0

As a follow up to this question. (Sorry if this linking other posts is wrong. I'm new to forums) Will this design for an XHP70.2 LED Driver with arduino control work to the params specified?

TL;DR: its a 15A~7V LED driver Current control loop controlled from an Arduino PWM. I'm new to op-amps.

enter image description here

CRE2512-FZ-R003E-3 - https://www.mouser.co.uk/datasheet/2/54/cre-1388683.pdf

INA180A2IDBVR - http://www.ti.com/lit/ds/symlink/ina180.pdf

I have found components for the first Current Sense Amplifier stage. It will output 0v - 2.25V and so, as suggested, I think a unity gain differential amplifier would work for the second stage. (with a potential divider to take the Arduino control from 5v to 2.25v and a potential divider to take the second stage output down to 2v)

  • What would be a suitable amp? (ideally something similarly small as the SOT-23-5 INA108A2)
  • Is there a way to reduce the number of resistors so I don't need All Of R9 & R10 & R11 & R12 & R13?

Thanks in advance

Wizzard_Dr
  • 69
  • 7
  • Not only R9-13, you don't need most of the other components either. I think your system design is way over the top. For instance, your ATiny outputs a numerically controlled PWM waveform, which controls a voltage controlled PWM waveform, think about that. However, the loop could be far tighter than that. If the ATiny outputted a current reference, and then the two FETs switched to ramp the LED current above and below that reference with a bit of hysteresis, you would have a simpler, safer, system. – Neil_UK Mar 11 '20 at 05:10
  • Thanks for the feedback. The reason for the pwm controlling the pwm is speed the VCO does 1mhz. I don't quite understand what you mean by the attiny outputting a current reference? Are you meaning to or not to use the ncp3420 synchronous buck driver? – Wizzard_Dr Mar 11 '20 at 05:39
  • The ncp3420 stays. Drive it from a fast comparator. One comparator input driven from the filtered PWM output of your atiny, the 'current setpoint'. The other comparator input driven from a current sense resistor in the ground lead of your load, the 'current sense'. Have some hysteresis round your comparator. When the current sense is below the setpoint, the upper FET is on, current rises. When sense is above the setpoint, the lower FET is on, freewheeling and current falls. The compator hysteresis controls the switch rate, together with the value of L and the voltages. Do you need a diagram? – Neil_UK Mar 11 '20 at 07:26
  • Oh I see. I think I understand. I was only going for the ltc6992 because I am new to opamps. I haven't used/made a comparator before. (I think that if I'm understand correctly, you are saying to replace the second opamp and LTC6992 with one comparator) As I am wanting it to be high side sensing would/could I still use the current sensing amplifier? Also how do I keep a 1mhz fixed frequency with a comparator setup? (or a frequency that maintains efficiency) – Wizzard_Dr Mar 11 '20 at 17:17
  • If you want a fixed fequency and high side sensing then it needs more components than I suggested. Presumably you're using the atiny for controllable brightness? You'll get much higher efficiency at much less than 1MHz, the FETs will suffer much less turn on/off loss, though you'll need a larger L (which will also be more power efficint (not space efficennt) at << 1 MHz. – Neil_UK Mar 11 '20 at 17:46
  • Unfortunately, space is the restriction and efficiency is the aim. I'm currently looking at the XAL7030-332ME_ https://www.coilcraft.com/xal7030.cfm Is this anywhere close to being a high enough value? I could probably go 10% - 20% bigger on total physical size but that's pushing it and might mean recessing it in the PCB. – Wizzard_Dr Mar 11 '20 at 18:15
  • https://i.stack.imgur.com/T6LYw.png Is this what you mean? – Wizzard_Dr Mar 11 '20 at 18:18
  • https://electronics.stackexchange.com/questions/184950/hysteresis-control-disadvantages/185019#185019 This is making me think hysteresis control is a bad idea? – Wizzard_Dr Mar 11 '20 at 19:47
  • Yes, that sort of thing. You've not shown any hysteresis on your comparator, but there's probably enough delay through your current sense amplifier not to need it. Self oscillating like that is no problem, as long as the variable frequency doesn't cause you interference issues, I don't know what environment your LED is used in. One problem I see with your system is a PWM-controlled buck is a voltage output, and LEDs take very rapidly varying current with voltage control. That's why I favour the current control of my approach, the control loop is easier to stabilise. – Neil_UK Mar 11 '20 at 20:01
