Power supply simulator. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Power supply simulator.

first previous 4 5 6 7 next last

Move Thread

LAN_403

Ash Small

Fri Jul 13 2018, 02:59PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Sulaiman wrote ...

In other words, I believe that the variable results that you got are almost entirely determined by the cores and number of turns.
=======================================
Abo ve a few MHz cores tend to be of low relative permeability so do 'leak' magnetic flux,
self-capacitance due to multi-layer windings or litz wire is usually avoided.

I guess I can test this by winding the remaining cores I have with ten turns, and seeing what the variation in readings is like. I'm not ready to wind another 500 grammes onto a core yet.

I plan to run some tests under load this weekend, maybe experiment with notch filters as well.

I'm guessing any capacitance in this application is only beneficial, as it adds to the total capacitance of the circuit.

In the photo above showing the two I recently wound, the one with one and two thirds layers is very unevenly wound, with some gaps on the outer face where it only has one layer. Surely this will have more leakage than the one that's evenly wound with four layers?

EDIT:- I seem to remember there's a reason why, when winding a flyback core, the winding is placed over the gap. It concentrates or focusses the fiels within the gap, improving the performance. With a distributed gap as you have in an iron powder toroidal core, it's important to have the windings spread evenly, but any gaps surely are going to allow the field to spread?

I will check the other cores I have with a few windings on each, and I'll also try to work out how much capacitance the one that reads 9.3mH has, I think I can do this using the 0.1uF and 0.01uF sections of the Tektronix test capacitor.

The smaller, 2" toroid I wound some time ago is reading 1.37mH. I plan on using this as a notch filter, and seeing what results I get. Plenty of tests to run this weekend

Sulaiman

Fri Jul 13 2018, 04:57PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Ash Small wrote ...
I seem to remember there's a reason why, when winding a flyback core, the winding is placed over the gap.

Exactly the opposite ...
try not to wind near the gap as the fringing flux cutting through turns causes emf reverse to the main emf, causing resistive losses in the copper near the gap.

Ash Small

Fri Jul 13 2018, 06:45PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Sulaiman wrote ...

Ash Small wrote ...
I seem to remember there's a reason why, when winding a flyback core, the winding is placed over the gap.

Exactly the opposite ...
try not to wind near the gap as the fringing flux cutting through turns causes emf reverse to the main emf, causing resistive losses in the copper near the gap.

Pages 3,4 and 5 of this application note cover this subject in some detail, it would appear we're both correct

It says that it some cases the winding should be over the gap, and in others, not.

I think it confirms that more of the field will have strayed outside of the windings in the choke I wound with 1 2/3 layers.

Ash Small

Mon Jul 16 2018, 07:15PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Well, I've tested it under load, and I'm getting 0.4V ripple at the second capacitor, which is exactly what the simulator predicts for a 9.3mH choke, so it doesn't appear to be saturating or swinging at all.

I've also built a notch filter using the 1.37mH choke, which apparently needs 1850uF in series for 100Hz. I tested it with with something like 1760uF, according to the marked values on the electrolytics, and got it resonating at 100Hz using the sig gen (electro's tend to be above indicated value, which presumably accounted for the discrepancy).

Then some very interesting events took place. I re-configured it in series, placed it in parallel with the second capacitor, and the ripple INCREASED. I played around with it, ending up with an indicated 974uF, which gave the least ripple, reduced from 0.4V to 0.16V, but don't understand why, according to the indicated values, it should be tuned for around 144Hz.

I've just ordered some low ESR polymer caps, which should arrive tomorrow, so I can see if they make any difference.

Have I missed anything obvious?

Shrad

Tue Jul 17 2018, 08:46AM

Registered Member #3215 Joined: Sun Sept 19 2010, 08:42PM
Location:
Posts: 780

maybe try a girator instead? noise floor reduction would be better but you'll have a voltage drop to account and some transistors to refresh...

Ash Small

Tue Jul 17 2018, 06:36PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Shrad wrote ...

maybe try a girator instead? noise floor reduction would be better but you'll have a voltage drop to account and some transistors to refresh...

I'd sooner just use passive components. I originally considered using a voltage regulator followed by constant current diode, but the voltage drop was going to be considerable, and I've read elsewhere that people don't get good results when using them for audio applications.

I think I have a plan, I've calculated that resistance of the choke is around 72milli-Ohms, and combined ESR of caps is about three milli-Ohms. I think by unwinding the choke, cutting the wire into three, and re-winding on a double core I can reduce DC resistance by a factor of nine, down to around 8mOhms.

I'd need more caps, but this just lowers ESR even further, I need to try some calculations first, to try to estimate what the 'new' inductance would be.

Any other suggestions?

EDIT: I think I have a plan. I currently have a capacitor in parallel with the notch filter, which provides a low impedence to the ripple. If I connect the notch filter between two chokes, like a 'T' filter, I guess, the second choke should present a high impedence to the ripple, and the notch filter will present a much lower impedence to the ripple, so the ripple will choose the path of least impedence. I know two chokes in series 'add', but they aren't strictly in series, they have a notch filter between them. I can't simulate this on the power supply simulator, so I'll either have to use LTSpice, or just build it and experiment. I'l get a schematic together later.

hen918

Wed Jul 18 2018, 09:35PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

It sounds like you are making an LC Butterworth or Chebyshev filter. a Chebyshev may be better for a PSU filter, as you want it to attenuate quickly.

Ash Small

Thu Jul 19 2018, 05:16AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

hen918 wrote ...

It sounds like you are making an LC Butterworth or Chebyshev filter. a Chebyshev may be better for a PSU filter, as you want it to attenuate quickly.

I guess it's a Butterworth filter with Cauer topology designed for a passband of zero Hz (actually it's a few Hz, but it approaches zero.

It differs from a multistage Butterworth because instead of having three capacitors and two chokes, the centre capacitor is replaced by a notch filter comprising a capacitor and choke in series. The notch filter needs a high Q in order to work, efficiency depends upon the high Q factor, I guess.

I can't actually find anything like it anywhere, maybe I've invented a new type of filter?

I think I can run some tests with chokes and caps that I already have, I don't think I'll have time to wind anymore anytime soon.

Something like this

EDIT: It shouldn't make any difference which way up the centre section goes, just a question of aesthetics in the schematic, I guess.

hen918

Thu Jul 19 2018, 02:48PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

Ahhh, I see. Simulation with some token values suggests that you need to keep the bandwidth of the notch quire high, as, whilst a rectified sine-wave is mostly twice the original sine frequency, there are plenty of harmonics at lower frequency.
I've attached the zipped simulation, the FFT of the transient analysis output is useful.
]draft9.zip[/file]

Ash Small

Thu Jul 19 2018, 08:33PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

hen918 wrote ...

Ahhh, I see. Simulation with some token values suggests that you need to keep the bandwidth of the notch quire high, as, whilst a rectified sine-wave is mostly twice the original sine frequency, there are plenty of harmonics at lower frequency.
I've attached the zipped simulation, the FFT of the transient analysis output is useful.

What's the best tool to use to open the ASC file, Hen?

first previous 4 5 6 7 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob