Electronics > Projects, Designs, and Technical Stuff

DC Bus Voltage sensing cirquit

<< < (3/4) > >>

jbb:
That circuit looks like a good start. I would add an LED which lights when the dump is on, a flyback protection diode across the dump resistor (it’s likely to be inductive) and a bit of noise filter capacitor (maybe 1uF film across the rails?).

The trip point of the clamp should be set with care. You should do some calculations to see how high the DC link might get if your incoming AC supply is 10% high. You don’t want the dump load trying to pull the AC line voltage down, it’s wasteful.

Hysteresis is good, and can be quite large. The power MOSFET should switch at quite low frequencies (100Hz or slower is my guess). Indeed, seeing the LED go blink blink would be fine. You should also look at how much margin you get between clamp voltage and over voltage for the drives.

Finally, I recommend an earthed metal box for this one in case something goes bang.

eliafavero:
Many thanks to all of you!
i'll run some simulations, prototyping and then i'll come back and post what configuration worked for me.
Very much appreciate all your contributions!

eliafavero:
Soo.. I am back as promised.
A couple of months have passed and things have changed slightly in the meantime.
I succesfully built and tested the small prototipe board. The circuit works exactly as intnded. Perfect, Right? Not really...
Unfortunately, it was only once the circuit was running, that I realized it wasn't what I actually needed. The circuit I need has to accomplish a slightly different
task than the one I initially designed with your help.
The balancer doesn't have to dump energy while the dc voltage is higher than a set voltage since the total 540V are regulated by the main power supply.
What I need instead is a way to keep the virtual "0"(middle Voltage of the DC bus) from drifting.


--- Quote from: eliafavero on April 09, 2019, 05:44:07 am ---I think i haven't done a good job at explaining the situation.
the AC/DC converter actually generates a DC Voltage of 540V which is then divided in two with capacitors in series. this gives me virtually 2 dc lines: 270-->"0" and "0"--> (-270). I then have several motor units on the "High"dc bus and the same number on the "low" dc bus. in normal operation every motor has the same mechanical, and therefore electrical, load. this keeps the bus and the "0" stable. so far so good.
it can occour that one or several motors need to stop for a while. This causes an unbalanced load and the vitrual "0" drifts. which leads the "high" dc bus to go up to 350V and the "low" one to go as low as 190V. Is there some sort of circuit used for this applications? does anyone have an idea on how to adress the problem or which approach should i take?

--- End quote ---

I searched for such a solution and i found the following paper: Active Voltage Balancing of DC-Link electrolytic capacitors.

https://www.pes-publications.ee.ethz.ch/uploads/tx_ethpublications/ertl_IET-PE_2008_.pdf

The proposed topology (attachment "DC_Voltage_Balancing.png") in the paper is not exactly designed for my application. I substituted the BJTs with two pairs of MOSFETSs and added a breaking resistor in series (see attachment"DC_Voltage_Balancing_modified"). I tested this design and it works. The only thing is that I don't like the way it works. On the first design the voltage would reach a certain level, switch the MOSFET On and all the energy would be dumped by the Resistor. With the second design 30- 60% of the power is dissipated by the "Switching" element, which is not actually switching. In fact the mosfet mostly works in the linear region. While there is (almost) nothing wrong with that, i still feel unsatisfied by the design, because that's not what MOSFETs are meant to do or be used for.

Can anyone propose some improvements or even alternatives to this design?

Elia

Marco:
The capacitors will be automatically balanced as the one with excessive voltage is bled, the one with a lower voltage will get charged from the rails and the virtual 0 gets pulled back to mid voltage. You just need one circuit for each capacitor.

You could create a version of the circuit I proposed which is more elegant that just copying it, but still dumps the energy into a resistor rather than the MOSFET. Leaving that up to you.

jbb:
Ah, I think I see what happened, eliafavero.  Did you just build one voltage clamp that went across the whole (-270V) <= (+270V) rail?  That wouldn't help with balancing.

I was thinking you could build 2 voltage clamps, i.e. one for the lower half (-270V) <= (0V) and the other for the upper half (0V) <= (+270V).

The ETH paper has a couple of nice features: few components and (hopefully) good tracking.  They do good work at ETH (especially papers by Kolar). However, it will dissipate power in the transistors.  This is fine if it's about balancing off capacitor leakage currents, which are quite low.  If you have an unbalanced load, you could wind up dissipating a lot of power in the balancing transistor.

So I'm going to ask very clearly: are ALL loads connected across the full (-270V) <= (+270V) supply, with NO connection to the (0V) mid-point?

I suggest you answer in the form of a sketch of your complete system, showing (as basic blocks) what loads are connected and where.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod