Gen1 DCDC Converter Troubleshooting and Repair

[kiev]

i've changed my mind on the identification of IC102 and IC203, from dual op amps to dual comparators. i see a common hysteresis design used on these circuits as on IC202 which seems to be a quad comparator. The only dual op amp seems to be IC208, etched and barely marked "1251".

Modified the IC202 comparator drawing to add a common via path that was found linking several transistors and CN201 pin 6 which is the +12V switched input signal.

This common via seems to have a latching feature. If any of the transistor paths are true, then the drain path for the diodes is completed. After it starts then the diode path will be latched as long as the key switched 12V is applied.

What this means:
The DCDC can start and turn ON without a key-switched 12V (for example during EVVSE charging of the pack), but a key will keep it ON (such as while driving). This is the section of control logic that makes this possible.

 

[eldenh]

So the switched 12v supplies several functions when the key turns it on, but can be powered from another source when charging with an EVSE. How is that commanded when there is no connection from CN201 and the EVSE connector?

 

[kiev]

So the switched 12v supplies several functions when the key turns it on, but can be powered from another source when charging with an EVSE. How is that commanded when there is no connection from CN201 and the EVSE connector?

The 12V on pin 6 is the Switched supply from a relay circuit [A-08X] that gets energized by the EV-ECU when it detects the key switch; the EV-ECU can energize that relay at anytime, for example whenever it detects EVVSE also.

But the DCDC will function without that signal altogether as long as there is sufficient HV to create Vsub, TP201 within range, and that there is a 12V on the Output side.

The switched 12V in the DCDC provides a latching action on the third transistor path of the common via. It is also somehow tied in to a signal for the PWM, still under investigation.

 

[kiev]

Added to page 1 schematics post: temperature and current sensor signal conditioning.

My crude little test data is not very useful to determine the cutoff temperature, but the reference voltage set point is at about 3V (3.042) with the Low Threshold at 2.6 and the High at 3.3V. This corresponds to a sensor resistance ranging from about 5k to 10kΩ, but i did not measure an intermediate temperature point to correspond to that range.

At room temperature i measured about 100 to 120kΩ, and you would not want the heatsink to get so hot as to boil coolant in the chill plate, so i would need to measure the slope somewhere between ambient and boiling water temperature. Maybe there is a standard curve for 100kΩ ambient thermistors.

The current sensor is an LEM hall-effect type. The gain of the current sense amplifier is about 1.7 and that signal is summed with some transistor logic that i will put all together on another sheet.

 

[eldenh]

I found a chart for 100K thermistor at https://mikroelectron.com/ProImg/X4/186d503e-bf2a-4279-ac77-9b94ce1f0c80.png. It looks like 10K corresponds to 180 F.

I've been puzzled that I had been reading 5v at the gate of Q203 and only 0.28v at the vias feeding 5v to IC202. Checking Q229, all I find is an open circuit. So I think I have damaged the 5v supply circuit. It seems to me that it should be possible to test the LV section by supplying about 15v to TP201. My ohmmeter shows 3.65K from there to LV ground. Is that likely to be damaging?

 

[kiev]

That temperature seems reasonable to shut off the DCDC to me. So a reading at 308F makes no sense for this device, and my laser IR thermometer didn't read my cold drink temperature very well. i dipped the thermistor into a glass of tequila on ice, so my reading was more likely at 27F rather than 36F.

i use the continuity beeper on my meter and sharply pointed probe tips (sewing needle sharp) to check traces. The fat-holed double vias are either +5, +15 (TP201) or Gnd, and the tiny-holed double vias are signal lines. i use colored sharpie pens to mark vias after i check them so i can quickly find one that i need during a trace.

The closest double via with the fat holes to the source of the +5V is just to the end of IC201 under some silk screen "C225". Once you find that you can trace down toward the LV section, there is a rectangular section near Q215 that has 4 vias at each end and is a big 5V pad. Then over by IC202 at the end near pin 8 is a double fat via near silk "C236" that should also show continuity to the 5V double via near IC201. If it doesn't then i suspect that a shorting event may have blown the 5V trace on one of the embedded layers between the PWM area and the LV section.

The only way i could recommend to connect a 15V source to back-drive the TP201 supply would be if you used a current-limited adjustable power supply that would protect against a short circuit condition. If all you wanted to do was verify the presence or absence of the 5V supply created from the 15V, that might be feasible with even a 12V supply, but it should still be severely current limited. And that assumes that Q229 on the bottom layer is a good part. A better option might be to just remove the IC209 from the bottom layer and test it separately on the bench.

i measured 3.5k from TP201 to ground also. Q229 is NPN and i measure ~0.6V diode drop from base to emitter and base to collector.

 

[eldenh]

When I got past the coating Q229 checked out OK. The vias showed continuity also. So I don't understand why I was reading 0.28v when I checked the vias when the board was powered up.

 

[kiev]

That may have been due to the insulation of the clear coat; one probe was not connected and acted as an antenna to pick up the HV electric field radiation?

 

[eldenh]

That seems to be unlikely to me. As I recall, the negative probe was in one of the screw holes of the chassis where it was for other checks and there was no reading until I twisted the positive probe in the vias to make contact on each leg.

 

[kiev]

Maybe check the volt meter function with a known voltage to make sure it wasn't damaged or has a fuse blown. i hope the 5V supply is not damaged.

There are 3 resistor networks on the bottom layer with a 10k, 10k and 12k [R258, 245, 241A] connected directly to that via but if ceramic capacitors C242 or C249 were shorted to ground then either could pull down that pad.

 

[eldenh]

I read 6.28k and 4.55k across C242 and C249 respectively. I assume those have to do with other current paths in the circuit. I've ordered a power supply with adjustable current limiting to try to test the 5v via TP201.

 

[kiev]

That's a good measurement that the caps are not shorted, i measured 5.7k and 4.5k on mine.

A power supply will be a good tool for this task. It will also let you decouple and adjust the input and output 12V later to detect and check circuit functions when you get it up and running.

 

[kiev]

Posted this schematic in the page 1 section also,

 

[eldenh]

The power supply finally arrived. As I considered how to hook it up I discovered that none of the GND pads on the top of the board are connected to each other or pin 1 or CN201. So it seems I must remount the board in the chassis or create a system to connect the pads. Is it possible that such isolation can be used to investigate some functions of the board in isolation?

 

[kiev]

It may be possible. It appears that the pads are capacitively coupled to the chassis, likely for filtering ac noise. GND1 couples the HV input after the EMI filter and is not needed for testing; GND2 couples the output 12V low side return thru some large capacitors; GND3 couples the LV section gnd plane to chassis thru a single small capacitor; GND4 couples the L201 inductor 12V output return to chassis before the current sensor.

If you wanted to just troubleshoot the LV section of the board without trying to start and run the PWM, then only GND2 would need to be connected to the 12V output return. But there are so many necessary conditions in the logic gates that i'm not sure how useful it would be. The main 12V output power return occurs at the center tap of the T103 transformer secondary, so that is where you need to be careful not to run the PWM without full continuity to the chassis ground for the entire board/system.

But if you just wanted to power up the low side supplies to check Vcc at the chips, etc. then that might be okay without the full-blown GNDs connected.

For example, 15V could be put onto the TP201 pad to measure if the 5V reference is being created by IC209. i would have the meter connected to measure before briefly touching the power supply lead to the TP201. Just a brief touch to see the result and then remove it. And have the 15V supply current very limited to just a few mA.

 

[eldenh]

My results are discouraging. I found 15v on the 5v buss. Q229 still checks as a good transistor, so I assume IC209 is damaged. Without being able to power up the LV section, I don't see a way to continue.

 

[kiev]

It's good that you found a problem with the 5V supply; You can replace the chip and move on to testing the LV logic section.

 

[eldenh]

When I looked for the part number in the list, the best I found was that it is an obsolete part. Where can I find a suitable replacement?

 

[kiev]

It looks like Arrow may have some in stock at $1.76, digikey and mouser showed 0 stock. Maybe i should buy a few also...

https://www.arrow.com/en/products/lt1461dhs8-5pbf/analog-devices


It may be that you could remove the chip and then power up just the 5V buss with your power supply if you wanted to check for a short on the 5V side. Also put 15V on TP201 with the chip removed and check for some other sneak path that could be putting 15 on the 5V buss (e.g. internal short in comparator or op amp, logic gate, etc.). i think i have drawn up nearly all the chips but need to double check. Go thru the ICs in order, determine Vcc and if there are input pins at 5, 12 or 15, etc and make sure there is isolation.

 

[eldenh]

Thanks for finding a source. But Arrow wants $50 for shipping! I have no experience with smd devices or multilayer pc boards. Rather than learn on this one, I prefer to buy one if possible or find someone to repair the one I have. I doubt I can effectively determine what was the original fault until it is as functional as it was before my probing damaged it further.