Choke-Input: How do I calculate the right snubber for high-current-low-voltage

[Blitz]

Originally posted by Blitz


Yes, full-wave with two secondaries and two bridges, balanced powersupply with +23V and -23 V, therefore two chokes and a capacitance of 100000uF behind the choke.

I have started to debug the power supply for high-frquency ringin with the osci, I first wanted to get rid of the buzzing (sound like high-frequency though) of the chokes.

This is per channel.

Correction: I have not started to debug for high-frequency ringing, how do you see it on the osci by the way ?

[JonMarsh]

Hmmmm, 2. 2 ohms with a DC current of 6A is a little higher resistance than I'd prefer. Power dissipation is I^2 X R, which is around 72 watts average in your case- this is why you have a 60 degree rise over ambient. But magnet wire is rated to ~200-250 C, in most cases, so you should be fine, apart from the power loss.

Another thought to ponder, and somthing you can measure- the copper winding has a temperature coefficient, and as the inductor heats up, the resistance increases. There may be some combined affects when you amp has been on and running for some time, that is reducing slightly the overall current through the choke, maybe less than 10%. But this could be taking you a little further away from the saturation point of the core, and quieting it down. Is your core rated at 6A max? The thing is, one usually wouldn't use the core at the full rated DC, because there is an AC ripple current which rides on top of the DC value (which is a function of the lines frequency and inductance); the actual working current will be then higher than the 6A rating. Is this happening in your case? You may be better off with an 8A rated inductor, for example, carrying 6A average DC current. The only drawback, of course, is that it's more expensive. But, you were saying something about being a perfectionist?

As long as you're doing a true choke input supply, ie. rectifiers to the choke, and it smoothing the current draw into the first primary filter bank by operating in continuous conduction mode, (which greatly lowers the peak rectifier on the AC mains and almost doubles the transformer utilization factor), you can have some potential for mechanical hum in the transformer, because the current isn't flat, at best, but has a sinisoidal ripple current which is a function of the AC frequency and the inductance.

What it sounds like you're looking at on the input side to the inductor to prevent interruption of the current path when the diodes are all in blocking is like the bypass capacitor used in a primary side active PFC circuit; this is usually one to two microfarads film capacitor connected across the output of the diodes. In this case the "switching" frequency is much lower, and the current higher- you might want to take that up to 10 uF or so. This should be a low ESR, high quality film cap. If the supply is singled ended, then wiring it to the diode outputs and to ground should do the trick.

If you could post a schematic of your basic supply configuration, it would be easier to comment.

[mwh-eng]

Glad Jon thought about the I^2*R loss. 79.2 watts dissipation is too high. Make sure you really have 2.2 ohms winding resistance and your amp is pulling 6 amps DC current. For both + and - supply this is 158.4 watts dissipation. Ouch! I would think the chokes would smoke. I'm sure they will cause a blister if you get too close. I think you need chokes with much less winding resistance for your application.

If your scope is not isolated and is a three wire, with earth ground connection to AC power, then connecting your scope ground probe to the secondary ground will tie your secondary(s) to earth ground. Should not be a problem. Look at the signal on each side of each rectifier. Better to use differential probe setup to measure directly across rectifiers. Don't connect the scope ground probe to any rectifier lead unless your scope is fully isolated. You should apply normal range of loading for measurements. Be careful with these measurements. Wait for someone else here, to post to either confirm or not confirm my measurement suggestions, unless you have such experience.

Would be good if you could post a schematic as Jon suggested. Do you have Spice, LTspice, etc.? Good way to quickly check for continuous current in the inductors. Only need to model one supply. You can calculate if you have continuous current. It's been 15 years for me, so I'm still scratching my head and rather use Pspice.

If you heard a high frequency noise from the choke, then you did have high frequency ringing. Need to find the source of this ringing. Your transformer leakage inductance and the rectifier junction capacitance is suspect.

Good luck and keep us posted.

[Steven]

Originally posted by mwh-eng
You can calculate if you have continuous current. It's been 15 years for me, so I'm still scratching my head and rather use Pspice.

A nice overview and calculation directions for the critical inductance for continuous current is given on: http://www.qsl.net/i0jx/supply.html

Steven

[Blitz]

Steven, very cool link, thanks a lot. When I calculated right, I can get away with a ten times smaller choke in inductance using a RESONANT CHOKE POWER SUPPLY. Wow !

I will make now some experiments. 72 Watts is a lot...fortunately the chokes are center-tapped, so I can use as well only half of the winding, bringing DC-resistance and inductance down. Let's see how this is. And I will test definitely the RESONANT CHOKE POWER SUPPLY...I could do this even with some coils I have from building some cross-overs ! Very cool.

[Steven]

Originally posted by Blitz
...fortunately the chokes are center-tapped, so I can use as well only half of the winding, bringing DC-resistance and inductance down.

Maybe you can separate the two windings that use the common center tap. Then use one for the positive supply and one for the negative supply. Make sure that the current direction is the same in both, so the magnetic fields should NOT cancel. In this way you double the inductance of each coil, just because there is current in the other one and they are magnetically coupled.

Actually, now I think about it, maybe you don't have to separate the windings. Use one in the negative rail of the positive supply and one in the positive rail of the negative supply. The center tap is the new ground where the supply capacitors are connected to.

Steven

[Steven]

I just forgot:

• Use a resistor in parallel with each choke as damper, guideline 10..100 Ohm. This allows for smaller snubber capacitors across the bridges.
• Connect a big 4.7u foil capacitor across each bridge, from plus to minus (more effective than across AC terminals). Normally such a capacitor needs a series R to dampen the LCR with the secondary of the transformer, but the optimum value of the resistor in series with the C is approx sqrt(L/C) with L is the secondary inductance and C the value of the C. The 4.7uF seems big enough to allow for a value of R that is close to the series resistance of the secondary, so effectively the LCR is dampened already.

Steven

[Blitz]

Steven, thanks again, these are very helpful hints. So, I experimented a bit:

- When I use two chokes with 40mH and 2,2 ohms with the two bridges, one in the + and one on the minus rail, I have a DC after the capacitor bank of 100000uF of approx. 22,5V. My ripple is down to 25mV. The choke is buzzing a little bit with 6A (being specified for 6A).

- When I use both choke in the same configuration, but taking the voltage out already at the center-tap instead of the whole winding, the voltage increases by 1,5V and ripple is now 33mV. The chokes are buzzing louder than before.

- When use the last configuration you showed in the picture, Steven: I have now 23,4V, so one volt more than originally, a ripple of 55mV and the choke is absolutely dead-quiet. It may be the case that the transformer is now buzzing a bit, but that may be an illusion as well.

Do you have any explanation ofr these observations ? I would prefe to go with the last option, but is this than longer a choke input-PSU ? How nuch bigger are the current-spikes asked from the transformer now in comparison to the original setting ? Is there any RF-filtering as there are now no series-elements in the + and - rails ?

[Blitz]

OK, just two additional observations:

- after an hours of operations, the choke is now buzzing as well, ripple has come down from 55mV to 48mV

- Even though I have a soft-start-circuit in place (0,5sec over a 27 resistor, than shorted by a relay), each time I switch it on directly without a variac, the automatic fuse of my house reacts and I am sitting in the dark. How do this come ?

Nevertheless I feel that this setting has its advanta´ges as I need only half of the number of chokes, which is max. half of the heat and I save two chokes for another projects...if there are no downsides like strong current-peaks etc ?

[Blitz]

OK, the turn-on problem is solved: I increased the time where the current goes through the reistor and that works.

Would be nice to get answers to the other questions above though