Varying current

[Flyback]

Advert says 32v max, but binning and labelling doc says vf max is 3.75 (albeit at 25c), so to be safe , I think you've gotta consider total voltage of lamp = 10* 3.7 + cable drops.....so say 38-39v...meaning you need an input voltage say at least 4-5v above that.....did you get an input voltage source above 44v?
Is the PCB really copper?.....aluminium is more usually used, copper is usually just used for the tracks. I think aluminium actually conducts heat better than copper.

 

[alec_t]

I think aluminium actually conducts heat better than copper.

Actually copper is ~50% better than aluminium.

 

[Flyback]

sorry, I meant aluminium is better at radiating heat than copper?

 

[ChrisP58]

Actually copper is ~50% better than aluminium.

Actually, it depends on how you are comparing the two.
Copper is a better conductor for both heat and electricity if compared by volume.
Aluminum is better for both if compared by weight.

As for radiating heat, the surface finish is the dominant factor, not the material.

 

[Flyback]

I think copper oxidises and becomes "dull coloured" , and less radiative, whereas aluminium is less that way inclined.

 

[Gregory]

It is Alliminum .
The batteries are 37v but initial charge is 40 v but this voltage only last for about 15 minites then it settles at 37 volts supply. The current must be regulated . The forward current is stated at max 10amp but this is to much causing the led to fail .you have indicated that the voltage in the batteries should be 40v or more . The forward voltage on the speckes is 37v
I am gathering the information to allow me to start to building the buckboost that you indicated in one of your previous post. If you think this is the better way to go could you let me know.
As I am not a design engineer As I appreciate all the help .I want to start making the project .

 

[Flyback]

the "binning and labelling" document does say 3.75V is the maximum Vf, (at 350mA), so with more than twice that in the leds, you have to suspect a potnential max string voltage of more than 37v, so yes, I believe you will need a battery of 40v or more.
Having said that, if its a real pain to change the battery, then i can run a design for your 37v as vin and you can just hook it up and see...you never know, your led lamp may have low voltage leds on it, and may run in a hotter ambient than the datasheet measuremental leds.

IVE RE-DONE THE EXCEL FOR VIN=37V...ITS WORTH A TRY.. (Sorry about caps)

Inductor is 100uH and 7 amp rated.
FET is 50v min and 6 amps min rated (HEATSINKED TO220)
Sense resistor 33milliohm and 2w, non inductive.

Give it a try, the lower vin certainly reduces the stress on the power diode.....just use a schottly diode, to220, and rated 2 amps, 60v plus.

remember to adjust the pot so the voltage at the LD pin of the hv9910b is 0.22v max....in fact, to get round the comparator delay, best make it 0.2v at the LD pin....if you wanted to be really brill, you could put an opamp ideal diode clamp on the LD pin so its voltage cant go any higher than 0.2v

so there it is, in the first place, not worth changing the battery , stick with what you've got and go for it.

 

[Gregory]

It is a problem changing the batteries as they come in a pack and room is the problem

 

[Flyback]

ok stick with the 37v sounds best

 

[Gregory]

Is the circuit you recommend
Which one would you recommend

 

[ChrisP58]

I think copper oxidises and becomes "dull coloured" , and less radiative, whereas aluminium is less that way inclined.

Oxidized copper is a better heat radiator than polished copper.

See: https://www.engineeringtoolbox.com/emissivity-coefficients-d_447.html

 

[Flyback]

I would recommend using the HV9910B.
CIRCUIT IS ON PAGE 1 (sorry about caps)
**broken link removed**
attach RT top the gate and RT, not to ground...ie, use constant off time.

you could use the zxld1370, but it is slightly more complicated.

don't forget the pot or potential divider into the LD pin.
Also, use a resistor of about 10k from fet gate to ground.
RT resistor = 68k
Also, the current sense resistor is 33 milliohm, 2w min, non inductive....likely to be a thru-hole part.
Use a zero ohm resistor to the gate. (in series)

RT is surface mount and small, so is the gate resistors.
Get the sense resistor close to the hv9910b.

 

[Gregory]

I don't like to ask , but could someone draw up a circuit digram using 16TC ZXLD1370 EST To suit the requirements of the led's and battery supply

 

[Flyback]

The thing is , I thought it was already agreed that you wanted to do Buck, due to lack of experience?
You cant do Buck with zxld1370 because your vin is too low (see fig 9 of datasheet).

You could do buckboost with the zxld1370, but...
If you want buckboost you will have to get your overvoltage protection circuit right or else you could blow stuff up.

zxld1370 datasheet
**broken link removed**
Remember also that zxld1370 has a minimum switching frequency of 300khz, so your inductor core losses and fet switching losses will be quite a bit higher. 300khz is quite a high switching frequency for a beginner.

