• State Resolved
  • Locked Locked
  • Replies 17 replies
  • Answers 3 answers
  • Subscribers 50 subscribers
  • Views 2995 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Original question:

OPA549: opa549

OPA549: Current Increasing by using pararell 2 OPA549

J.Hoo Lee
Prodigy 30 points

Part Number: OPA549
Other Parts Discussed in Thread: TINA-TI, LM3464, TPS92518, TPS92692

hi team,

My goal is to source a constant current of 1A to 15A using two OPA549s in one high power LED continously. (The specifications of the LED are shown in the figure.)
The figure below is a circuit using two OPA549s in parallel. (I Refered to the suggestions in the datasheet..)
Since I only use constant current, I want to keep the voltage fixed and control the ILIM as a DAC.
Actually I already implemented a 0.5-6A constant current source in the same way using one OPA549 and it works just fine.
(The additional circuit is configured as shown in the figure, and this circuit has an accuracy of less than 1mA if the drift caused by temperature is ignored.)

My concern is that the datasheet provides two parallel figures of OPAMP, but it's hard to find material or reviews implementing them. From some of the stuff I've checked, they all seem to be struggling and not exactly resolved.

Before I build the actual circuit, I need as much information as possible to consider. Maybe in the part of implementing my purpose "Voltage controlled continously constant current source <Vmax=5V, Imax=15A (the best if up to 20A)>", I think that the method of using the OPA549 may not be suitable. Actually 15A is considering the specs of 2 OPA549s, 20A is better for me.

to sum it up,
1. In the above configuration, what am I missing or something to consider?
2. Any suggestions from TI that are more efficient or effective for my purposes?

Thank you.

hoo

  • 0
    TI__Expert 4720 points

    Hello,

    The team is on vacation. You can expect an answer before the end of the week. Thank you for your patience.

    Regards,

    Castrense

  • 0
    Guru 324920 points

    These opamps are not in parallel; pins 1/2/3/4 of IC1002 are not connected correctly.

    This circuit will dissipate lots of power inside the opamps. This might be outside the SOA (see figure 6), and in any case is not efficient.

    If you do care about efficiency, you need some kind of buck converter. See LED drivers for devices that allow dimming, but 20 A appear to be beyond their capabilities. I'd guess that PWM dimming would be easier to implement.

  • 0
    Guru 140200 points

    Hi Junghoo,

    I'm also a bit confused:

    J.Hoo Lee said:
    My concern is that the datasheet provides two parallel figures of OPAMP

    I only see the master-slave circuit from figure 14, which you seem you are using? And as Clemens already mentioned R1005 is not connected correctly.

    I think you will go too far to the limit of OPA549, not only because of the huge heat dissipation you have to deal with but also because the current limit can set in unevenly in the two OPA549 additionally increasing the heat dissipation in one of the OPA549.

    You loose up to 4.8V output voltage at an output current of 8A. So, provided you choose a supply voltage which is just 4.8V higher than the wanted output voltage, the OPA549 has to dissipate 4.8V x 8A = 38.4W. With a 1°C/W heat sink this will heat up the junction of OPA549 to about 38.4W x 2.4C/W = 92°C over the ambient. So at an ambient temperature of 30°C the junction will heat up to about 122°C, which is quite a lot.

    And when you take a higher supply voltage to allow for some headroom of output voltage the heat dissipation will increase even more.

    Kai

  • 0 in reply to kai klaas69
    TI__Guru* 93105 points

    Hi Junghoo,

    Clemens and Kai have addresses the many concerns and issues with your OPA549 constant current application. The high power dissipation levels is likely to become a problem because it is often difficult to achieve the ideal thermal mechanical requirements on a consistent basis. I think Clemens' suggestion about the PWM dimming is better approach from a power dissipation standpoint.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

  • +1 in reply to Thomas Kuehl
    TI__Guru* 93105 points

    Hi Junghoo,

    If you still would like to try using two OPA549 power op amps to generate a high constant current I show below a circuit I came up with some time ago that is a modification of the Improved Howland Current Pump. It differs from the conventional Improved Howland Current Pump in that it incorporates a two power op amp, Leader-follower configuration in the pump. I have not built this circuit on the bench, but it simulates nicely in TINA-TI so I do think there is a high probability that it will do what is intended. My main concern is the the thermal issues brought up in the earlier posts.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

    OPA549_Howland_Parallel_02.TSC

  • 0 in reply to Clemens Ladisch
    Prodigy 30 points

    Hi Clemens,

    Thanks for let me know of my silly. It was a mistake and now i corrected my schematics.

