the effects can be reproduced in simulation.

 

[betwixt]

Collector to base negative feedback works when the transistor is configured in common emitter mode where base and collector signals are inverted relative to each other. Essentially you are using some of the inverted signal to selectively cancel the base signal. In your design, there should be no signal on the collector and the transistor does not invert the voltage so applying negative feedback wouldn't work.

The mechanism causing oscillation isn't clear and as FvM states, the regulator shouldn't amplify at such high frequency. Possibly the relatively high impedance at the transistor base at high frequency and the parasitic capacitances are forming an oscillator circuit. I would still go for a single capacitor across the TLV431, following advice in the data sheet. If the probes are sufficient to stop it, even a small capacitor value might be enough to do the trick.

Brian.

 

[treez]

Thanks, i had another look today......in fact, now i am finding that both oscilloscope probes (both diff probe and normal) connected across the 82R resistor causes the 50MHz oscillation.
Sometimes i have to grab the probe ends in order to start the oscillation off.

I didnt read your above till now, and so i had connected a 10R,10nF series RC circuit from collector to base of the transistor, and it did remove the oscillation.

I do however, prefer your idea of a single capacitor across the TLV431 (from reference to anode) so will try that tomorrow.

 

[treez]

I presume the effects can be reproduced in simulation.

Thanks, but as the attached sim (ltspice) shows, its almost impossible to get it oscillating in the sim.

Adding Anode to cathode capacitance, as well as the RC across the base resistor, did not stop the oscillation.
As such, we will try the attached modification next. These are to slow down the transistor.
(The oscillation does not stop the DALI rail being pulled up and down, but we still want to solve it.)
The oscillation (with no mods) happened with FMMT491A and MMBT222A-7F, but not with BC847B. However, I would like to increase the circuits stability margin even with the BC847B being used.

Incidentally, on the PCB, the TLV431’s anode is “white-wired” to the PCB with a short piece of wire…because the layout guy did the footprint the wrong way round…. I am wondering if the stray inductance of this wire is doing it…….even then, adding stray inductance to the sim doesn’t set it off oscillating.