Inductor selection in a Boost switched mode power supply?

[jwhitmore]

I'm starting to look at a boost switched mode power supply for a few projects, one of which involves Valve/Tube preamp. I've not selected a controller IC as yet but just picked a datasheet to look at, to get started [1]. One of my first questions was what value of inductor I need. Since the output voltage is dependent on duty cycle of the switching I assume the required Inductance is dictated by the input and output power requirements. Page 13 of the data sheet [1] gives a formula for the ideal inductance of:

L = VOUT / (4 x IOUT x fOSC)

The controller IC switches in the range 100KHz to 500KHz so if I say 300KHz and my Valve/Tube requires 250V and say 10mA then plugging in those numbers gives me an Inductance of about 0.02H or 20mH That seems like a quite high value. Because of this high value the inductor will probably have a quite low current rating. Now given the Valve/Tube preamp application then the output current is very low, (10mA), so that's fine but both the input and output current will flow through the inductor. Have I got that right? I think that's correct, the inductor is a bridge between the input and output side of the boost. If that's correct then the output is not a problem, but the input might well be. My output power is 250V/10mA so 2.5W if I was running this off an input voltage supply of 9V I'd need to pull an input current of 2.5W/9V or nearly 300mA. That's a bit more then the avarage 20mH inductor can handle.

Perhaps I have that wrong, or I've made assumptions which aren't valid? Obviously one solution is to use an input voltage as close as possible to the output voltage of 250V but I've taken an extreme value of 9V just to get an idea about this. Additionally this project is for a guitar pedal and the standard power supply in those effect pedals is 9V, so as an example I used it for some rough calculations.

[1] https://www.farnell.com/datasheets/2345060.pdf

[thm_w]

22mH 350mA: https://lcsc.com/product-detail/Power-Inductors_TDK_B82791H2351N001_TDK-B82791H2351N001_C117734.html
22mH 400mA: https://www.digikey.ca/product-detail/en/murata-power-solutions-inc/1422604C/811-3849-ND/5798114

If you go up to 500kHz then recommended inductor drops to 12mH.

There is another calculation in the datasheet for inductor current, but yeah its around 300mA+.

[T3sl4co1l]

Yeah it's a terrible equation.  Inductor value depends on Vin as well as Vout, and they don't explain what the constant '4' is doing (it happens to be the inverse ripple fraction, which is why they mention slope compensation).  It's correct for Vo = 2Vin.

Why are you looking at such an obscene voltage ratio?  Why not use a flyback?  Transformers are commonly available off the shelf.  Anything rated for a few watts, mains voltage stepped down to 5-15V, will do.  Works the same backwards as forwards, the transformer doesn't know the difference!

If you intend to run this off battery, first of all consider using lower load voltage, and current, and for that matter, battery operable tubes (or skip the tubes entirely, but I'm going to guess that's non-negotiable).  A few hours run time from a 9V battery might be achievable, but you'll want to also consider something with meaningful capacity, say for a whole evening's use, or several days?  Li-ion is great, with a few 18650 cells, or a pack or pouch, being quite affordable, easy to use* and able to run for quite some time, even in the same size as a 9V.

*For discharge, battery management is basically turning it off below 3V/cell or thereabouts.  Super simple.  You can get a chip that does this (the MAX668 has a similar UVLO threshold, in fact), or use a micropower comparator to disable your controller.  If you want a "gas gauge", you're going to want an IC for that (and probably another chip, an MCU, to translate it to something like an LCD display..).  For charge, it's still fairly simple, but you may be more comfortable with a chip to handle everything for you.

Tim

[Siwastaja]

Start with the universal physical truth:

V = L*di/dt

This simple equation catches the inductor behavior, and has the advantage of being an actual law of physics, not a "design equation". Rearrange it to your needs.

As an example:

di/dt = V/L

Apply 10 volts to an inductor of 1 Henry -> current rises at 10V/1H A/s = 10A/s

As you can see, if you apply higher voltage, the current rises more quickly. When designing a converter, you likely don't want that, so need a larger L to counteract.

Now a small practical task for you:

Let's say it's a buck dropping the voltage to half, or a boost doubling it. Duty cycle D = 0.5.
Say f_sw = 100 kHz.
How long is the period?
How long are the switch on-time and off-time?
Now say you decide the output current is 1A, and let's say you want the inductor current - the triangle wave - to ramp between 0.75A and 1.25A; current ripple is 0.5A, which is 50% of the output current.
Try to apply the basic equation now and solve for L.


On the other hand, getting 250V from 9V is almost impossible, or at least hugely inefficient and difficult, with a boost converter. Unfortunately you need a more complex topology. If you want to use boost for its simplicity and learning, nothing prevents you from cascading two separate boosts after each other, such as 9V->48V, 48V->250V

[jwhitmore]

Thank you all for your responses. That really helped me get to grips with this.

Cheers all.

