• State Resolved
  • Locked Locked
  • Replies 13 replies
  • Answers 3 answers
  • Subscribers 62 subscribers
  • Views 1171 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.

TPS62822: Output voltage not held at higher loads

Rutger van Dalen
Prodigy 130 points
Part Number: TPS62822
Other Parts Discussed in Thread: LMR33630, TPS62088, TPS62140, TPS82140, TPS62130

Hi,

I've been running some tests with the TPS62822 EVM, and changed the output voltage to 3.3V by changing R1 to 453k. Without load and loads up to 100 mA, the output voltage is a near perfect 3.3V. However, at higher loads (850mA and 1.5A), using resistors as loads, the voltage drops to 3.18V and 3.01V, respectively, even though these currents are well within the device spec. Please advise.

Thanks

  • 0
    Prodigy 130 points
    Forgot to mention that Vin = 3.6V, using both a lab power supply and a Lithium battery.
  • 0 in reply to Rutger van Dalen
    TI__Expert 3465 points
    Hello Rutger,

    it's my pleasure to help you.

    Could you specify at which exact pins on the EMV are you measuring the input and output voltage?

    Could you also check whether you see a value change in the measurement of the input voltage?
  • 0 in reply to Emmanuel Granatello
    Prodigy 130 points
    Hi Emmanuel,

    I measure the input voltage on VIN/GND of J1 and the output voltage on the VOUT/GND pin of J2. I see no change in the input voltage.

    Thanks,


    Rutger
  • 0 in reply to Rutger van Dalen
    TI__Expert 3465 points
    Hi Rutger,

    may you please measure at the S+ and S- pins of the respective voltage headers and tell me the result?
  • 0 in reply to Emmanuel Granatello
    TI__Expert 3465 points
    I mean S+ vs S-, of course.
  • 0 in reply to Emmanuel Granatello
    Prodigy 130 points
    I don't understand, according to the schematics, S+ is the same net as VIN/VOUT on J1/J2 and S- is GND on both headers. How would this change the result?
  • 0 in reply to Rutger van Dalen
    Prodigy 130 points

    Never mind, looking at the layout rather than the schematics, I see what you mean...

    Here are the results (Vin/Vout, measured on S+/S-):

    Running on lab power supply:

    zero load: 3.589V/3.332V
    850 mA load: 3.446V/3.324V
    1.5A load: 3.349V/3.249V

    Running on Lithium battery:

    zero load: 3.665V/3.332V
    850 mA load: 3.308V/3.240V
    1.5A load: 3.197V/3.081V

  • 0 in reply to Rutger van Dalen
    TI__Expert 3465 points
    Hi Rutger,

    you found the reason: the layout traces do not have zero resistance, and at the same time the feedback resistor R1 is tapped at the output capacitor. The voltage will be then regulated at that capacitor.

    I can also notice that your lab power supply has a discrete output resistance. Indeed you see that the voltage measured at the power supply is lower than the set 3.6V, causing the device to work in 100% mode and deteriorating the output voltage regulation. This effect is of course not present when Vin is 3.6V.

    So, if your lab power supply has sense lines (aka Kelvin connection) for its own regulation, please use it when evaluating our EVMs (that's another reason for the S+ and S- at the input).

    Regarding the lithium battery case, I was expecting the high resistance to cause the drop on Vin. Due to that, the device is going to work in 100% mode at higher loads. If you desire to avoid that, you should use an higher voltage or lower output resistance battery.
  • 0 in reply to Emmanuel Granatello
    Prodigy 130 points

    Hi Emmanuel,

    Thank you for sharing your conclusions. What I'll try is to adjust the power supply output voltage to a level where the voltage at S+/S- is 3.6V and then measure again. I'm not sure what you mean with "the feedback resistor R1 is tapped at the output capacitor. The voltage will be then regulated at that capacitor." Could you please clarify?

    In our scenario, we are in hibernation mode (1µA current draw) around 99% of the time, so we're looking for a power supply that has a good performance in these situation. At most a couple of times a day, our transceiver draws 1500 mA for a duration of 300 ms, so the power supply needs to be able to deliver 2A (to be on the safe side), but doesn't need to be very efficient in this phase, although most converters tend tp perform better at the higher loads. I think the TPS62822 is a good fit, but if you have any other suggestions, I'd be happy if you could share your thoughts.

    Regarding higher voltages, a series configuration of n batteries is also used. Which converter would you recommend in this case. I've evaluated a couple already, and have ordered a LMR33630 EVM, but if you have other suggestions, I'd welcome those too.

