[SOLVED] 220V Battery Charger Design Help

[vishweshgm]

Hello,

I am trying to design 220V DC Battery charger. I have little experience in power electronics, but I have earlier successfully designed 112V,10A battery charger using SCR. The SCR charger took 140V,50Hz AC from transformer as iput and using half controlled (diode+scr) rectifier bridge and a capacitor delivered up to 135V 10A Current without any issues (Here is the simple schematic of working unit. Gate trigger is done using micrcontroller+pulse transformer driving circuit.

With that confidence I am now trying to take up 220V, 60A project. 230V,50Hz rms AC will be input Considering huge current, I have to go for full controlled bridge rectifier for sure. Also there is no room to add capacitor after rectification, as I feel it will be very huge value and I cannot add 100A rated inductor to manage capacitor charging load (cost issue). Here is my planned simple schematic:

I have not finalized parts to be used yet. But since the project is huge for my experience I am seeking help (1.)if its good idea to make 220V,60A DC Charger with simple topology shown above. I also intend to use a microcontroller based triggerring circuit (same as I did earlier) and this requires me to have a common ground (This is unavoidable, else efficiency of feedback voltage loop (PID control) decreases). (2.) Is it okay to have common ground when handling this much huge current (I am asking w.r.t functioning of the device ,not safety).
I am currently planning schematic, will post once its done.

 

[FvM]

Not clear what you mean with common ground related to this circuit. SCR gate voltage is referenced to cathode, at least two of the SCR need isolated drivers. The circuit can however work with two SCR as well.

Charging a battery without filter inductor causes high peak currents and doesn't comply with power quality standards.

 

[vishweshgm]

Hi Fvm

Not clear what you mean with common ground related to this circuit.

[Vishwesh] : I meant, zero voltage reference point of rectified AC and ground of power supply given to microcontroller are common. In some circuits I have seen micrcontroller ground is galvanically isolated from rectified AC ground. They have used linear opto like LOC110 to measure the feedback voltage.

at least two of the SCR need isolated drivers.

[Vishwesh] : Yes,indeed. I plan to use 2x 1:1:1 pulse transformer to trigger SCR gate. So it is isolated driver as shown below

Charging a battery without filter inductor causes high peak currents and doesn't comply with power quality standards.

[Vishwesh] : This is the part exactly I am afraid of. Using inductors rated as high as 60A, not only addup cost but also will be very bulky causing my product to be huge. Can you help me understand what problems I might face if I don't use inductor and what its work around might be?

One way I have seen is by adding AC capacitors parallel at input itself along with a relay in-series. This way capacitors can be introduced/detached from circuit by turning off/on the relay. I am still studying this method. But my point is I am looking for ways to handle this via software as I am good at it and want to eliminate bulky inductor.

 

[cupoftea]

I am sure you know that a battery load is not like a resistor load.
A battery is basically a voltage source, so you need to limit the current, or better still, control the current into the load (battery). I presume you are doing that with the controlled rectifiers.

You could just put six 2.2kw PFC'd offline battery chargers in parallel...they would also give you isolation at the output...and be a lot lighter in weight overall...if you dont need isolation, then you could put six 2.2kW boost PFC's in parallel.....and regulate the current of each.

The power you describe, when done with a single converter, usually uses a 3 phase input supply...and a simple 3 phase rectifier will give you a pretty good power factor.

I meant, zero voltage reference point of rectified AC and ground of power supply given to microcontroller are common. In some circuits I have seen micrcontroller ground is galvanically isolated from rectified AC ground. They have used linear opto like LOC110 to measure the feedback voltage.

Put the microcontroller ground wherever you like....rectified DC negative is fine....as long as you isolate the necessary signals and drive waveforms. You can even use say a little SMPS to give you an isolated ground point....then you can earth this ground and connect comms cables that have an earthed ground to it.

 

[Easy peasy]

A half controlled bridge is all you need for battery charging or resistive load for heating only - consider 8 - 12% leakage on the main Tx - this will give much less peaky currents in the Tx and the semi's, resulting in much lower rms heating of the Tx ! and the semis ..! - and the output and mains input wires too ...!
This is a very industry standard way of achieving what you want.

