Varying current

[Flyback]

I don't know what the B means.....I just know that all the other variants of this chip don't do constant off time. I used the Clare part once and spent an hour wondering why the board didn't work, then looked at the bom and saw it wasn't hv9910b....changed it and it worked.
Give us a shout if you need the inductor smaller....the calcs for that are a few minutes at most.

Interesting to see you used a 220uH inductor (instead of my 100uh recommendation) , and that you also adjusted the RT resistor to suit this.....this suggests that you have more knowledge about this than I first thought?

INDUCTOR:
To get the inductor smaller, use a Epcos PQ3535 core (N92 material) and put in a 0.6mm gap, and have 19 turns of 1mm wire. That will give you 100uH , start of saturation is 8.35amps.

epcos pq3535 core:
https://www.epcos.com/inf/80/ds/b65881a.pdf

 

[Flyback]

Just had a look on digikey to see if there was any inductors off-the-shelf, and to my surprise, there is...
Bourns 1140-101k-rc:
https://www.bourns.com/data/global/pdfs/1140_series.pdf

so this part will do the job for you, if its not in stock, then just search on digikey for

"Inductor" -> "FIXED INDUCTORS" ...select something (in ferrite) that doesn't saturate up to 8 amps, and with say less than about 0.04 ohms of resistance or so.

 

[ronsimpson]

100uH and 220uH
Are you running slow, say 20khz?
The duty cycle is very high near 100% so why not use a smaller value of inductor?
I think, during ON TIME there is very little voltage on the inductor.
During OFF TIME there is 35V across the inductor but this could be 1uS or something even shorter.

It is hard to find a shielded 100uH 10A inductor.

I would push the frequency up if possible to use a smaller inductor.

 

[Flyback]

The thing is , Gregory has no experience with switchers, so his layout might not be that great.
So I went for a switching frequency of 32khz.
This means that Gregory will be able to put in some series resistance with the gate and slow down the drain transistions.
Also, the HV9910B has a weak gate drive anyway, so cannot drive at high frequency as there would be too much switching loss.
We've found an 100uH, 8 AMP inductor off the shelf for Gregory, a bourns one...so he doesn't have to make that.

So this is the spec for Gregory
Vin = 37v
vout = 32v
iout(max) = 6A.
OFF TIME = 3.6us (RT = 68K)
Inductor = 100uH
...this off time gives an f(sw) of 32khz.

However, sometimes greg will have a lower output current, and a lower output voltage,
This will mean an increase of the frequency above the 32khz quoted above.

When Greg adjusts the led current to just 1 Amp, the load voltage will be around 23v, and in that case, his switching frequency will then be 100khz, which is well high enough for the weak gate drive of hv9910b. So I believe we should stick with 100uH and 3.6us off time.
-And remember, when the lithiums are fresh they will be 42v and not 37v, so the switching frequency will be higher still, so I believe we should stick with the 3.6us off time and the 100uH inductor.

HV9910B LED DRIVER:
**broken link removed**

 

[Cicero]

Just to throw another option into the mix, if in fact the switch mode doesn't work out. Gregory could do a linear solution more easily built. This is a fellow forum members design, but I can't remember his forum alias unfortunately, so apologies for that. https://angeliselectronics.blogspot.com/2014/11/pictures-ill-explain-later.html

 

[Flyback]

regarding linear solution, as you know, this must regulate the current from the maximum battery voltage with the minimum vf led bank, and that means a lot of dissipation.
I reckon the HV9910B solution is easily manageable, specially now that we've identified an off-the-shelf inductor. I'd be amazed if the hv9910b solution did not work.....the gate drive is weak so the drain transistion will be relatively slow, and can be slowed further by making the gate resistor higher. Basically, as long as Greg gets the power components in close together, and starts off by getting the sense resistor and the hv9910b close to each other, then i'd be amazed if it didn t work. The only problem is the pcb layout, but with the eagle program, anyone who knows the basic structure of a pcb, and knows how to use windows, can use eagle. That is the beauty of eagle, (IMHO) it is the only pcb layout program which is made to make a layout engineers work easy.