    Thanks again,

    Rutger

  • 0 in reply to Rutger van Dalen
    TI__Expert 3465 points
    Hello Rutger,

    As you have experienced, the resistance of the layout traces influences the output voltage regulation. Actually, the traces that make this effect are only the ones not included in the feedback loop. The part of the loop that one can influence in this circuit is the resistor divider, which is the "sensor" of our system.

    The point of the layout where this sensor is connected is the exact point that will be regulated, since it's voltage, once scaled down by the divider factor, will be equal (with a certain very small error) to the reference voltage. For example, if one would connect R1 to the Vout header, you won't see any drop with increasing load current.

    The reason why we don't do this is because we make our devices fitting highly integrated applications, where the sensing point is the output capacitor (which in turn is very close to the load). Our EVMs instead have kind of standard dimensions to be of practical use and evaluation for our customers. But still, we want our customer to see what actually happens on their application, so we furnish them with the S+ and S- lines, which connect directly at the output cap.

    Regarding your specific scenario, my understanding is that you have to deal with a battery operated object which is in standby the most of the time. I guess you want to have the longest battery life possible, and at the same time you are evaluating the possibility of having higher battery voltages.

    You did a great choice with the TPS62822, but if you really want the best in class you shouild have a look at the TPS62088 (available in fixed versions for 1.2V ,1.8V and 3.3V Vout, that means no external resistor divider needed!).

    When evaluating higher Vin, then I would suggest the TPS62140 or it's module version TPS82140. I would really suggest the module, since you can take advantage of an easy layout plus do not need to search and buy a suitable inductor. Your choice, both are really good up to 17V Vin.
  • 0 in reply to Emmanuel Granatello
    Prodigy 130 points

    Hi Emmanuel,

    Thanks again for sharing your thoughts and expertise! I've selected the TPS62822 because in some cases we need other output voltages (e.g. 3.0V), so this gives us some flexibility. Nevertheless, I've ordered the TPS62088 to give it a spin.

    When I use WEBench to create the design with the TPS62822, it suggest the 744373240047 from Würth as inductor. This is different inductor than the one used on the EVM. I assume that TI has selected the Würth part because it has the best performance. What can I expect in this respect, if I use the Würth inductor?

    The TPS62130 (the 3A-counterpart of the TPS62140) and TPS82140 were on my shortlist, and I've done some tests with both EVMs. How do you rate the LMR33630 in comparison? Based on the collateral, I got the impression that this device would also serve us well.

    Cheers,

    Rutger

  • 0 in reply to Rutger van Dalen
    TI__Expert 3465 points

    Hi Rutger,

    Webench gives you the possibility to optimize your design from different aspects and priorities using the main knob. I guess you went for the max efficiency option, since this wuerth inductor has a very low DCR, but quite high price and dimensions. Anyway, you must take care of how efficiency optimisation is done.

    Indeed, in a first instance, webench calculates the efficiency based on the max current (which you chose as the main input values). In my example below, you would hence get the project that maximizes the efficiency at 1.5A. But you want information on a much lower current. 

    So I would suggest to leave the max current as 1.5A in the box highlighted below, but use the advanced options, in which you can give a specific operating point for the efficiency evaluation and design scoring. You will get something like this when the operating current will be 100uA:

    NB: better ti not go below 100uA because the current range included in the database might be incomplete for lower currents (you might lose the evaluation of some devices not because they are not fitting your requirements).

    As you see, the efficiency is lower, but the TPS62822 and TPS62088 are still top choices. Anyway, the inductor you get is still the wuerth, because of the max efficiency. But you can evaluate easily if a smaller and cheaper project can still give you a good result at low load (which should because the losses on the DCR are really small at those levels). For example you can go for the central knob position, like this:

    and a slightly higher DCR inductor is suggested (but smaller and cheaper), plus the efficiency at 100uA is still the same!

    You could also go further, opening a design and searching for a smaller inductor evaluating the effects immediately. The choice is yours!

    Regarding the LMR33630, I would prefer it only for Vin>12V. For Vin<12V the TPS62 and 82140 are better in terms of IQ and efficiency (assuming they work at the same Fsw range of course, which one usually wants to be high in order to save price and space on the passives).

  • 0 in reply to Emmanuel Granatello
    Prodigy 130 points

    Hi Emmanuel,

    Thank you so much for the additional information, very helpful! For now, you've answered all my questions, so we can close this case.

    Kind regards,

    Rutger