Thinking widely is often the key
pwrtrnx.com

Thinking widely is often the key

We get asked to solve a fair few power electronic issues,

from the fairly straight-forward, right up to the tricky and involved where hardware & software interact due to timing and interference issues ( and mis-applied control theory ).

There are a few key things that will assist the power electronic engineer in getting to the bottom of niggly issues that prevent things working as hoped.

1) Hard switching converters; these generate noise as a result of too fast mosfet turn on, and where diodes are suffering reverse recovery as a result of overly aggressive mosfet drive - you have lots and lots of noise right up to 300MHz and beyond.

Some of this noise is near field magnetic from di/dt in current loops - which you really can't easily see - and this can really get into everything - from gate drive IC's, BJT's, bond wires inside IC's and even control boards and wiring some distance away.

In a prototype situation - make sure your heat-sinking is very good to allow you to slow the turn on under hard switching - so that you can at least run your circuit at moderate power to see what is going on.

Allow for fitment of snubbers - hopefully you won't need them ( unlikely ) - but it will at least allow you to test your control scheme.

Excessive dV/dt at mosfet turn on and sometimes turn off creates near field electric RFI that can get into and onto pcb tracks and even bond wires inside IC's. Again easing gate drive and use of snubbers can alleviate electric near field RFI, to allow you to test the control.

If your hard switching converter has mosfets and diodes that are heat-sunk directly to an earthed heatsink - this provides a capacitive propagation path for EMC - which can really upset things when the time comes to try and pass EMC.

Hard switching converters connected to transformer wires and similar provide another path for radiating high dV/dt throughout nearby spaces.


2) Soft switched converters; usually these are fully ZVS above 50kHz and although they do not often suffer from overly high dV/dt on the switching nodes - they can have high di/dt current loops which can get into nearby gate drive IC's and onto control wiring.


3) Current sensing; newbies rarely put enough thought, time or attention into current sensing,

for pwm type control - often there needs to be some filtering of the leading edge spike at turn on to prevent turn off a few 100nS after turn on.

For average current sensing we often see far too much filtering of the signal which puts delays in the control loop - which can be tricky to overcome and lead to much slower overall response of the converter if not fixed.

Local dV/dt and dI/dt can often upset current sensing, especially for poorly located CT's, or specialist current sensing IC's.

Control IC's that rely on 30mV of signal in a low ohm shunt for control of full power are often a poor choice for converter control. IC's that can amplify a shunt signal need to be fast and some what RFI immune.

Putting your gate drive current spikes through your current sense is not a great idea.


4) Voltage sensing; again we often see far too much filtering of voltage signals, delaying the signal and leading to poor control - often oscillatory - which leads the designer to try and compensate in the feedback loop, often not very successfully.


5) Not designing for extra output capacitance; it often happens that the end user adds quite a bit of capacitance to the output of your converter sometimes at the end of long-ish cables,

if you do not test and design for this scenario - you can end up with a power oscillator instead of a power supply - this is quite often true for heavily inductive loads too - so do your testing !


6) Heat-sinking; you always need more than you think - so work out the worst case losses and then run resistors on the heat-sink to simulate the semiconductor losses and look at the temperature rise - is it too much ? time to re-think the case.


7) Magnetics with gaps; make sure that sensitive parts, e.g. CT's, control wiring, any type of sensor, are not near gaps in magnetic core structures - this can really be a hidden problem.


8) Ripple current in capacitors; many newbies simply do not do the proper calculations to estimate the ripple current and therefore heating in capacitors, at higher voltages de-rating must occur, the losses then are: Vac x Iac x DF as given by the manufacturer. Electrolytics at higher voltages have very poor ripple current ratings and need proper film capacitors to handle frequencies above ~ 30kHz, as many electrolytics are series resonant near or above this frequency and therefore present only a low inductance plus internal resistance to high frequency currents ( hence the popping sound after a few hours at full power ).


9a) Software issues; it is a good idea to make a low power proto-type of your hardware to test the functionality of the controlling software - this allows the designers to work out the bugs in the uP or FPGA software, with hardware that can be protected from blow ups and that makes very little RFI - and so won't upset the chips doing the control.

If you can bullet proof the control software at low power - this will assist greatly on the full power embodiment - although interference can still occur.

Thus it makes sense to make the 1st full power converter as quiet as possible so that you can test the control fully.


9b) Software issues; with the best will in the world, very subtle errors and bugs can be left in control software - whether assuming certain states at power up, overflow errors, A/D errors, dealing with fluctuating LSB's ( noise ), faulty WD Timers, or logical errors in the code that look very logical at first glance.

It pays to get your software checked by peers in the same field - more than once - this can save huge headaches later on and is simply good engineering practice !

and lastly:

10) Crappy test and auxiliary power supplies; if your power converter is so noisy it upsets any common or garden power supplies - or power sources - that you are using to test your system - then you will spend a lot of time chasing your tail on non existent issues !

You would be surprised how often this happens, whether isolating gate drive supplies, control power supplies, or 12V 50A bench supplies that act as the power source - do not trust them until you have seen clean outputs with your converter running at full power - this is one of the more subtle and overlooked issues in trouble-shooting - RFI causing Vout reduction in aux power supplies, or oscillation, or current limit kick in ( sporadically ) - so - user beware !


pwrtrnx.com


see also:

https://www.dhirubhai.net/pulse/niggles-viewing-high-side-gate-drive-colin-j--39slc/?trackingId=4lInmb7XbG681rYrZnSgmw%3D%3D





great to see all the reads on this topic !

回复

要查看或添加评论,请登录

Colin J. Tuck ( Snr VP Global Corporate Engineering )的更多文章

社区洞察

其他会员也浏览了