You can see from your scope traces that a large current is drawn just as the voltage reaches it's peak value - this is normal for a rectifier feeding  a capacitor. Put the inductor BEFORE the capacitor and it will help a bit.

Try simuating your cicruit (LTSpice is free from www.linear.com).

Why do you care about the power factor - are you planning to build a lot of these ?

Is cost an issue ?

Michael Kellett

Just found an interesting paper which addresses your problem and explains things rather nicely.

www.coilws.com/Publications/ImprVF.pdf

Michael Kellett

Thank you for your answers.

The application is a LED substitute for flourescent tubes, and hence supposed to fit inside an existing flourescent tube fixture. This means:

1. The input capacitance and inductance cannot be changed. These are mounted in the fixture, and therefore not a part of my actual printboard.

2. Power factor matters, as these are supposed to replace, say, 100 flourescent tube in industrial (or private) buildings.

3. Cost is an issue, but only to the extent that implementing an active PFC circuit (i.e. a few dollars worth of components) in the input of the PCB is no problem at all.

I did actually add a ValleyFill circuit to the existing PCB, but it did not improve PF, when measured.

I happen to have read something about harmonics in the current, not present in the voltage, from e.g. switching decrease PF (http://www.cip.ukcentre.com/Power%20Factor.htm).

Do you know if this is true, and could it be the main reason for the low PF measurement (on Power Meter)?

Thanks once again for your time!

David Larsen

Hello David,

Certainly you can have harmonics in the current which are not present in the applied voltage. If you are designing  a circuit for mass production you need to invest in some serious test gear - starting with something which can measure harmonics like the

AIM-TTI INSTRUMENTS - HA1600A

(one day they'll add a paste as text function to this awful editor)

With a line conditioner this will cost you about £2000 (Farnell sell them).

If your valley fill cicruit didn't help then you got something wrong !

If your plan is to offer  a substitute for a normal fluorescent tube that uses LEDs and goes in a standard fitting  I don't think it's comercially viable - LEDs are only of comparable efficiency with florescents so the pay back time will be very long - BUT - the reflectors etc are designed for a tube which radiates all round and will not be optimum for LEDs so the efficiency of a LED tube in a fluorescent fitting may well be seriously compromised.

I can't really help you more without an answer to the how many - a good design for mass production will use custom silicon. If it's for a special purpose it may not. If you are up to 50W or more into the LEDs you will have to use  a conventional power factor corrected switching supply.

Michael Kellett

Hello again,

I am terribly sorry about the late reply, but I have been rather preoccupied the last few days.

So what I hear you are saying is: Essentially a Valley-Fill solution should suffice?

And active PFC might actually be money out the windows?

It is definitely true that actual power saved is not really that much, however, depending on production cost (I am not producing these myself), lifetime of tubes might make such lighting preferable to some people.

Apparently there is an actual demand for this!

As an example (in reaction to your post), a single order of 40000 units has just been placed.

Custom silicon is probably not the optimal solution, since the tubes are already in production and basically just need an optimized PFC due to (relatively few) large scale implementations.

With money being tight, though, £2000 equipment is probably out of the question right now, although I do appreciate the need when working more permanently with these types of problems.

Hopefully, I'll be able to obtain a solution using normal a Power meter!

Thank you once again for your help og advice!

David Larsen