• State
  • Locked Locked
  • Replies 14 replies
  • Subscribers 64 subscribers
  • Views 5497 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TPS40210 DC DC conversion from 19-32V TO 48-120V shutting down problem

N S
Prodigy 110 points
Other Parts Discussed in Thread: TPS40210

I have build a circuit using TPS40210 Boost converter from TI. The circuit specs are:

Vin = 19VDC to 32VDC (Could be any DC level in between these limits)

Vout = 48VDC to 120VDC (Variable output voltage obtained by varying DigiPOT value in feedback path) (Max Load current is 100mA)

The circuit is as shown in the attached picture (This is the circuit of DC-DC converter side only). There is other circuitry on the same board that contains processor and also downstream DC-AC conversion circuitry using PWM from processor. But, the problem I am facing currently is there even when I isolate the DC-AC conversion circuitry

I have also attached the layout of the PCB

This is a 4 layer PCB with 1oz Cu

I have tried to copy the layout from EVM module as much as I could.

Processor changes the Digipot values in steps to vary the output coltage from 48V to 120V apprx

This circuit has been working fine previously on about 20 different PCBs.

We have just ordered some more of these PCBs, and all of sudden this circuit became too noisy

To start with it induced a noise on processor Reset line causing it to go in reset state randomly. that never happened before I managed to get around that by putting a10nF cap on processor Reset line.

So, processor is now working but DC DC converter keeps going to shut down mode when Digipot changes from value of 30 downwards corresponding to 70V apprx upwards.

In previous PCBs it worked fine when Digipot was changed from 255 down to 0 in steps of 10 and DC DC converter never went to shut down mode BUT

In these PCBs, it started to go down to shut down mode. I tried few changes as:

I changed the step size from 10 to 2 (used 5 before and sometimes it worked fine for few cycles but at other times, it went to shut down mode). Couple of PCBs worked fine at this step size (although they were still noisier and had to use cap on Reset line). All other PCBs are still problematic as in DC DC converter keeps going to shut down mode.

I also tried disconnecting DigiPOT and connected an external analog POT and found that it changing the POT abruptly caused the unit to go in shut down mode so decided to use the smaller step size

Changed inductor, 10mR current feedback resistor, TPS40210, current sense cap from 180pF to 47pF (but problem is still there)

The output of this circuit is also monitored by using potential divider, buffer and ADC (i2c – comms with ADC )

Our s/w engineer has helped me in modifying the software for the processor, that if it ever detects less than <40V then it Disables the converter (Enable-DCDC line shown in the schematic) and enables it again. This helps for quick discharge of 10uF cap connected to ‘Soft Start’ pin as shown in the diagram. But problem is still there and sometimes it works and other times it keeps collapsing even when it is sitting at one voltage level (usually 70V upwards i.e., anything from 70V to 120V). It collapses, goes to disable mode and then enables again (done by processor for faster response to avoid the wait time until soft start cap discharges)

There is a series resistor of 220R (R16) connected in feedback that limits the maximum voltage that could be obtained when digipot is sitting at 0

The component values are indicated on the schematic. I designed this circuit originally by using Webench from TI and also with the kind help of TI sales rep from Arrow electronics who helped us in obtaining EVM modules for this converter. As I said this circuit has been working fine before but all of sudden it decided to have a tantrum. Please help

PCB is 4 layer, see attached pictures indicating each layer

Also, there is a picture showing the plated through hole vias used on DC DC converter side connecting the copper areas on all 3 layers

  • 0
    Prodigy 110 points

    Red is Top copper

    Yellow is Layer2

    Green is Layer3 (with GND plane)

    Blue is Bottom Layer

  • 0 in reply to N S
    Prodigy 110 points
    As per datasheetpp13, UVLO, OCFault or DIS, could cause the converter to go in shutdown. Clearly, its not UVLO or DIS so it has to be OC-Fault, i.e, Isense is detected >150mV. But can't figure out the reason for this going above in these latest PCBs. I checked it on oscilloscope and both (old working PCBs) and new (problamatic PCBs) show noisy waveform at Isense, (its already in the noise spectrum of oscilloscope TDS2024C) that its hard to see if its that noise causing it OR is it something else that I cant find!
  • 0 in reply to N S
    TI__Mastermind 35135 points

    Hi Naeem,


    Sorry for the long delay in response on this. I had to think through it a bit to figure out where to start.

    One of the first things I'd like to do is check if it's noise or an actual over current condition causing the TPS40210 to hiccup.

    • Is there any load on the output when the TPS40210 hiccups above 70 V output?
    • Could you share some of the waveforms you took of Isense on the new and old PCB? I would recommend using a voltage scale of around 100 mV so it's easy to see the 150 mV current limit threshold. It would be good to also have the GDRV waveform. Alternatively to the ISNS voltage, if you have a current probe I would like to get a look at the inductor current measured on the VIN side of the inductor since this can give us a clearer current measurement.
    • Is it possible to remove D2 to isolate the TPS40210 circuit and check its performance again?

    Best Regards,
    Anthony

  • 0 in reply to Anthony F
    Prodigy 110 points

    Hi Anthony

    Thanks for the feedback. I have done some tests today and have some results for your attention. Tests were performed on Old Good board marked as GB and New Bad boards marked as BB. For both boards Diode D2 was disconnected so DC DC converter is now working in an isolation with only load of voltage divider network that goes into ADC.

    NB: There are some videos that also show the situation on oscilloscope when the Vout collapses as it was hard to capture the trace at the moment when it is collapsing. Also, unfortunately we don't have any current probe in house that could be used to obtain Inductor current's trace. So, its GDRV, Isense and Vout on traces. You could see Vout value going up in videos each time Digipot is changed

    Plots are obtained as under: (GB=Good board, BB=Bad Board) (Ch1-Yellow=GDRV, Ch2-Blue=Isense, Ch3-Pink=Vout)

  • 0 in reply to N S
    Prodigy 110 points

    Its not letting me attach the traces or videos. I will try another way

  • 0 in reply to N S
    Prodigy 110 points

    See video below

  • 0 in reply to N S
    Prodigy 110 points

  • 0 in reply to N S
    Prodigy 110 points
    I have managed to compress one video and uploaded it. This is the one that shows Vout collasping and Isense & GDRV traces
  • 0 in reply to N S
    TI__Mastermind 35135 points

    Thank you for all of the information Naeem. One thing I noticed is the waveforms show more noise at ISNS on the BB waveforms. A useful experiment to run to find the source of the noise would be to move the components on the bad board to the good board and see if you can make the good board go bad. Specific components I expect to contribute to this first are the TPS40210, MOSFET, and diode.

    Best Regards,
    Anthony

  • 0 in reply to Anthony F
    Prodigy 110 points
    Hi Anthony
    Thanks for the prompt reply. I did some more tests today as per your recommendations. I replaced components on the Good board one by one with the components from Bad board. I swapped TPS40210, no issues over the whole range from 48-126V. I then swapped MOSFET, still no issues. After that I swapped the diode, Hey presto, the good board turned into a bad one. To start with, it started to reset the processor (as I didnt have any protection caps on old boards) due to noise even at lower voltage outputs in range of 50-60V. I installed a cap on reset line and found that good board was now behaving like bad board as in the Switch controller kept going to shut down mode. I repeated this test with diodes from few more new Bad boards, and all diodes showed same problems on Old good board. I soldered the original diode back on old good board and its working fine.
    So, the conclusion is that culprit is Diode. But the question is why? This was the recommended diode 'Ultra Fast Rectifier --- US1G-13-F' from DiodesInc by Webench. It has been working fine so far on all previous boards but the decided to become problematic on this batch of 25 PCBs. Diode looks fine from simple DVM test (OK in Forward bias, blocking reverse bias). From that point of view I can't send them back to the supplier classing them as bad diodes unless I am missing something here. What are your thoughts? Do these diodes need to be tested some other way so that we are certain they would be fit for purpose? Should I use any other diode in similar package (so that I don't have to change the PCB layout? Any recommendations?

    Many Thanks
    Best Regards
    Naeem
  • 0 in reply to N S
    TI__Mastermind 35135 points

    There must be a variation in the reverse recovery performance of the diode causing more noise. More reverse recover charge or longer trr would cause more noise. Changing to a schottky diode will help to improve the switching performance and reduce the noise. Unfortunately WEBENCH can select non schottky diodes due the parts it has to choose from. I'm also forwarding this to the team who works on WEBENCH to see what we can do to improve the selection.

    I took a look on digikey for some schottky diodes you may try and the link below is the full search results.

    Specifically the SS2150 or SS1200 look like good options.

  • 0 in reply to Anthony F
    Prodigy 110 points
    Hi Anthony
    Apologies for delayed response as I was waiting for different diodes to be delivered. I have now tested the following diodes:
    Farnell-2211965 --- rated600V/1A (Worked OK at 24V input but @ 19V input collapsed twice when digipot going down from 10 to 0 apprx 110V to 120V --- No extra noise generated that could reset the processor
    Farnell-2452035 ---- rated200V/2A ----- Generates extra noise -- Not Good
    Farnell- 2211963 --- rated400V/1A (Worked OK at 24V input but @ 19V input collapsed twice when digipot going down from 10 to 0 apprx 110V to 120V --- No extra noise generated that could reset the processor
    Farnell- 23111234 -- rated300v/3A ----- Generates extra noise -- Not Good
    Digikey - SS1200-LTP --- rated200V/1A (Worked OK at 24V input but @ 19V input collapsed twice when digipot going down from 10 to 0 apprx 110V to 120V --- No extra noise generated that could reset the processor
    Digikey- SS2150 -- rated150V/1A (Worked OK at 24V input but @ 19V input collapsed twice when digipot going down from 10 to 0 apprx 110V to 120V --- No extra noise generated that could reset the processor

    All of the above were tested on Bad board. It then triggered a thought to check a good board @ 19V input (as I have been testing it at 24V that is a nominal recommended for this apparatus but must be able to work between 19V-32V) with all original components and it was found that that Vout on that board also collapses when digipot goes down from 10 to 0. It was intended for original design to work from 19V input to 32V input to generate 48-120ish V output. Over the past few months it has only been used @ 24V input and no problems were observed but it is now found that at Low voltage input (i.e., 19V, 20V also tested) the Vout starts to collapse indicating the controller going to shutdown mode. Is there anything that could be done to rectify this issue?

    Many Thanks
    Best Regards
    Naeem
  • 0 in reply to N S
    TI__Mastermind 35135 points

    Something which can help further reduce the noise from the schottky diode is adding an RC snubber in parallel to it. Typical values are around 5 ohms and 1000 pF but it is best to tune it for the given layout and power stage. The following is a good guideline for choosing the values.