[jwhitmore]

Hello again, I've been looking at this project, off and on, when I get the time, but still struggling with it. DC/DC boost conversion to power a Valve/Tube is not a trivial exercise, well not for me at least. This is potentially going to be a long post, as I have a few questions. Some might find that offensive, in which case move on, but I'm simply trying to understand a bit better.

Previously 'T3sl4co1l' mentioned a flyback design and I looked at that and it makes sense, as the turns ratio of the flyback transformer might get me close to the Voltage I need. In this design I'm trying to power a Valve pre-Amp so the output Voltage is sort of fixed at 250V DC. The input Voltage is not fixed, but I've mentioned 9V DC, as guitar effects pedals are usually powered by a 9V DC mains plug adapter. A lot of the analog pedals use less then 100mA, but I'm certain to break that by a bit. Now as previously mentioned 9V to 240V might well be asking too much so I can up the input Voltage to 18V DC or even 24V, using a Laptop power supply specifically to power this design. With that in mind I got a Boost IC to try breadboard this, selecting the ON Semiconductor's NCV887103 [1], for no other reason then is has a fairly wide input voltage range 3.2v-40V DC. That gives me a bit of wiggle room with input Voltage.

Having selected a Boost IC the next thing is a flyback transformer. I'm just browsing through the local farnell to get ideas. I came across the coil craft NA5920-ALD [2]. With a turns ration of 1:24 I'm thinking that's gotta get me close to 250V output. It seems to be rated for 100-400V so plenty good enough. It's only when I look at the data sheet [3] and see that the input voltage is 2V - 8V that I'm a bit surprised. In order to use that part I'd have to actually step down the 9V input voltage? I wasn't expecting that. I'll keep looking, but does that appear to be quite a low input voltage. I'll be the first to admit that I'm only at the wrong end of a learning curve when it comes to boost converters, and clearly I have assumptions that aren't valid. 8V might be a huge input voltage. Another question I have about that transformer's datasheet [3] is that it is 'designed to operate at 140KHz'. I got the NCV887103 variant of the Boost [1], because it can potentially run the duty cycle up to 93%, however it's opeating at 340KHz, unless I add other circuitary to feed it the 140KHz that the flyback transformer is 'designed for'.

I should forget that transformer and maybe go for the FA2470-AL_ [4]. It doesn't mention a 'designed for' frequency, and whilst it only has a turns ratio of 1:8 it does have an input voltage of 8.5V - 12V, (forget the laptop power supply)

I might leave this at that for the moment. I've read and re-read the datasheet [1] numerous times and I don't think it's a classic. Maybe it's just me, but it's not too bad until you get to the compensation network. Both figure 12 and 13, of [1], show a compensation circuit of a Cap C2 in parallel with a Resistor and a Cap, R2 & C1. No idea what that does but I'm hoping it becomes clearer when I have the other values and plug some numbers in the equations given in table 2. The obvious answer to that question is that those three passives create a compensation network. Like I say hopefully it gets clearer. Because the compensation circuit in the datasheet mentioned 'poles and zeros' I assumed the circuit was a filter of some description, but it's not one I recognise.

Thanks for making it this far. Apologies for the number of questions, as I have mentioned I'm at the wrong end of a learning curve.

[1] https://www.farnell.com/datasheets/2255395.pdf
[2] https://ie.farnell.com/coilcraft/na5920-ald/flyback-transformer-1-24-100-400v/dp/2526826?st=flyback%20transformer
[3] https://www.farnell.com/datasheets/2017944.pdf
[4] https://www.farnell.com/datasheets/1870409.pdf

[thm_w]

Coilcraft likes to make inductors/transformers for a specific IC, in the first case, LT8584. In the second case STLC3075. So they don't give you generic specs, they provide data only relevant to those ICs (to simplify design).

I would care about the:
- output voltage/pri:sec rating
- power rating
- inductance
- srf

I wouldn't care if it says 9V primary, as any insulated wire will easily handle 50V+.
140kHz is going to be close enough to 340kHz, you'll just have to assume that, as they don't provide any frequency graphs. That or look for a more generic transformer with full specs.

However, I can't say if NCV887103 is going to work with those transformers or how to lay it out, maybe someone else can advise.
https://www.maximintegrated.com/en/app-notes/index.mvp/id/1109/
https://www.ti.com/lit/an/slva649/slva649.pdf

The compensation network is basically a filter of feedback for stability, yeah. The full explanation is more complex but you got the idea.
https://www.digikey.ca/en/articles/designing-compensator-networks-to-improve-switching-regulator-frequency-response

[jwhitmore]

Thanks thm_w for all that info. I remembered by junk bin today. I tend to get items people are dumping and remembered that there's a 12V inverter in that bin that got water damaged. Now in that circuit is a small transformer that went from the 12V side to the high voltage. There's no identification at the moment, but tomorrow going to desolder it and see if I can make sense of it. See if I can stick it onto a bread board with the NCV887103 and make something happen. It's got three pins on one side and 5 on the other. Couldn't be easy with 2 and 2, but I'll buzz it out and then if I can't find identification maybe stick a signal generator on it and feed it a really low voltage sine wave. See what comes out the other end. I might be able to work out a turns ratio and got from there. It might have been effected by the water damage incident but who knows I might learn something.

