[SOLVED] Isolated Feedback for High Voltage Charger

Hi All,

I am looking for a very reliable design using isolation barrier for voltage feedback on a industrial battery charger.
Normal operating voltage to monitor ranges from 120 V (Battery low) to 180 V (battery fully charged)

I guess that I do not need any sophisticated compensation on the optocoupler because of the stability of the load.
Architecture of the charger is classic:

230 VAC Mains-> Rectifier and Caps -> 325 VDC bus -> 20 KHz IGBT Full Bridge -> HF Trafo -> Secondary rectifier -> Output Filter
PWM chip should be a TL494. Control strategy is Voltage mode, with a external overcurrent trigger.

Feedback excursion from an opto output should vary from 1 to 2 volts range when output varies from 120 to 180 V.
What do you think of this first draft:



I Have also included the LTSpice model.
Thanks for your help.

I think that the spread of the characteristics of optocoupler and temperature instability will lead to the need to customize the scheme and its complexity.
In the standard libraries LTSpice not taken into account the temperature characteristics of optocouplers. I changed a few standard library (not sure that my changes accurately match the real characteristics)


I too include the LTSpice model.

-UPDATE
* Copyright © Linear Technology Corp. 1998, 1999, 2000. All rights reserved.
*
.subckt 4N25 1 2 3 4 5
R1 N003 2 2
D1 1 N003 LD
G1 3 5 N003 2 .876m
C1 1 2 18p
Q1 3 5 4 [4] NP
.model LD D(Is=1e-20 Cjo=18p EG=1.11 XTI=3.0)
.model NP NPN(Bf=610 XTI=3 XTB=1.5 Vaf=140 Ikf=15m Rc=1 Cjc=19p Cje=7p Cjs=7p C2=1e-15)
.ends 4N25
Given by **broken link removed**

Do you have an idea for a better design ?

Try this

A reasonable controller design operates the TL494 with PI error amplifier and thus doesn't depend on opto coupler linearity for the static performance. That's how the industry standard TL431 opto isolated feedback works. Opto coupler transfer gain and gain non-linearity still matters for stable loop closure, but as said, this isn't a problem for a battery charger.

The Z-diode circuit in the initial post suffers from type variation and temperature dependency and should be replaced by a TL431 or similar.

Thanks to all of you.

@pplus, I will change the original schematic to use a TL431 instead of the Z-Diode.
@FvM, do you have an idea of how many milliamps are drawn on the V-iso line ?
My initial intention was a self-powered solution on the high-voltage side but your proposal looks also interesting.

FvM said:

A reasonable controller design operates the TL494 with PI error amplifier and thus doesn't depend on opto coupler linearity for the static performance. That's how the industry standard TL431 opto isolated feedback works. Opto coupler transfer gain and gain non-linearity still matters for stable loop closure, but as said, this isn't a problem for a battery charger.

The Z-diode circuit in the initial post suffers from type variation and temperature dependency and should be replaced by a TL431 or similar.

Click to expand...

I know as work couple TL494 and TL431 in SMPS. As i think, problem is different - measure battery voltage, no save battery voltage constant, because the charge discharge cycle requires knowledge of current battery parameters. Correct me if I'm wrong.

Modern battery chargers are able to provide the optimum charge for your leisure battery, through the use of intelligent multi-stage charging.
**broken link removed**
1. Bulk Charge - The full output current is provided to the battery and the voltage is allowed to increase until it reaches a pre-set level. This stage restores the majority of the batteries charge.
2. Absorption Stage - The voltage is held at constant voltage of 14.4 volts until the current flow into the battery falls below 2 amps.
3. Float Stage - The voltage is held at a constant level of 13.8 volts to help keep the battery topped up.

Battery voltage measurement hasn't been required as far as I see, otherwise we need to specify accuracy and resolution. A linear opto isolator could be a reasonable option in this case.

Battery voltage measurement hasn't been required as far as I see, otherwise we need to specify accuracy and resolution. A linear opto isolator could be a reasonable option in this case.

Click to expand...

andrelux said:

Control strategy is Voltage mode, with a external overcurrent trigger.

Click to expand...

For this case the right, but

andrelux said:

Feedback excursion from an opto output should vary from 1 to 2 volts range when output varies from 120 to 180 V.

Click to expand...

Some precisions, guys:

The need is to measure output voltage of the charger which can vary from 120 to 180 volts.
Yes, I use battery cells in this voltage range. Of course, for a constant current, voltage may increase during the charging process.
This voltage will be fed back to the error amplifier of the PWM chip with the appropriate ratio, through the optocoupler.
The PWM chip will maintain a constant voltage at the output, assuming that the reference voltage of the PWM chip will be kept constant.

In this application, I will have a MCU whose main job is to generate a reference voltage that will be used by the TL494 PWM chip.
Output current is also read by the MCU, in order to generate the appropriate voltage reference.

I do not want the MCU to be part of the PWM voltage control mode. I have used the TL494 for smaller projects and I trust it as a good SMPS chip.
I must also admit that I am not fluent :sad: with a full digital PWM control relying only on a MCU.

The other reason is that I do no want to be "tied" for 20 years with a MCU manufacturer when using full digital control.
In this application, the keyword is reliability (This is one of the reasons I have chosen voltage mode vs current mode on the PWM chip)
The MCU acts only as a supervisor for generating the voltage reference and also displaying output current and output voltage.

Thanks for your feed back (with optocoupler;-)

A lower part count solution for the intended function would be to use digitally isolated I and V measurement (e.g. SPI or SD-bitstream interface) on the secondary and utilize the processor as digital PWM controller.

But a classical analog design with linear optocouplers or e.g. an isolated DC current sensor can work as well. It's surely more easily to debug.

@FvM Here is the version using a TL431 instead of Z-diode



Results looks not so bad:



Here is the LTSpice file

@pplus= I am interested by the temperature simulation you did with the previous model. How can I predict the temperature contribution in this one using TL431 ?

@ALL= Do you think I have now a reliable design for picking up the charger output voltage ?

andrelux said:

@pplus= I am interested by the temperature simulation you did with the previous model. How can I predict the temperature contribution in this one using TL431 ?

Click to expand...

Thermal drift of the photodiode and the phototransistor is still not compensated. I neglected temperature drift tl431, considering it insignificant. In the attachment is a modified model CNY17, put it in the LTC \ LTspiceIV \ lib \ sub \.

andrelux said:

@ALL= Do you think I have now a reliable design for picking up the charger output voltage ?

Click to expand...

Why did not you transfer the controller over to the the battery? This would greatly simplify the measurement.

Does it means that power ground could be common with control ground ?
Of course, in this case I need a simple voltage divider. Seems more reliable.

The drawback is the need of a RS232 isolated connection on the CPU. Feaseable.

andrelux said:

Does it means that power ground could be common with control ground ?

Click to expand...

Yes.

andrelux said:

The drawback is the need of a RS232 isolated connection on the CPU. Feaseable.

Click to expand...

I mean its it a good practice. This is part of APC UPS for example.

Here is the version using a TL431 instead of Z-diode

Click to expand...

It doesn't consider TL431 maximum ratings. TL431 works effectively as a threshold switch in this circuit, this isn't of any use if you want to setup a continuous transfer characteristic for voltage measurement.

Thanks for the comment. How should I change my schematic to make the TL431 allowing a "continuous transfer characteristic" ?
If possible, I would like to stick to the "self-powered" concept on the output side.

I think the self-powered discussion is pointless unless you clearly specify the intended analog coupler operation mode, transfer an error signal only (usual operation mode of self-powered TL431 circuits) or transfer a linearized, reproducable scaled voltage measurement.

In the latter case, something like the circuit suggested in post #4 will be needed. It might be still self-powered if you don't rely on measurements below downto zero output.

Thanks FvM for this input. Now I have all the elements to take decision about final architecture for voltage feedback.

I would like to thank again FvM and pplus for their valuable contribution. Now, it is time to close this thread.