Current limit issues with high voltage LED Boost Converter circuit using HV9911

Thread Starter

mvElectron

Joined Aug 23, 2019
9
The project here is an LED boost driver, for the purpose of running a 72vDC LED string (a 4' 18W "fluorescent" LED tube) from a 48vDC nominal battery bank (40v-60v design range), using the Microchip (Supertex) HV9911. Yes, I've tried posting on the Microchip forum, but only ended up with the forum coming up with ridiculous server errors and rejecting my post...not to mention that there isn't much life on the Microchip "analog design" forum category.

I've had PCBs made via JLCPCB.com, based on the sample design circuit for the Microchip (formerly Supertex) HV9911, with lots of help from the Supertex AN-H55 application note. Technically, the circuit functions, as it can easily light up the 72v LED tube from 12vDC (current limited around 0.8A by my power supply). However, the problem is that neither of the HV9911 current limiters seem to be functioning. If I turn the power supply current limit up, the LED tube gets brighter. But I designed this circuit to run from 55v, not 12v--meaning that at 12v, the required input current will probably blow the FET before reaching the 18w output.
I tried setting the input voltage to 36vDC, and then turning the driver on: the LED tube flashed, and the driver stopped working (likely blew the FET, as I still have voltage at 2: VDD).

What I find weird is the potentiometer for setting the LED current (yes, I know that I should have put a resistor on the top lead for better range!): If I set it low enough to regulate the LED current at a low level (which should avoid the current limit on my power supply, allowing for higher input voltage)...the HV9911 does not even attempt to turn on. However, if I set it far too high, the HV9911 starts...but obviously hits the current limit of my power supply.

Several things I've wondered about:
  • Is there an oscillation on the LED current feedback circuit? If I physically touch the LED current potentiometer with a screwdriver, the HV9911 instantly faults off (requiring a restart). I've tried a capacitor from the pot wiper (15: IREF) to ground, but with no change. (It is worth noting that that is specifically discouraged in the HV9911 datasheet.)
  • Is the LED current feedback resistor too small? I used 0.15 ohms for a quantity discount with the FET current sense resistor, though the AN-H55 note calculations specified ~0.6 ohms. (Seemed like a waste of power to me.) Like maybe there's a "lower threshold" on the current sense feedback? I've tried replacing this with a 0.5-ohm resistor...but again, no change.
  • I know that the LED output cap should go to GND, not the top of the current sense resistor. I've tried changing this, but again...no change.

Here's the schematic:
1584029137898.png
(PWM control circuit off the left; it is not currently implemented. I'm just manually shorting PWMD to VDD to turn the chip on. Also note that I have renamed pin 16: IFDBK -> FDBK)

Any suggestions as to what is amiss?
 

BobaMosfet

Joined Jul 1, 2009
1,220
Can you repost the schematic- a little bit off the top is clipped. And thank you for a schematic and a very sensible description.
 

Thread Starter

mvElectron

Joined Aug 23, 2019
9
(edit: redid the picture again, bottom was also cropped. Added a few more clarifications and specs.)
Sure can, though the only part missing on the top is the "F1" designator for the input fuse.

1584032648400.png
VIN, GND: power in, 40-60vDC (technically could go up to 68v before the TVS clamps)
L+, L-: LED output
R7: LED current adjust
R12: MOSFET current sense resistor
R13: LED current sense resistor
Q1: switching FET, rated 100v @ 4.5A continuous (14A peak). Very tiny, about the size of a SOT-23, with on-resistance of 65milliohms @ 4.5v gate threshold. Only rated 1W power dissipation.

Off the left side is a PWM controller that I have not soldered into the board yet, so isn't applicable.

I will note that I have successfully made a buck LED driver with the HV9910 (again, 40-60v input, currently driving 38v 25W LED fixtures. They work like a charm...currently have about 6 of those drivers in use, no issues.) Just this boost driver that's giving me headaches.
 

Attachments

Last edited:

BobaMosfet

Joined Jul 1, 2009
1,220
Just an observation:

Datasheet shows a Cin capacitor on Pin 1 for Vin:

1584037768121.png
Notice that it is POLARIZED. In fact all caps shown in the above 'typicals' are polarized. A question I have is- where are your calculations. How did you calc the component value for each component needing calc'd? I know for my part, when doing this, I create a spreadsheet so I can take everything into account.
 

