BQ24650EVM-639: BQ24650EVM-639

Hello,

For L1, what input and output voltage do you intend to operate at? Section 9.2.2.1 of the datasheet details an equation for the inductor ripple which will affect the peak inductor current as well as the general inductor selection. Please make sure you have a good margin between the DC current through the inductor and the inductor current rating. Please ensure a good margin on the peak inductor current and the saturation current. In our EVM with a maximum 28 V input we have a DC current of 2 A designed with a ripple of 1.167 A (or 2.583 A peak). In our case our peak current did not exceed the rated current. I would consider giving more headroom between the maximum charge current and the rated current.

For R6, if you happen to exceed 5 A during operation you could exceed the current rating of the sense resistor. I would consider going with a higher rated sense resistor if you can.

You just need to replace the above mentioned items and follow the user guide and you should be good to go.

Regards,

Mike Emanuel

Hello,

I appreciate your comments.

In the test we will use a Solar Array Simulator set to a PV with 30W or 60W (Imp = 3A max).

The PV open voltage will be 22V and Vmp = 17.20V.

We will use a Battery Simulator set to end of charge of 14.60V (LiFePO4 pack). 

These are the conditions under which the MPPT solar charger will work in a real application.

Using the spreadsheet provided by TI we obtained a ripple of 1.20A, so it seems to us that the chosen Inductor with 4.5A maximum current and 8.0A saturation current will be a good choice.

On the other hand the Inductor has the same footprint which will be advantageous for the replacement of the demo board.

Under the above conditions, do you have any other suggestions on how to perform the test?

regards

Jose

Hello Mike,

I appreciate your comments.

We tested the demo board, without any changes, and verified correct operation as stated in the manual.
We noticed good thermal performance of the circuit, as well as a very high efficiency.

Already the accuracy at setting the Vmp voltage and the battery end-of-charge voltage was not as high as we have achieved in other MPPT chargers we have designed in the past.

Please note in our circuit we will use an Inductor with a maximum current of 6.64A and saturation current of 8.68A.
However it is of larger size and we will not be able to solder it on the demo board.
We will therefore use the inductor that we mentioned to pay special attention when the current approaches the maximum current.

We will divide the tests into two phases, the first keeping the demo board as is, using microcontroller control.
In the second phase we will then replace the Inductor and sensing resistor to obtain the 4A maximum current.

We have two questions for which we seek some help before proceeding with testing.

- It makes sense to use 0.5% or even 0.1% resistors to set the battery and solar panel voltage. In other words, if we use resistors of this tolerance, will the internal accuracy of the BQ24650 be compatible with these values ?

- Can we add a current sensor to the terminals of the sensing resistor of the BQ24650, in order to measure the charging current of the battery, without affecting the performance of the circuit ?

regards

Jose

Hello Mike,

I appreciate your response, so let's use precision resistors.

As for the current measurement, my question was not clear enough about what we intend to do.

What we need is a regular charging current measurement during the normal operation of the system, not just during testing.

Our current system makes real-time measurements of the current supplied by the solar panel, and the charging current delivered to the battery, so that we have real-time calculations about what is going on. We are currently using a component other than the BQ24650.

What we want to do in the new project with the BQ24650, is exactly the same measurement, for example by placing a High-Side Current Shunt Monitor connected to the sensing resistor of the BQ24650.

Hence our question about whether this connection will affect the performance of the BQ24650.

regards

Jose

Hello Mike,

I appreciate your answer.

Does that mean that it is not appropriate to put the current sensor to the terminals of the sensing resistor of the BQ24650 ?

Hello Mike,

We are using  Diodes ZXCT1082 with 1.7 uA (typ) and 5 uA (max) current in Sensing pins.

We use this component to measure the solar panel current, and the battery charging current in the project currently being manufactured.

If you suggest us a similar one from TI, and one that does not lead to a price increase in manufacturing, we are willing to change.

regards

Jose

Hello Mike,

We will then evaluate which type of component is the most suitable for our application.

We have some strange behavior on the Demo board that I will now explain.

We put a SAS with 22Voc and 500mA, and a battery simulator with 13,20V limited to 4A.

When we activate the SAS the BQ24650 sets the voltage to 17.17V (R17 = 133K and R19 = 10K) and the current to 500mA.

So far so good.

Then we applied a 5V signal to the CE test point to shutdown the BQ24650, and found that the shutdown did not happen, and the SAS indicated 5.43V and the same 500mA.

We measured MPPTSET and it had 378mV, and therefore above the 75mV to shutdown the BQ24650.

As the Vgs(th) of the 2N7002 is 2.5Vmax, we understood the 5V signal would have to activate the FET.

Do you have any explanation for the reported fact?

Of course, if we apply Gnd directly to the MPPTSET test point, the BQ24650 will shutdown.

regards

Jose

Hello Mike,

We found we need to remove jumper JP4 to work as expected.

regards

Jose

Hello Mike,
We are finishing testing with the Q24650EVM board.
In one of the tests we applied a signal to the CE test point to control the shutdown of the BQ24650 (CH3 - Red).

When the control signal at the CE test point drops from High to Low, we found that the BQ24650 takes about 1.5 seconds to set the MPPT voltage (CH1 - Vin - Yellow). Delta X = 1.54s
You can see in the figure the voltage drops from 22V to the 17V MPPT (CH1 - Yellow)
When the MPPT voltage is set, the charge starts and we have current in the battery (CH2 - Blue) during the time that the BQ24650 remains active.
Everything works as expected, we just don't understand why after Shutdown is removed the BQ24650 takes about 1.5 seconds to set the MPPT voltage and start charging.
In the datasheet we did not find any parameter that indicates a time of this magnitude.

What are your comments to what we reported above ?

regards
Jose

Hello Mike,

Hello Mike,

I finished my post without asking a question that I had in mind.

What I described is what is indicated in figure 2 on page 10 of the BQ24650 manual?

regards

Jose

Hi Jose,

I will get back to you by Wednesday. 

Regards,

Mike Emanuel

Hello Mike,

I appreciate your response, so we would like to know if there is any way to reduce that time, or is it an intrinsic characteristic of the BQ24650 ?

We have a last question:

As for the TERM_EN pin, it can be connected to either Gnd or Vref (3.3V) depending on the desired function.
As our microcontroller runs at 3.3V, I am asking if we can use an I/O of the microcontroller to command the TERM_EN I/O selecting the function by software (via microcontroller).

If it is possible do you recommend that we use a pull-down or pull-up resistor on the TERM_EN pin to power-on the microcontroller and before it takes over the I/O's ?

If the solution of controlling directly to the TERM_EN pin via a microcontroller I/O is not recommended, what approach would you suggest, in order to be able to select the function via the microcontroller ?

After these questions I think we have finished all the tests, and we will go ahead with the design of our circuit.

regards
Jose

Hello Mike,

I appreciate your comments, and we will follow your suggestion regarding TERM_EN.

Regarding your curiosity, you answered it with your question. As you say, since each customer chooses a mode of operation, our charger is programmable so the customer chooses what they want.

We have already finished testing the BQ24650EVM-639, and the operation and performance is excellent and as expected.

We will now change it to a charge current of 4A.

However we have two questions that we don't have answers to yet, and your comments may help.

We are actually developing two projects, the first being an MPPT charger.
In this project we will follow the BQ24650EVM-639 layout for our PCB as far as the BQ24650 zone is concerned.
We found that the demo board has some components on the bottom side, which for us is complex in terms of manufacturing.
Generally speaking, do you think it will be problematic to put those components on the top side (continuing with a 4 layer board) ?

The second project is more complex, and features an MPPT charger with the BQ24650, and an electronic controller for controlling various devices where there are current peaks on the order of 3A at 12Vdc.

The controller part has two 150uF capacitors in parallel at the power input, where the output of the BQ24650 and the battery is connected.

This system works 24H/7D in remote locations, outdoors, without maintenance or access to it, and the charger circuit with the BQ24650 is always connected to the battery, and always connected to the electronic controller I mentioned.

What concerns us is the fact that the BQ24650 'sees' these capacitors 2 capacitors causing changes in the 12KHz to 17KHz rule stated in the manual.

We have run tests and found no malfunctions, but we are concerned about what might happen in a manufacturing run after several units have been installed.

Do you have any suggestions for this type of application I mentioned ?

regards

Jose

Hello MIke,

I appreciate your valuable comments.

We were doing the last test with the board BQ24650EVM-639, and we damaged it.

We soldered two 2K2 resistors to the terminals of the sensing resistor R6, and a current sensor to these resistors.

We don't know what happened, but the board stopped charging the battery.

Although the STAT1 led remained ON, the charge was intermittent, that is, it measured the charge current in the battery simulator, and then it went to zero. And it stayed like that forever ... an intermittent charge and with very low current. 

We checked the signal at BTST pin, we found the expected signal (when there was charge),  and a damped signal (when there was no charge)

Charge

End charge, but led STAT1 still ON

We thought we had damaged the BQ24650 at the SRP and SRN terminals, so we removed it, and soldered a new one.

The surprising thing is that after testing with the new BQ24650 we still had exactly the same problem.

We checked the diode which is in good condition, as well as the Mosfets.

There are no shorted capacitors, so we believe the remaining components are in good condition.

What do you think might have happened, and how can we get the board working again ?

regards

Jose

Hello Jose,

Can you please provide a schematic diagram of the modifications made?

Can you please provide a waveform of the input voltage, PH voltage, battery voltage, and charge current on the same plot?

What is your input voltage and your output voltage when this is happening?

Regards,

Mike Emanuel

Hello Mike,

Helo Mike,

I can send you the schematic drawing with the changes made if you give me an e-mail address.

The input voltage was always 22V (max.) with the voltage dropping as a function of current, and the board sets the MPPT voltage at 17.18V which is the Vmp voltage of our solar panel. Please note that we used a Solar Array SImulator for the tests.

We will prepare the waveform you ask for, being that we have no way to show the load current on the oscilloscope. We will try to show the voltage at the terminals of the sensing resistor with a high impedance probe. Let's see if it doesn't change the behavior of the circuit.

I will post the waveform as soon as it is available

regards

Jose

Hello Mike,

Please find below some waveforms of the failed board (Vin active)

1 - Using a Battery Simulator

2 - Using an Electronic Load (CV @ 13.20V)

3 - Using a real battery pack 

If you think it is important we can solder some terminals on Rsense and get the waveform.

regards

Jose

Hello Jose,

Can you please report the status of both of the STATx pins?

Please friend me on E2E so you can share the schematic.

What happens if you remove the high side current sense? Does the part work then?

I would like to see the current waveform if possible.

Why does the input voltage keep dipping?

Regards,

Mike Emanuel

Hello Mike,

I will remind you what happened to avoid an incorrect analysis on your part.

1 - We changed the BQ24650EVM-639 board to maximum load current of 4A, and MPPT voltage of about 17V. We ran a wide range of tests in that configuration, and everything worked as expected.

2 - To measure the battery charge current, we added a current sensor to the sensing resistor R6 (10 milliohm).

3 - The current sensor was made with the ZXCT1082 and two 2K2 resistors, one on each side of the sensing resistor R6.
Note that we have used the ZXCT1082 to measure current elsewhere in the circuit, such as the solar panel current, and have experience with the component that guarantees good measurement accuracy.

4 - To avoid measurement errors we soldered the two 2K2 resistors (in 0805 format) directly to the pins of the sensing resistor, and the ZXCT1082 over the two resistors.

5 - When we connected the board to make the first measurement, the problems we described started.We thought we had damaged the SRP or SRN pin probably due to some unintended contact between the ZXCT1082 circuitry and one of those pins.
We changed the circuit, and switched to feeding the ZXCT1082 with Vref from the BQ24650 and not with input voltage (SPR pin) as in first test, but it had no change, the problem remained. So, we decided to remove the whole circuit with the ZXCT1082.

6 - After removing the circuit, and to our great surprise, the problems continued and were identical to when we had the circuit with the ZXCT1082 connected to the sensing resistor.
Please note we have already tested the circuit with the ZXCT1082 disconnected from the BQ24650EVM-639 board, and it works fine, after all it is only one circuit with 4 components ...

7 - We thought we had damaged the SRP or SRN pin probably due to some unintended contact between

8 - We removed the BQ24650, and soldered a new one to the board. When we went to test the board, it still had exactly the same problem. Note that we never again resolder the circuit with the ZXCT1082.

9 - We have verified that the Mosfets are not damaged, nor the diode and there are no shorted capacitors, so we cannot understand what is going on to have the same behaviour with the new BQ24650.

Please note all the waveforms we sent you are from the board without the ZXCT108. The first ones still with the original BQ24650, and the second ones with the new BQ24650 - the behavior is the same.

As for your questions:

- the STAT1 signal goes to LOW (led on) when Vin is put in the circuit, and therefore as in a normal charge.
The STAT2 signal doesn't present any changes and stays at High.
We checked both signals with the oscilloscope and there is no noise or any pulses, perfectly stable.

- As I said above all tests of the reported problem were performed without any circuitry added to the BQ24650EVM-639 board, i.e. the board does not work and is damaged after the initial test with ZXCT1082.

- What it looks like to us in the input voltage variation is that the BQ24650 tries to fix the MPPT voltage and fails to maintain it.
Notice that the channel 1 waveform (yellow) voltage drops from 22V of Vin to about 17V of MPPT voltage and then return to 22V.

We can't figure out what is going on, and in the meantime we have already bought a new BQ24650EVM-639 board, but we are afraid of damaging it again if we connect the circuit in question - the board is too expensive to take those risks again.
So we wanted to understand what damaged the current board so as not to make the same mistake.

Since we have the whole project stalled due to this 'small' detail, we have to make decisions on how we are going to measure the battery charging current.

We are left with adding another sensing resistor in the battery to measure the current, it is not the most efficient way to do this, because of the impact it may have on the efficiency of the charging circuit. However it will be an option to consider.

We will await your comments to make a decision.

regards
Jose

Hello Mike,

If we have already replaced the BQ24650, and the diode and Mosfets are not damaged, what else could it be ?

With the description of the malfunction I made earlier, what is your suggestion for us to try to repair the board ?

regards

Jose

Hello Mike,

When you say the 'board is damaged' do you mean the PCB ? If not, can you be more specific ?

If we don't want to use the TS pin function, and the TS pin function is disabled forever, what should we do ? 

regards

Jose