    As expressed in the schematic, the DUT is an 'white LED' and this is not just a light. It judged that the high FPS of the inspector measuring LED is not suitable for PWM Dimming. This is why I use analog dimming.

    If 20A is out of the allowable range, can you tell me the information of the LED driver that has the highest current?

    Hoo.

  • 0 in reply to kai klaas69
    Prodigy 30 points

    Hi Kai,

    thanks for your reply.

    Maybe I didn't understand your question.

    kai klaas69 said:
    I only see the master-slave circuit from figure 14, which you seem you are using? And as Clemens already mentioned R1005 is not connected correctly.

    I made and verified a device using one opa549, but I haven't actually tested using both in parallel. The circuit is still under design. And Clemens' point has been corrected.

    As you are concerned, many threads already know that thermal issues in this way are a topic. Of course, this is very difficult. However, my device is not severely limited in space. This means that a large area is allowed for heat dissipation.

    Of course, I'd like to consider more methods, but it's hard to find a solution that works for my goals (15A-20A).
    I currently in the process of checking which methods can be used.

  • 0 in reply to Thomas Kuehl
    Prodigy 30 points

    Hi Thomas,

    Thanks for your support.

    about the parallel connection of OPA549,
    I understood that if the thermal management was overcome, there was no problem with the implementation.

    Your simulation will also be of great help to me.
    For my 2nd question in the text, should I get the help through another thread?

    J.Hoo Lee said:
    2. Any suggestions from TI that are more efficient or effective for my purposes?

    Hoo

  • +1 in reply to J.Hoo Lee
    Guru 324920 points

    There are LED drivers that just control external current sinks or DC/DC regulators, and where the specified current limit just indicates normal usage. It might be possible to scale up the current by using different components (i.e., smaller sense resistor, larger MOSFET) and inserting a stronger gate driver.

    Look at devices like the LM3464, TPS92518, or TPS92692. For those, please ask a new question in the power management forum.

    Also see Dimming in switched-mode LED drivers.

    Are you considering multiple LEDs? What would probably the easiest solution.

  • 0 in reply to Clemens Ladisch
    TI__Guru* 93105 points

    Hi Junghoo,

    The switched-mode PWM solutions that Clemens suggests will certainly provide much higher efficiency and produce much less heat than an analog solution. Those are certainly very important considerations for any modern electronics application.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

  • 0 in reply to Clemens Ladisch
    Prodigy 30 points

    Hi Clemens,

    Thank you for your continued interest and help.

    Clemens Ladisch said:
    Are you considering multiple LEDs?

    I only considering one high-power(spec in the above picture) LED. 

    PWM Dimming is not considered for me. There is no dispute that PWM Dimming method is very efficient in terms of power and heat.

    However, as i explained earlier, My Device(the LED) use for lighting Inspector(an camera).

    Inspector is kind of camera has high resolution and high Frame per second. So, It may be detected the light on/off at PWM Dimming.

    At this time, the specifications of the camera may vary, so increasing the PWM frequency is not a solution. and analog dimming provided by a general LED driver's resolution is very lower, so it is difficult to use.

    This is why I ruled out PWM.

    Additionally, the current must be very stable. (Several mA or less) I have already configured a device that is very stable (less than a few mA when the ambient temperature is static after reaching thermal equilibrium) at a maximum of 7A using one OPA549, and it is very stable under continuous operation for at least 20 days.

    Clemens Ladisch said:
    What would probably the easiest solution.

    I can't know which one would be easier, but I'm also considering a current mirror combined with an opamp and FET.

    I'll ask new question at power management forum as per your advice.

    Thank you.

    Hoo

  • 0 in reply to Thomas Kuehl
    Prodigy 30 points

    My further question seems to be off topic in the thread. thank you.