Thread Starter

mvElectron

Joined Aug 23, 2019
9
Here's my calculations from following AN-H55:

Minimum input voltage: 40v
Maximum input voltage: 60v
Maximum LED string voltage: 90v
Minimum LED string voltage: 60v
Maximum LED current: .5A
Minimum efficiency: (90%)?

Step 1: Select the switching frequency
-> 200KHz should be fine

Step 2: Compute the maximum duty cycle
Dmax = 1 - ((90%[efficiency] * 40vInMin) / 90vOutMax)
-> 0.60

Step 3: Calculate the maximum inductor current
IinMax = ((VoutMax * IoutMax) / (90%EffMin * VinMin)
-> 1.25A.

Step 4: Compute input inductor
L1 = (VinMin * Dmax) / (0.25[ripple] * IinMax * f)
-> 384uH. Standard sizes are 330 or 470uH. Hmm...

To get 90% efficiency, power loss has to be 2-3% maximum.
For 3% loss, calculate the maximum wattage loss...
Pind = 0.03(%) * VoutMax * IoutMax
-> 1.35W

So...DCR < (0.8 * Pind) / (IinMax * IinMax)
-> DCR < 0.7 ohms

Saturation current has to be at least 20% above the peak current...
Isat = 1.2(120)% * IinMax * (1 + (0.25 / 2.0))
-> 1.7A saturation current MIN.

384uH (330uH probably OK), at least 1.7A SATURATION current, under 0.7 ohms...

Step 5: Switching FET
Maximum voltage across FET is equal to the output voltage; give a 20% safety margin
Vfet = (120%) * VoMax
-> 108v

RMS current through FET is...
Ifet = IinMax * (sqrt(Dmax))
-> 0.96A

For best results, it has to have a current rating 3x that...
...minimum gate charge < 25nC
(OK, maybe my 100v FET is a little undersized, but that's not the problem I'm having)

Step 6: Switching diode
-> just pick a Schottky, at least 1A.

Step 7: output filter capacitors (REQUIRED)
-> skip the math, just use 2.2uF, 100v metal film cap, NOT ceramic
Or one of the 100v, 100uF electrolytics on hand...?
(well, I am missing a special cap here. Trying the 100uF, 100v electrolytics I have on hand.)

Step 8: Disconnect FET
would like to leave it out, but it assists with dimming. AND saves capacitor life.
at least 100v, 1A. Other specs not important.
---left it out---

Step 9: Input capacitor
...just use a 10uF, 100v cap + 0.1uF polyester.

Step 10: Timing resistor
(1/Fs) = Rt * 11pf
cheat, use their answer: 453K

Step 11: Current sense resistors
(LED) Rs = 0.15W / (IoutMax * IoutMax)
-> 0.6-ohm calculated; USE 0.15 OHM.

Probably could use a 0.1 or 0.2 ohm resistor and calculate the difference

(FET) Rcs = 0.25v / (1.125 * IinMax)
-> 0.18 ohm

Use 0.15 ohm, 1/2-watt resistor for a little input current headroom.

Step 12: Selecting reference voltage resistors
Rr1 & Rr2 = trimpot, 25K (per spec)

Step 13: Slope compensation resistors
DS = (VoMax - VinMin) / L1
-> (90v - 40v) / 330 = .151[A/us]

RSlope = 49.9K (default value...?!)
Rsc = (RSlope * DS * 1000000 * Rcs) / (10 * Fs)
-> (49900 * 0.151 * 1000000 * 0.15) / (10 * 200000) = 560 ohms

Step 14: Setting inductor current limit
Vclim >= 1.35 * IinMax * Rcs + ((4.5 * Rsc) / RSlope)
-> 1.35 * 1.25 * 0.15 + ((4.5 * 560) / 49900) = 0.303v

Use 22K / 7.2K for 0.308v bridge from 1.25v ref. Over 25K total, so we're fine.

Step 15: Capacitors...
VDD = 1.0uF unless FET's Qg > 15nC (then 2.2uF)
REF = 0.1uF
(well, umm...I have a 0.47uF cap here. Need VDD to fall rapidly when power is lost for the PWM controller)

Step 16: Overvoltage trip point (Rovp1-2)
120K / 1.5K = 1.25v reference "tripped" at 101vDC
NOTE: Will vary +/- 3% due to reference voltage changes. Leave headroom!

Step 17: sensory overload ;-)
Can't figure out what "S" is, so just use their numbers...
Cc = 2.2nF 50v
Cz = 6.8nF 50v
Rz = 20K


Hmm. Maybe I made a few errors from my calculations. VDD cap might be an issue...
 
Top