TNY278/279/280 drain current shows ringing
Hi,
I am investigating a circuit using the TNY278/279/280, in which this device increases in temperature quite significantly under moderately heavy load (7W output) when the input voltage is increased (30-40 degrees C rise from 110VAC to 260VAC input).
I realise this is not much information to go on by itself (I will send schematic and layout later, if needed to answer this quesiton)... but my concern at the moment relates to the drain-current waveform captured. In this case, a TNY280P, in reduced-current mode was used.
The attached image shows the current into the TNY280 drain-pin as the yellow trace at 200mA/div. The drain voltage here is the blue trace, at 200V/div. The current was measured using a Chauvin Arnoux E3N AC/DC current probe - set at 1A/100mV; zeroed when the SMPS is off; polarity correct (+ve side is the transformer pin and clamp, -ve side is the TNY280 drain pin).
The waveform appears quite different compared to the waveforms shown in the datasheet and the PI university videos. Does this look about correct, even for a transformer with shield windings, or do the graphs suggest some issue here? Could the large 'on' time be a cause of the increased temperature with input voltage? I am also concerned by the significant negative spike that occurs, and the ringing which seems to be present throughout operation.
I'd much appreciate any comments on these waveforms to know if the device drain current looks more or less correct, or is a concern here.
Kind Regards,
dave-ml
Comments
The Traveler:
Thank you for your reply. I've double-checked the oscilloscope and probe for AC coupling - it doesn't seem to be an option on the probe and the oscilloscope is definitely set to DC coupling. Sorry, I didn't word very well what I mean't about the 'on' time in my previous post (but thanks anyway) - I meant that the current probe suggests current flow for a significantly longer time than the MOSFET is on.
Unfortunately, I'm unable to post the requested files publicly, but could e-mail them confidentially to you if I may do so?
Many thanks for your time,
dave-ml
dave-ml:
I did a quick search on the current probe you're using and I think that is most likely your problem. The bandwidth on the current probe you're using is only 100kHz and it looks like that particular clamp on probe is better suited for low-frequency high-current measurements.
In our lab, we use mostly Tektronix current probes and current probe amplifiers. The model I have on my bench is a TCPA300 current probe amplifier with a TCP312 current probe. This setup gives me 100Mhz bandwidth at up to 30 amps.
Looking at their website, I wouldn't use anything from Chauvin Arnoux, at least in regards to their current probes. The highest bandwidth probe they have is the one you're using and you need something capable of going up into the MHz range. Power supplies are unique in that you need tools to look at very slow AND very fast current and voltage signals. In a typical power supply, you have everything from DC up into the 10-30 Mhz frequency spectrum.
If you don't have access to a better current probe and if you're *VERY* careful about how you setup your connections, you could insert a small value series resister between the transformer and drain pin. This would let you read the drain current as the differential voltage produced across the series resister.
-The Traveler
I had wondered about the bandwidth for this clamp but had expected it to result in some attenuation more than anything. I'd hoped the clamp rating of 100KHz would provide some insight into the circuit behaviour, which I believe runs at approximately 130KHz. However I can see how this probe would be inadequate if something with significantly higher bandwidth is needed. I had actually looked into the exact model of current probe and amplifier you use - but opted to try the Chauvin Arnoux for cost reasons.
Prior to acquiring the current probe, I had tried using a 0.1 ohm shunt resistor as you mentioned; a screenshot of one of the oscilloscope captures is attached. I noticed that the SMPS circuit did not operate correctly with the shunt in place (although I haven't yet looked into exactly what was happening, I could see output was not reaching regulation). Also, the waveform for the voltage across this shunt again suggested some odd behaviour - peak drain currents of 15A (negative, for some reason!). This appeared to be the case regardless of if the probe and oscilloscope were set to 10x or 1x. The oscilloscope used is rated as a 25MHz model and the circuit under test was completely isolated from mains.
I will try this method again using a smaller shunt resistor when possible. Again, thank you for your time in helping with this issue. Kind regards,
dave-ml
dave-ml:
A couple of suggestions:
-Use a high-impedance differential probe if you have one. If you don't have the differential probe, use two seperate scope probes referenced to SRC on each side of the shunt resister. Then you can use the math function in your scope to subtract one WF from the other. You need to do this to avoid putting additional capacitance from the scope into that circuit node.
-I would try using a slightly larger resister. Maybe around an ohm or so. This will give you a large signal to look at and hopefully less noise.
- Don't use the clip-on ground leads of your scope probes. These will pick up radiated EMI with their large loop area. Use a scope probe socket instead. Something like on the bottom of this page should work well. http://www.probemaster.com/product_info.php?cPath=1_10&products_id=266
- I would definitely recommend picking up a good current probe amplifier setup. Having the right tools will make your job much easier and the quality of your work will be much better as well. Along the lines of good tools to have, a good high-impedance, high-frequency, high-voltage differential probe can be really helpful. I have several of these in the lab: http://www.probemaster.com/product_info.php?cPath=21&products_id=62
-The Traveler
Thank you again for your reply and suggestions. It sounds like a bit of a gear upgrade is in order for us!
The following current probe (CP8030): http://www.szjwj.com/Product-2.asp?sendid=57 (requires a page translator) is something we are looking to purchase. From the specifications and signal response graphs given, it looks like this probe meets the requirements you've previously suggested - would you mind terribly taking a quick look to see if there is any reason why this probe would not be suitable before we place an order? It would be much appreciated.
Kind Regards,
dave-ml
dave-ml:
I'm not familiar with that brand but quickly looking at the specs, I think it should work a lot better for you than the one you're using now. :-)
Let me know how the new probe works out.
-The Traveler
Hi Traveler,
This current probe finally arrived and does indeed work much better than the previous one. I've attached the drain voltage (yellow trace) and current (blue trace at 100mA/10mV) waveforms with this new probe, which are much closer to what we expect. I hope these images help anyone else who may have had the same issue trying to capture drain current waveforms.
Thank you for your help with this issue,
dave-ml
dave-ml -
Wow! The differences between the traces in those scope plots is like night and day! It's amazing how helpful it is to have the right tool for the job.
Was there anything else I can help with, or did this answer your question?
-The Traveler
dave-ml:
I agree that your drain current waveform looks very unusual. You "on" time is actually very small. The MOSFET "On" time is the time that Vds=0V. During the time while Vds=0V, your drain current should be a fairly linear ramp with an initial spike at turn-on when the MOSFET Coss gets charged.
Just out of curiosity, is your drain current waveform AC coupled? This could be a setting on your oscilloscope or on your current probe. This might explain the unusual nature of the waveform.
If you can, please attach you PI Expert design files, a PDF of your schematic and PCB layout as well as any other information you have.
We'll get to the bottom of this!
-The Traveler