  • That makes sense. So is it right that the hysteresis is to limit the frequency? How do I calculate/anticipate the min & max frequency for different reference inputs, undershoot and overshoot? How do I limit the minimum frequency when there is a reference drop or current spike? How will the inductor deal with the ranging frequency compare to a fixed frequency? – Wizzard_Dr Mar 11 '20 at 23:36
  • The hysteresis is not to *limit* the frequency, it's to stop the loop settling at a mid-point, a buck converter needs to switch. However, it's likely there's enough delay round the loop to do the same job, and extra hystersis would modify the frequency down further. Frequency is a non-issue, the questions about how you limit or calculate it, how the inductor deals with it are not a problem. Design for energy storage in the inductor, and the frequency is then whatever it is to meet the resulting on and off times given the voltage, programmed currents and L value. Unless EMC matters! Does it? – Neil_UK Mar 12 '20 at 04:47
  • EMC doesn't matter in the sense of regulations but there are I2C lines near the switching and it would be generally good not to have it produce heps of EMI. What value do I look for in the current sense amplifier datasheet to see how much delay it will add? Or is there a small smd current sense amplifier you would recomend that would be fast? -Thanks for all your help BTW – Wizzard_Dr Mar 12 '20 at 05:18
  • I meant EM radiation picked up by sensitive receivers, not logic-level circuits nearby. If you have to worry about the frequency of aggressor circuits to get your nearby I2C to work, then you have big fundamental problems in both circuits and are going about it the wrong way. You avoid producing heaps of EMC by understanding the paths the currents take and change between on switching events, keeping those loops small, and proper decoupling, the frequency is essentially irrelevant. If your EMC sits right in your AM band and trashes radio reception, then it might be worth moving it. – Neil_UK Mar 12 '20 at 05:25
  • No there's no sensitive receivers and it's all going in a metal case anyway. It's tricky though because it all needs to fit on a long 9mm wide PCB. – Wizzard_Dr Mar 12 '20 at 05:37
  • The parameters you need to worry about are maximum and minimum programmed LED brightness level, not seen those mentioned yet. When running, you need to avoid too high LED peak current. My route handles that directly by switching off at a current threshold, yours indirectly by choosing a switching frequency, L value, voltage levels, then slowly controlling a mark/space ratio. If you need the size, then go small L high frequency, at the expense of higher FET switching losses. It's the topology and control method I'm advocating. Run a SPICE simulation of both topologies and see for yourself. – Neil_UK Mar 12 '20 at 05:38
  • What do you mean brightness level? Is that not the 0a - 15a? I'm currently going down the route of hysteresis control. The inductor I found that is small enough is the 3.3uH XAL7030-332ME_ coilcraft.com/xal7030.cfm is this inductance high enough? (I'm confused, is the hysteresis control method or the PWM control method you are advocating?) – Wizzard_Dr Mar 12 '20 at 06:09
  • So you reckon the 3.3 μH XAL7030, with its 12.3 A saturation current, and > 80 °C rise at 15 A is going to be suitable do you? I'm still advocating current rather than voltage control, and free-running rather than fixed frequency. – Neil_UK Mar 12 '20 at 07:06
  • Wrong one. I meant the 2.2uH XAL7030-222ME_ . It states 18A Isat and 40deg C rise at 12.9A in the air, but I'm hoping for better temps with a heat sink. I can't work out whether that value L will have any effect on the square wave of the output not? I think it should give a 3% current peak to peak but i may be magnitudes off. – Wizzard_Dr Mar 12 '20 at 08:01
  • Why such huge values of L? You're making a rod for your own back. What ripple at the LED do you really need? If you want low ripple, then a small C across it would be useful. Even 2.2u is massive. Have you worked out \$I^2R\$ for 15mohm at 15A? 1uH looks more the ballpark. – Neil_UK Mar 12 '20 at 08:20
  • I hadn't calculated that. 3.5w doesn't seem too bad with a heatsink. Either way, it seems that I am looking at it wrong. I was thinking 2.2uH was tiny. I have no idea what ripple I need. I was thinking the less ripple then the higher I could push the led because the peaks wouldn't be so high. Is 0.45a peak to peak not a lot? – Wizzard_Dr Mar 12 '20 at 08:49

0 Answers0