Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

LTC4020 not regulating, blowing sense resistors during debug

Status
Not open for further replies.

MARGD

Newbie level 4
Joined
Aug 14, 2019
Messages
5
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
88
All,

I'm currently working on a design that's trying to use the LTC4020 battery charger, and I'm at a loss to understand what I'm doing wrong. Here's my schematic.

4020 schematic.PNG

When I apply 48V to Vin, I get between 24V and 38V on Vout, and nothing on the battery terminal (with no battery connected) and it's very unstable.

When I probe at the gates for the four FETS, here's what I see:

M1 Gate.PNG
M2 Gate.PNG
M3 Gate.PNG
M4 Gate.PNG

Which is clearly wrong. At first I thought it was because I mistakenly set my Vfloat to 57V, so I corrected that (Vfloat was reduced to 54V and Vmax was set to 55V) but the problem remained. It's not a damage issue: First, Linear's application engineer assured me that there would be no damage to the part unless my input exceeded 60V, which it did not and second I tested the updated values on an untouched board and it still didn't work.

If I apply 46V instead of 48V to the updated 54/55 circuit I get results that are closer to what I expect (52V out on an open circuit), but still unstable and the gate control lines do the same as the above screenshots.

I've been in touch with Linear Tech support and they recommended changing the circuit to decrease my Vshutdown threshold (set for 32V by changing R5 to 385kohms) and change my switching frequency (R8 = 47.5kOhm) and reduce my charge current to 400mA by changing R11 to 125mOhms.

I still don't get the correct output. I tried connecting to my battery, which was discharged to an open circuit voltage of 52V. When I connected that to my circuit, and then applied 48V (current limited to 700mA) to VIN, the circuit popped and my sense resistor was blown open like a fuse.

I'm currently waiting for replacement sense resistors to see if I damaged the LTC4020, which I expect is the case.

The thing I don't get is: Why isn't the circuit performing? I followed the datasheet for all calculations. Linear's tech support found no significant differences between my schematic and the application examples (in fact I mostly just copied the typical design example from page 38 of the datasheet, subbing component values based on the application section guide) The layout has one needed correction that I noticed after it was already fabbed (my high-side gate lines cross a split in a plane but I'd expect that to give me efficiency reductions, not catastrophic faults.)

I've designed synchronous buck converters before, and never had a problem like this. Clearly I'm missing something, but I have no idea what.
 

Attachments

  • 4020 schematic.PNG
    4020 schematic.PNG
    122.6 KB · Views: 157

Hi,

Did you try to use the I_limit feature?
Try it. Start with very low current ... then increase.

Show us your PCB layout and a detailed photo of the circuit.

Klaus
 

Current sensing is very important - read the data sheet carefully on using caps to filter the signal ...

reading the whole datasheet very carefully will likely reveal the causes - the chips can be damaged fairly easily - esp is a fet goes bang ...
 

I'm not using the Ilimit pin because my understanding was that the inductor sense resistors were supposed to handle that. Also, none of the application examples in the datasheet use it.

I'm not legally allowed to share the layout.
 

I have an update:

I tried modifying the DC2134A evaluation board to match my board more closely using components that I had on-hand

I replaced RCBAT1 with 100mOhms (.5A charge current limit)
I replaced R8 with 360kOhm and R10 with 17.8kOhm (Charge voltage ~53V)
I replaced R9 with 390kOhm and R11 with 20kOhm (Max voltage ~55V)

I could not replace the RCBRT1 or RCTRT1 sense resistors, as I'm waiting for an order.

Everything else on the eval board (the inductor, the 13V minimum for Vshutdown and the soft-start and ILIM capacitors) are unchanged.

I was able to get between 50 and 53V out with no load and no battery. When I connect a battery with an open circuit voltage of 52V it charged with a current of 0.17A, which is expected. The power supply was set to 48V with a current limit of 0.7A, and under those conditions the power supply clamped down to 15V with the max current of 700mA going to the board.

I expect to receive new sense resistors tomorrow, at which point I'll run the experiment to limit the inductor current with 0.075 ohm sense resistors (max inductor current ~0.66A in boost mode)

The primary differences between the eval board and mine are:

1) VC has a 47kOhm resistor in series with a 47pF cap to ground, while my design has 33kOhm resistor in series with a 680pF cap to ground.
2) ITH has a 36kOhm resistor in series with a 6.8nF cap to ground, while my design has ITH connected to CSOUT through a 2nF cap.
3) The decoupling cap between CSP and CSN is 4.7uF, while my design has 0.33uF
4) SensBot is decoupled to ground with 0.033uF versus my 0.013uF
5) My Vshutdown is set to 32V instead of 13V

Oh, and I just found out that the company that supplied my PCB was having quality issues, so I can't be certain that the raw boards are even any good.

So I've got a lot to look at right now. I'll post back with updates when I have them. If anyone can recommend most likely starting points, I'd be much obliged.
 

I've updated the Vshutdown on my board to 14V minimum, set my max V to 55 and my float voltage to 54, using the same values from the modified DC2134A above. It started working, if inefficiently, when I applied 48V to the input, similarly to how the modified evaluation board behaved (The power supply clamped at 700mA with a voltage of 15V)

I was unable to get a current measurement, however, because I turned the power supply off for a few seconds and when I turned it back on the battery charge sense resistor blew. I replaced it, and it looks like the whole board is dead.

I've lost three boards to this. The first one I thought was an error in my connections, and the second I thought was the result of having my Vfloat and Vmax resistors set incorrectly. I have one more board left, and I don't want to power it up until I at least have an idea of where this defect is coming from.
 

you may need to have a power supply that can supply lot more current to keep the ckt working - try adding 1000uF or more, temporarily to the input wires to your board ( right at the board ) - this will supply the switching current peaks and allow your power supply to supply the average current only, once your power supply takes a dive in volts the system will latch up as your board is trying to draw constant power ( i.e. higher amps at lower volts ) so turn up the current limit ...!
 

It looks like there were a couple of issues with the design:

First, the schottky diodes were not sufficiently derated. (60V diodes for a 54V output). I confirmed that the diodes across M2 and M4 were blown short.

Second, the spacing rules for the PCB layout was too close (they were at the minimum spacing for 100V according to IPC.) This appeared to conspire with an assembly issue to cause a soft short between INTVCC and ground.

Finally, The compensation components for ITH and VC pins were apparently incorrectly set. Linear provided a very helpful app note (AN76) that explained the rationale for choosing them. The values on VC and ITH are heavily dependent on my ouput capacitance and ESR, and AN76 explains the relationship fairly well. This is probably what led to the diodes blowing. Unfortunately, the note doesn't provide calculations for what values to use, but rather just says to do what it shows on the datasheet and modify as needed. I'm updating the schematic to match the simulation that Linear's field applications engineer sent me.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top