If you use hv9910b you can do Buck from 37VIN.

 

[Gregory]

Thank you for that information I will go with the HV1190B. I will draw the circuit up and post it with to make shore that it is correct befor going to copper. I will do the circuit board by hand.
Do I discard the PWM that I am using at present.

 

[Flyback]

yes I wouldn't use that pwm item with the hv9910b circuit.

 

[Gregory]

What is the difference between HV9910B and HV9910 .When I searched for the IC in question the supplier had advertised as the above .
You have written there is a 0 ohm resistor in the circuit that would be a piece of wirer.

 

[Gregory]

I got a circuit for led driver would this do the job

 

[Flyback]

looks good....but.
Make sure to put a ceramic cap (10n say) at the LD pin. (...oh you already have, nice work)
Also, make sure you put a cap right at the vin pin....eg a ceramic of a couple of microfarads.
Also, your input caps are 35v rated, that is not enough.
Remember a ceramic cap should have a voltage rating at least 20-30% more
than the rail it sits across.
Put a 27k (apprx) signal , eg 0603 resistor or 0805 at the fet gate, from gate to ground.
Have a zero ohm resistor in between GATE pin and gate of fet...that's just in case you need to damp the switching transistion....have an actual zero ohms resistor,
and if you need to damp, because of noisiness, then make it a 4r7 say.
ensure c3 is near the pin.
Put a cap (ceramic) of 100n, 50V min, across LED- and LED+, but make it near the Power diode and not near the leds.....then take the wires to the leds off as a twisted pair....actually no, make this cap at least 63v rated...because in the event of someone accidentally yanking the load connector out whilst the led driver is powering the load at full power, then there will be an overvoltage ringing on this capacitor, so make it 63V rated.
Make the pot a multi turn pot if possible.
When you power it up, first have the pot to give 0v at LD pin, then wind it up to your max power.....which I think we calculated was 0.2V?... see calcs previously.
Remember the inductor is 100uH....ok you can use 220uh and the 124k off time resistor if you want.......it does bring the switching freq near the audible range, but is maybe better for noise reasons etc so go for it if you want.....220uh and you may get too much winding loss though and core loss...depending)
Yes you absolutely must use the "HV9910B" from supertex (though I think supertex is microchip now?)
Don't use the Clare one , or any other, it must be the HV9910B only.
the other ones aren't constant off time.

Did you follow the power ratings for the sense resistor?
Check the part number for the input caps to make sure enough ripple rating, though we know that some of the ac will be drawn from the batt....ensure the input wires for power, if possible are twisted or at least run side by side.
Use a bigger shcottky, bigger fet.
Presumably you are going to wind the inductor yourself, using which core set? (must be ferrite)

regarding THE INDUCTOR....
https://www.epcos.com/epcos-en/5294...ites-and-accessories/pq-cores-and-accessories

You could use 13 turns on the PQ50/50 core of the above link.......but it needs an "overall" air gap of 0.5mm...the gap must be in the centre leg only.....so you shave off 0.25mm of each of the core half centre leg surfaces....to give the overall 0.5mm air gap. Otherwise if you don't use the gap it will saturate...I doubt you will be able to put the gap in yourself...it is very difficult as ferrite is rock hard and brittle...a transformer winders would need to do it for you.
So that's 13 turns (FOR 100Uh) of enamelled copper wire round the gapped PQ50/50 core.

The PQ5050 CORE with 0.5mm gap gives you up to 10.9A with the 100uH...this gives you nice safetymargin........if you want smaller inductor give me a shout and ill do it again with the samller members of the PQ core family.....
The inductor (gapped) calc sheet is as attached..
Use minimum 1mm diameter enamelled copper wire.

If you cant afford to send the core halves off for gapping, then I suppose then you have to just put spacers in between the core halves to give an overall 0.5mm gap.
Remember with such spacers, its the total gap you have to get to 0.5mm, not every individual gap.
If you want you could just get an inductance meter and do it imperically....keep gapping it and measuring the inductance...but beware, if you gap less than 0.2mm then you could saturate when at max current when in the lamp.

The alternative is just to send the core off to a winder and ask for inductance of 100uh and rms current 7amps (say). Say it should not saturate up to 7.5 amps.
You don't need to use litz wire as your frequency is so low.

 

[Gregory]

Thank you very much for your persistence and help .
When I have drawn up the circuit in the correct sequence I will post it to see that it is correct.
I am a bit slow doing the circuit but will give it a go.
What does the B mean on the part number HV9910B