 

[Cicero]

regarding linear solution, as you know, this must regulate the current from the maximum battery voltage with the minimum vf led bank, and that means a lot of dissipation.
I reckon the HV9910B solution is easily manageable, specially now that we've identified an off-the-shelf inductor. I'd be amaxed if the hv9910b solution did not work.....the gate drive is weak so the drain transistion will be relatively slow, and can be slowed further by making the gate resistor higher. Basically, as long as Greg gets the power components in close together, and starts off by getting the sense resistor and the hv9910b close to each other, then i'd be amazed if it didn t work. The only problem is the pcb layout, but with the eagle program, anyone who knows the basic structure of a pcb, and knows how to use windows, can use eagle. That is the beauty of eagle, (IMHO) it is the only pcb layout program which is made to make a layout engineers work easy.

Appreciated, just thought I'd add it in as a side note. Since the adoption of the switchmode as the preferred solution I've been staying out of this thread for a while, and just monitoring from the outskirts to see the outcome. I'm very interested in how it all pans out in the end.

 

[Gregory]

I have tried to use the cad system but have failed. I will post the circuit I drew with values and Ic drawn in by pen. Could somewon help me out with a cad circuit and board layout . I am not shore weather I have the circuit correct . Please change
Thank you

 

[ronv]

Maybe there is an easier way.
How many of these do you need to build?
Do you mind if they are not quite so bright?
Do you still need a dimmer as well?
Do you have it mounted on a big heat sink?

 

[Pommie]

Just caught up with this thread. Is there any reason the origin arrangement could still work with the addition of 5 x 8A heatsinked diodes in series with the battery? Or, to put it another way, can anyone see why this simple solution wouldn't work?

Mike.

 

[Cicero]

I have tried to use the cad system but have failed. I will post the circuit I drew with values and Ic drawn in by pen. Could somewon help me out with a cad circuit and board layout . I am not shore weather I have the circuit correct . Please change
Thank you

Bummer Greg, I was concerned it would be too complex from the get go. I still think you need to go the linear route but I've been staying out of the thread giving it the benefit of the doubt.

Just caught up with this thread. Is there any reason the origin arrangement could still work with the addition of 5 x 8A heatsinked diodes in series with the battery? Or, to put it another way, can anyone see why this simple solution wouldn't work?

Mike.

Mike, as for 8 (or more if need be) heatsinked diodes, not a bad thought and I like it, but its the same as putting a power resistor (or two) into it essentially and just dropping the input voltage accordingly, thereby dropping the current. Which is what I suggested towards the start of this thread, be it by a set amount though if it were diodes. It can work, both the diodes or resistors, and would be FAR easier to prototype and test for just a few dollars than trying to build a whole new switching regulator.

 

[Gregory]

Yes there is a big heat sink and it is sealed in a canister which is in water . Yes a dimmer is required and the maximum brighten is required. I required 4 of the boards.
I apologise but I did not post the circuit but will now.

 

[Gregory]

Cicero could you post a circuit for me as to what you suggest. Will this prevent the led's from flickering at low voltage.
What size heat sink would you recommend .
Does this mean that I can use the original control system.
Thank you greg

 

[Flyback]

looking at your schem of hv9910b you have a cap from vin to input_positive, and that is not good.
The Eagle (cadsoftusa) cad system is very easy once you know how.
It has more beginners tutorials on the web than any other.

Try the tangentsoft.net tutorials for it
He does one for schematc drawing
one for pcb layout,
and one for putting the components in the library (drawing the footprints)
They are very good

Do you know the basic structure of a double sided pcb?.....if you do not then you will never be able to use any package.

A double sided pcb is a piece of fibreglass sheet, with metal tracks and pads on each side, and holes which are copper coated going from one side to the other, and all over the pcb, except on the pads, you have a film of material called solder resist, which weirdly repels solder and stops it from forming unwanted shorts between the pads.
There is also the silkscreen lettering.
Each thing, (like eg the solder resist), is deemed to be a "layer".
A little confusing is the fact that there is not a solder-resist layer, but instead a "solder mask" layer, and "solder mask" is those bits where "solder resist" doesn't exist.

Here is some eagle tutorials...
schematic creation in eagle

creating pcb in eagle

creating circuit symbols

creating component footprints

here is the route page
https://tangentsoft.net/elec/movies/

Here is the schematic...attached
Also attached is a simulation of how the HV9910B led driver will work, in LTspice, a free download from linear.com

 

[Flyback]

Cicero could you post a circuit for me as to what you suggest. Will this prevent the led's from flickering at low voltage.