  • +1 in reply to J.Hoo Lee
    Guru 140200 points

    Hi Junghoo,

    ok, if PWM rules out and this seems to be a physical research application were costs, heat dissipation and miniaturisation don't play any role, then I would think about using more than two OPAmps in parallel. You could take the circuit from figure 14 and add two additional slave buffers. Increase the equalization resistors (also called isolation resistors) from 0.1R to 0.4R then.

    As a rule of thumb, the bigger the equalization resistors the more stable the circuit. But carry out a phase stability analysis first and perform thorough testings of the real circuit with a dummy load. For this you will need an as correct as possible equivalent circuit of your "LED" load.

    Increasing the gain of master can also improve stability. The recommended circuit in figure 14 shows a gain of 5V/V.

    Kai

  • +1 in reply to kai klaas69
    TI__Guru* 93105 points

    Hi Junghoo,

    Then, using Kai's recommendations for multiple power op amps you could apply a second OPA549 follower op amp to the parallel output, Improved Howland Current Pump circuit I proposed. Certainly the total number of OPA549 op amps becomes 3, but then each would provide 6.67 Amps towards the 20 Ampere total. That would help keep the individual power dissipation down to more reasonable levels and allow somewhat closer output voltage swing to the supply rails for a little more compliance range.

    I am providing an updated schematic for the parallel output, Improved Howland Current Pump employing 3 OPA549 power op amps. The maximum output current has been increased to 20 Amperes. I left the series output resistors at 200 milliohms for each of the three op amps. The only other components changes are the lowering of the two leader amplifier feedback resistors from 300 Ohms, down to 267 Ohms.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

    OPA549_Howland_Parallel_03.TSC

  • 0 in reply to kai klaas69
    Prodigy 30 points

    Hi Kai,

    I really appreciated your advice.

    Your advice that increasing the gain and equalization resistance improves stability is very helpful to me.
    However, it will be difficult to obtain the exact equivalent circuit of the LED, so this part (tunning of equalization resistance) seems to have to be actually tested.
    The following figure is the IV curve of the LED I am currently using. The maximum rating is 18A.

    Hoo

  • 0 in reply to Thomas Kuehl
    Prodigy 30 points

    Hi Thomas,

    Thanks for the updated simulation. This will be of great help to me.
    However, there are additional requests.


    1. I use a single power supply(positive) for my OPAMP. Can you provide a simulation under the condition that VCC is 12V?
    (Since the voltage of the load is less than 4V, the power loss due to too high a voltage drop cannot be ignored.)

    2. In your simulation, Is the equalization resistor(R3, R5, R11, R12, R4, R6) on the IOUT pin absolutely necessary?


    Hoo

  • 0 in reply to J.Hoo Lee
    TI__Guru* 93105 points

    Hi Junghoo,

    Come on - you really should be able to use the OPA549 TINA-TI circuit file I provided and change the supplies to +12 V. Note that the Improved Howland Current Pump solution I proposed uses dual supplies (+/-V). I expect that the LED load will be connected from the circuit's output from the LED's cathode to GND. Then the OPA549 Op amps will be sourcing the output current upwards to 20 Amperes to the LED load, and they will not be sinking current. Nevertheless, the OPA549 will require a negative supply in addition to the +12 V supply to maintain operation. I suggest using -5 V for -V if my assumptions are correct.

    Using +12 V/-5 V supplies will much reduce the power dissipation compared to +/-30 V supplies. Additionally, if you plan to use a +12 V supply the positive output swing will be limited to several volts below that level due to the OPA549 Vo swing capability limits. Originally, you indicated a 0 to 5 V input to the amplifier. Since you are proposing lower supplies the voltage could not be that high without hitting the Vo limit. You will have to adjust the circuit's V-to-I transconductance gain (ΔIout / ΔVin) if you still want to use a maximum +5 V input to source a 20 Ampere output to the LEDs.

    R3, R5, R11, R12, R4, R6 are 0 Ohm resistors only resistor place holders in the TINA-TI simulation schematic. Just in case we wanted to give them a value other than 0 Ohms. They are not necessary in a real applications circuit. The series output R14, R15 and R17 ballast resistors are a must just in case you are thinking about them.

    Regards, Thomas

    Precision Amplifiers Applications Engineering