If it's like some of the parts I've been looking at then the two center pins on each side will be dud, so two usable pins on one side and four on the other. The unit might have generated actual AC hence the 4/5 pins on the AC side. Mess with it in the morning and hopefully learn something.

Thanks again for your help.

[Benta]

A few very basic facts that you need to realise are, that a boost converter or a flyback converter are operationally the same thing. The flyback just provides insulation between input and output.
Next, boost or flyback converters need an inductor/transformer that can store the full energy needed at the output. Thus, it must have a certain size (switching frequency dependent).
This in contrast to buck or forward converters, where the inductor/transformer only needs to handle ripple+DC or the magnetising current.

It's a question of output power. Rule-of-thumb is that below 100 W boost/flyback are the economical choice. Above that, it gets interesting.

I wish you luck. I've designed plenty of switching PSUs, both low- and high-voltage. The big problems is, that it's not possible to do a closed-loop calculation of the inductors, it's an iterative process. And when it's on the prototype PCB, it doesn't behave as your calculations said it would.

Ah, power inductors, it's a love/hate relationship

[jwhitmore]

Thanks Benta, unfortunately this has not been a productive day. I breadborded the circuit, but it didn't give me the output I'd hoped for, actually any output. That's not fair. I started with an input Voltage of 5V and all I got was an Output Voltage of 5V, so achieved zero.

I might rubber duck this here and try figure this out. Whilst I desoldered the transformer I mentioned it didn't make much sense to me, as I wasn't getting any output from it, with a signal generator feeding it a 200mA 440Hz size wave. So I left that to one side. I do have a fairly beefy inductor [1] and a Power MOSFET [2] kicking about so decided I'd try them. To keep things light for the moment I used a 1M Ohm output Resistor in parallel with a 470uF Cap. Used a 1N5819 schottky diode to the output.

Given all that I used feedback resistors of 5k1 and 54k, hoping to get an output of about 15V. The only thing I'm missing is the compensation network. I turned it on without anything connected to that pin of the NCV8871 because the datasheet [3] has a table of formula for both CCM and DCM modes of operation. I'm not even sure what mode I'm in. I might have missed that in the datasheet but continuous or discontinuous. The design process is for a continuous mode so I assume I'm in that mode. I decided I'd just give the circuit an input voltage without a compensation network for the moment expecting a really bad 15V output. I think it's fair to say I got no output.

I'm struggling to make sense of this. The Boost chip is meant to turn on the N MOSFET if the feedback voltage is below a reference voltage (1.2V). I measured the voltage at the Vfb pin (0.2V) so the gate drive pin should be driving the FET on, but it's sitting a 0V. The Boost chip does have a VDRV pin which should have the gate drive voltage on it. That gate drive voltage is internally derived from the input voltage, according to the datasheet that should be 8.4V but is 0V.

All I can think is that even with an input voltage of 5V I've managed to blow this chip. I'll maybe order up another and look for a suitable flyback transformer and try in that configuration. Is there another name for 'flyback transformer' I read somewhere coupled inductor or something. I think this adventure will be on hold for a few days till I get a fresh IC.  Stupid chip.

[1] https://ie.farnell.com/murata-power-solutions/32220c/inductor-22uh-15-3a-th-toroid/dp/2062786
[2] https://ie.farnell.com/on-semiconductor/mtp3055vl/mosfet-n-ch-60v-12a-to-220ab-3/dp/2454583
[3] https://www.farnell.com/datasheets/2255395.pdf

[jwhitmore]

Cheers blueskull, unfortunately I can't find anyone who stocks it. I guess I could as TDK for samples, but there must be something suppliers stock which does the job. I had a look on banggood and found a DC to AC converter [1]. That makes me think that this is doable if I can only find a suitable transformer / coupled inductor. Maybe I should just get getting a few bits and winding my own. Actually perhaps that is the solution, figure out how many turns I need had current capability of the wires involved. I should be looking at ebay for an inductor winding kit. If  such a thing exists.

[1] https://www.banggood.com/40W-DC-AC-Inverter-Power-Supply-12V-Liter-220V-Step-Up-Transformer-Boost-Module-Support-in-Parallel-p-1424445.html?rmmds=search&cur_warehouse=CN

[jwhitmore]

Result in my shopping cart with digiikey.

I'm afraid I have one more question. I tried to go from 5V up to 15V with ON Semiconductor's NCV887103 [1], That for what ever reason didn't work out for me, perhaps I blew it. Anyhow wondering should I move to the LT1171 it might be easier for the novice to use. I guess I'll just get it and try it. Even if I can go from 5V to 15V be a major victory.

Thanks a million again for the pointer to that part

[1] https://www.farnell.com/datasheets/2255395.pdf