I think a linear solution is likely to have a higher dropout voltage than the hv9910b switcher solution. (ie, the linear solution will need the vin to be higher in order to drive a certain current through the leds, than the hv9910b smps solution)

eventually, as the vin drops so low, you will get flicker whether linear or switcher.....its worth remembering that linear solutions as in linear voltage regulators also have a feedback loop in them.

 

[Cicero]

I think a linear solution is likely to have a higher dropout voltage than the hv9910b switcher solution. (ie, the linear solution will need the vin to be higher in order to drive a certain current through the leds, than the hv9910b smps solution)

eventually, as the vin drops so low, you will get flicker whether linear or switcher.....its worth remembering that linear solutions as in linear voltage regulators also have a feedback loop in them.

You're right, and I do like your buck method dont get me wrong. Its definitely the most efficient, and the majority of us would've probably gone that route. However considering Gregory's level, it may be a slight step too big. The pure resistor method is a little crude, and you'd have to design for the max voltage in, and then your minimum voltage would then be affected like you say - same with the diodes in series as well.

I posted that "Electronic load simulator" ( https://angeliselectronics.blogspot.com/2014/11/pictures-ill-explain-later.html - which I cannot take credit for, it was a fellow forumite who came up with it - forgive me for not remembering his username). Its similar to one of your suggestions of a feedback power transistor circuit, right in the beginning of this thread. I've posted a simulation which is an adaptation of it, but could be a starting point for a linear solution if needed.

Cicero could you post a circuit for me as to what you suggest. Will this prevent the led's from flickering at low voltage.
What size heat sink would you recommend .
Does this mean that I can use the original control system.
Thank you greg

Just have a look at this as a suggestion...

The LED is simulated as a 30V zener diode, max input voltage of 42V to the circuit, and the current from 0-6A by controlling the "ADJUST" point from 0-5V (this can go higher for more current, and you could use a simple resistor divider with a pot for that) on the opamp input. Each line of FET's takes a third of the current - which drops the power/heatsinking needs of the individual FET's.

This isn't the most efficient, and it'll waste energy in heat, but it is simple to build by hand and can be wired in easily on veroboard with very readily available parts.

EDIT: Old attachment removed

 

[Flyback]

I like the circuit Cicero, though as you know, it starts dropping out at vin=39v, and is significantly dropping out at vin = 38v
On the other hand, the simulation here shows the switcher going good down to vin=32v for a 30v , 7A load.
I realise that you already appreciate this though, and you're offering is certainly easier to layout, and does not need a PCB to be layed out, the reasons no doubt that you presented this nice linear solution.

 

[Cicero]

I like the circuit Cicero, though as you know, it starts dropping out at vin=39v, and is significantly dropping out at vin = 38v
On the other hand, the simulation here shows the switcher going good down to vin=32v for a 30v , 7A load.
I realise that you already appreciate this though, and you're offering is certainly easier to layout, and does not need a PCB to be layed out, the reasons no doubt that you presented this nice linear solution.

Hmm, shouldn't do that, the FET's should just switch fully on, maybe the zener isn't the best method to simulate the LED's. Have a look at this one...I just strung six 5V LED's together instead.

That switcher simulation of yours is awesome, saved it for myself, and it looks easier than I thought it would be. I guess its just the PCB and layout thats a potential problem.

 

[Flyback]

yes, I haven't yet had time to see why the linear circuit you kindly provide started dropping out at 39V.....is it that the opamps are not "input and output voltage swing" rail to rail?
Or is it something to do with the drain voltage of the fet going below the gate voltage at low vin? Surely the current sense resistor voltage is only 0.5v, and the fet's DS drop should be able to be controlled down to 0.5V say?, so where the other few volts is going......I will look into it in good time.

 

[Cicero]

yes, I haven't yet had time to see why the linear circuit you kindly provide started dropping out at 39V.....is it that the opamps are not "input and output voltage swing" rail to rail?
Or is it something to do with the drain voltage of the fet going below the gate voltage at low vin? Surely the current sense resistor voltage is only 0.5v, and the fet's DS drop should be able to be controlled down to 0.5V say?, so where the other few volts is going......I will look into it in good time.

Have a quick look at the latest circuit I posted above. It doesn't have that drop out problem.