 

[vishweshgm]

A half controlled bridge is all you need for battery charging or resistive load for heating only - consider 8 - 12% leakage on the main Tx - this will give much less peaky currents in the Tx and the semi's, resulting in much lower rms heating of the Tx ! and the semis ..! - and the output and mains input wires too ...!
This is a very industry standard way of achieving what you want.

Hi,
I took advise and have designed half controlled bridge. Today I ran prototype test with 50V, 20A battery pack.
It worked fine.

But I need all of you guys expert opinion to move further.
I spoke with industry vendors and they told battery chargers have trickle charge and floating charge and boost charge.

What I did was trickle charge, where I was giving pulses of incoming rectified sine wave like this

They asked if I can provide floating charge. This is used to charge a dead battery too.
So what approach can be used to provide floating voltage?

One approach is adding capacitor after bridge. But how do I decide the value? (mind you .. its maximum capacity required 220V,60A)

 

[Easy peasy]

If you want to provide a flat steady DC trickle charge, or float charge to the Pb-acid batteries - there are two options - 1) use an 8% leakage transformer and put Cout = 4,700uF per amp of full load current.

or 2) for a standard transformer, use Lout = 1mH, followed by 4700 - 10,000uF of caps to smooth the DC current and voltage.

 

[vishweshgm]

Hello,

use Lout = 1mH, followed by 4700 - 10,000uF of caps to smooth the DC current and voltage.

• Can you please explain me how you arrived Lout = 1mH (please provide any link that I can refer and understand)?

• Also, during normal working (i.e. without Cout or Lout), above 20-25A, There is very sharp peaks that I am unable to measure via microcontroller adc. I understand without inductive filter peak currents would occur. But an old system that my customer has doesnot have inductor, yet able to provide and control constant current requirement. So i reckon there must be some way it can be done.

From my HallCurrent Sensor , I read adc like this

Due to peaky currents, this circuit can only hold until a point. However peaky current is not the problem here, but its measurement. after above filter circuit output at 20A is like this.

If I put a clampmeter, the 20A DC current is constant. Basically I need to measure peaks and map it to dc current. If anyone has idea on these please do share. I can sample adc at 10kHz.

 

[KlausST]

Hi,

There is very sharp peaks that I am unable to measure via microcontroller adc.

What do you mean with "sharp peaks"?
These cones in the scope picture? I don´t consder tham as sharp peaks and they can easily be measured with a mircocontroller ADC.

yet able to provide and control constant current requirement. So i reckon there must be some way it can be done.

How do you know how "clean" the constant current is?
And for sure you can builld a transformer in front of the SCRs with high series inductance to act like a smoothing inductor.

Hall test circuit:
Basically there is a 159Hz filter followed by a 1590 Hz filter. The second is almost useless.
your charging signal is like in the scope, repeating with 100Hz.

So with a 159Hz filter you suppress the 100Hz just a little and you suppress higher frequencies.
Are you satisfied with this, or what do you really want to achive?

If you want to get a smooth DC output (= averaging the peaky shape) then you need a much lower cutoff frequency and a higher order. Maybe a second order 3Hz filter. But for sure it will slow down reaction for a regulation loop, maybe causing instability.

You may also use software filtering.

Klaus

 

[vishweshgm]

Just for completeness, I am talking about lead-acid batteries in this whole thread.

What do you mean with "sharp peaks"?

Thanks a lot for all the inputs. Please give me time till monday, where I get batteries again for testing. I'll try to capture more screenshots and scenario of my issue.

How do you know how "clean" the constant current is?

I only measured current using clampmeter and this old device maintained the current set by user in its display. My customer was showing it to me as an example when I told him we cannot do it without inductor.

For eg. In this machine I set 20A current, press start and that's all. It starts increasing voltage from 0 and triggers SCR from 180,179,178,...90,89,88..degrees (every rectified halfcycle)--(until current setpoint is reached)--(I hope you get what I am trying to say). In clampmeter It showed between 19A to 20A, but never crossed it. (I saw voltge across two end terminals of series (90x2V cell), it was 197V and increasing...)

As far as "clean" is concerned, unfortunately I don't afford a current probe. So can't tell for sure. But at-least I want to achieve this design first.

And for sure you can builld a transformer in front of the SCRs with high series inductance to act like a smoothing inductor.

Yes. We have a transformer in-front of SCR already, which takes 3 phase input and gives out 230Vrms output.

So with a 159Hz filter you suppress the 100Hz just a little and you suppress higher frequencies.
Are you satisfied with this, or what do you really want to achive?

I have put filter just to avoid noise if any. The filter frequency values are not designed carefully, I just put a low R&C values. Reason is that, there is a very low chance that I am going to see high frequency noise , as there is no high speed switching involved. Just a 50Hz Rectified AC being delivered to battery without any filters at output. That's all. I admit I may not a good designer. But I try to learn from mistakes.

 

[KlausST]

Hi,

Just a 50Hz Rectified AC being delivered to battery without any filters at output. That's all.

Again my question to this:
Do you want to see the 100Hz ripple or do you want to see the flat (average) of the charging current? (= a DC line)

Klaus

 

[Easy peasy]

you can control the AVERAGE current quite nicely - that fact that it is made up of sharp peaks that cannot be seen with a simple clamp on meter seems to have escaped your notice ...

--- Updated Mar 10, 2022 ---

an averaging ckt on the current sense signal will give an average value - this is what a slow error amp effectively does to the current signal - its integrating action averages the input signal ( and compares it to the ref )

 

[vishweshgm]

Hi @KlausST , @Easy peasy , sorry for late reply. I wanted to study and test more before I discussed further.

Do you want to see the 100Hz ripple or do you want to see the flat (average) of the charging current? (= a DC line)

On Mar10,2022, The answer I had for this question is I want to see 100Hz ripple on the battery (I mean if I put scope probes between terminals of battery, I want to see rectified AC ripple). But I did not want to see the same ripple in my feedback circuit, as that would cause issue in my adc read. This is the reason I had C34 in my feedback circuit. I actually replaced C34 with 100uF capacitor in hope that it would act like "peak detector", which is just enough to map & calibrate my adc to amount of Amps being drawn.



So without C34 being replaced 100uF, waveform from Hall Sensor would look like this:




After replacing C34 with 100uF, is the image I posted earlier. I saw flat line here until some15A point and after that peaks above flat line started to appear. That is where I thought it was due to "peak current".




But today Mar14,2022, I got the batteries again for test today and this time I took good-old shunt resistor along with its feedback circuit withme.



Its waveform in same conditions (I think around 20A) as above looked like this:





Here I sampled adc at 10kHz rate and applied averaging technique in SW and it worked to pin point perfection. I could achieve 100,30A today. Next plan is on wednesday to run full 60A.

Sidenote : I was automotive embedded sw engineer for 5years , now working on power electronic products. This is so much fun and frustrating at same time. Just amazing!

So I must re-phrase my question and need your expert opinion,
1. Why the averaging technique works on shunt but not on Hall Sensor (Hall sensor I used was CSNP661 - 90A and given 12V stable input). Or Am I doing something wrong here?
2. Why the Hall sensor waveform different from that of shunt? Notice that Hall sensor has ripples and also has a dc offset. Shunt doesnot have a dc offset it falls to ground. So might there be some technique should I use? If some one has experience on SW implementation side please share.

 

[KlausST]

Hi,

(I mean if I put scope probes between terminals of battery, I want to see rectified AC ripple

We did neither talk about a scope nor about the voltage.
We talked about your current measurement circuit.

Klaus

With C34 directly at the sensor output (not in the feedback as you wrote) you most probably overload the sensor and risk oscillation.
--> A C alone is no low pass filter, it needs an R or an L in series.
Also for tau calculation or cutoff freqyency calculation you need the R or L.

Klaus

 

[FvM]

Current sensor CSNP661-90A has current, not voltage output. It doesn't work in the circuits you have posted so far, missing load resistor Rm. Please clarify about the sensor connection.

 

[KlausST]

Hi,

Current sensor CSNP661-90A has current, not voltage output.

I have to admit I didn´t read the datasheet until now.
I guessed it was a voltage output.

Thanks for clarifying.

Klaus

 

[vishweshgm]

Current sensor CSNP661-90A has current, not voltage output. It doesn't work in the circuits you have posted so far, missing load resistor Rm. Please clarify about the sensor connection.

I blundered here big-time. I didnot know hall sensor would give current as output. Thanks a lot. I'll rectify the mistake. Thanks again.