Electronics > Power/Renewable Energy/EV's

News from the GTI design workbench, Blog

(1/11) > >>

fourtytwo42:
Hi all just so you know I am busy rather than asleep here are some updates to my ongoing GTI designs, basically I have one incarnation in service whilst designing its improved replacement, an ongoing process.

Those familiar with the typical LCL filter used for grid interconnect will understand that whilst it has many advantages it has severe penalties caused by its resonance and phase shift making the feedback around the current loop very difficult to stabilize.   This does not really have a parallel amongst simple feedback discussions such as type I, II & III. 99% of related treatise assume your a mathematician and have a copy of mathcad neither of which is true in my case. However reading material endlessly can procure results and I found some references mentioning the use of LCL capacitor current as a method to assist loop stability and the lights switched on for me!!

Capacitor current has a leading phase wrt to line current and solves the basic problem of attaining a feed-forward effect that I have long sought after, as in predicting line current before it happens :) I hope the enclosed plot illustrates the effect of mixing a proportion of capacitor current into the feedback loop ( the really nasty waveform is just line current, the better is with a proportion of capacitor current).

The next issue is that of sine reference, I must admit to being guilty of simply using the grid voltage!! Fool I am sure many say well OK and now I understand if your grid has a lot of harmonic distortion or your grid coupling impedance is highish your GTI acts as a harmonic amplifier!! So I shall finally resort to synthesized sine reference by micro phase locked to zeroX but why the hell does reference material scare people off with crazy wording such as "rotating reference frame" etc, for gods sake its a simple pll right!! maybe I grew up to long ago :)

Finally I would like a word about the venerable ACS712 hall current sensor, I used this in innocence in the presence of high powered ferrite inductors and transformers OMG I know I should have known but as warning guys it doesn't like external magnetic fields! Anybody know of a good analogue isolation barrier as I am thinking of dumping them in favor of resistive shunts, I have loads of high voltage auxiliary power available ?

Edit so there is the HCPL-7510 from Broadcom (Eww) but it's optical and there must be similar from real designers somewhere :)

   

dmills:
The HCPL-7510 was an Avago part that I think originated with HP/Agilent, it is probably fine (Those guys were good at isolation amps).

TI have some of the old Burr Brown iso amps still in the catalogue, may be worth a look?

Or you could maybe go old school with an IL300 or suchlike?

A 'Rotating reference frame" is just saying 'phasor diagram' to me, nothing too scary there.

Regards, Dan.

capt bullshot:
Use a closed loop magnetic current sensor like the ones from LEM - these are quite insensitive to external fields.
The HCPL-7510 is also a good and reliable device, though there are newer parts with better linearity and bandwidth available from HP/Avage/Broadcom. Better not use the magnetic or capacitive coupled ones from TI  / ADI (to name the most known brands), they might be nasty sources of high frequency EMC. Stay away from coreless or open loop sensors in your environment. With some exeptions, e.g. when properly used, the Sensitec ones may perform well in your environment though beeing coreless.

Anyway, I'd prefer a LEM current transducer of appropriate size. Easy to use, +/- 15V or single supply 5V versions available, and isolation is for free.

Edit: If it is just the capacitor current to measure, there's no need for a DC path, you could use a simple small CT here.

Another one: If you have paralleled capacitors in your LCL, you can use these as a current divider (smaller cap in parallel to a larger has proportionally less current) together with a larger value (easier to get a larger signal amplitude) sense resisor.

fourtytwo42:
Many thanks for your ideas and help :) I must admit to being slightly biased against certain companies that have acquired technology through aggressive acquisition followed by downsizing of employees but sometimes there is no choice :( As for the name changing game, just Grrrrr

I like the idea of a small CT for capacitor current and I have never had a problem with small CT's in SMPS's so that should work :) As for the output current it is good to have the LEM recommendation although I recall these can be expensive OTH in a noisy environment a shunt can loose considerable power to provide enough signal.

I must confess to currently (pun) working in other area's of the project, namely the MPU that requires considerable overhaul and for me a completely new assembler to digest :)

fourtytwo42:
Things are once more moving to the prototype construction phase having created yet another new topology and control scheme. I have been looking at PCB costs and they are ever rising, the pain in part is caused by the size of the components mounted on the PCB so I have decided to revert to my valve ancestry and chassis mount the major components, in this case being the transformer, inductors, large capacitors and also avoid using a PCB for mounting heatsinks.  The only reason for a PCB at all is the voltages are to high (~500V) for stripboard construction and a compact layout is required around the switching devices/drivers.

Another lesson learnt is to divide the power circuits into two PCB's as this will make re-spin half the cost. I made the mistake before of trying to fit a prototype into a fairly tightly dimensioned enclosure and although this worked the PCB costs were much higher than had I given myself more space for a more distrubuted mechanical design.

The control function has not been an issue as it is constructed on stripboard and plugged into the power PCB, very simple to re-work at almost zero cost.

Hope this gives food for thought for others trying to reduce PCB costs in high voltage power circuits :)

Navigation

[0] Message Index

[#] Next page