In Reply to: Re: Regulation over long time periods for high DCR high L supply posted by mikeyb on November 1, 2006 at 13:39:43:
I'm really very uneasy with this forum and prefer not to stick around. In the spirit of sharing, I will give you the answer, and then I'm gone.From studying the problem, I see there is an optimal value for the input capacitor that maximizes the DC supply output voltage. If the capacitor is too small, the output drops because the cap can't sustain its voltage under the current load. If it's too large, the voltage drops because some of the higher-order raw supply harmonics get filtered out.
In this respect, ychumgmd's observations may have merit, although they were incomprehensible to me on first reading, and he has failed to follow through to the conclusion I'm about to offer.
In my opinion, the most interesting characteristic of this design is how the peaks of the input cap ripple voltage waveform vary with load current. Without the input choke, the positive peak voltage on the input cap is essentially equal to the transformer secondary voltage. As the DC load current increases, the ripple voltage increases as well. But because the positive peak is pinned, the ripple waveform can only grow downward, causing the average to drop and resulting in bad regulation.
With the input choke in place, the voltage on the input cap "floats" or is decoupled from the secondary voltage. You will see that the positive peak voltage on the input cap actually goes up with increasing load current at about the same rate that the negative peak voltage goes down. The average remains relatively constant, and this is why the regulation is good. This is only true for a limited range of input capacitor values.
Depending on the circuit parameters, current in the second choke may actually reverse, so that the input cap charges from both the upstream and downstream branches. In this case, the magnitude of the reverse current in the second choke increase as the load current increases. This is a non-intuitive result.
Someone posted today the observation about the optimum capacitor value for DC regulation. I don't recall seeing this mentioned here before. To the best of my knowledge, though, nobody to this point has ever tried to explain why this should be so.
It's worth noting for Tre's benefit that the small input choke significantly improves the peak amplitude and duty cycle of the diode charging current. This should mitigate his "hash" concerns.
The simulation doesn't give us the fine details about the diode switching noise. Careful evaluation and prudent application of snubbing components would be required for best performance.
If you're not comfortable with chaotic ringing in your ripple filters, you can increase the capacitor values and experiment with staggered tuning frequencies to get a better behaved response.
I can't comment on the sonic advantages or disadvantages of this approach. It does seem to have some technical benefits. I am especially hesitant to say anything that might appear to vindicate Jeff and Dennis. I hold these two in very low regard.
I think this approach should work well even with transformers of ordinary DC resistance. The design rule is to start with a small input choke, then vary the input capacitor to maximize DC output voltage. Next, check the input capacitor ripple voltage peaks as a function of load current to insure the condition for good regulation. Finally, add one or more LC sections to achieve the desired final ripple voltage and step response.
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- Re: Regulation over long time periods for high DCR high L supply - Kcanretsap Yrneh 14:42:06 11/01/06 (20)
- Or, just regulate the daylights out of it :) - BBeck 06:42:18 11/02/06 (2)
- Re: Or, just regulate the daylights out of it :) - Urael Cnfgreanpx 19:16:58 11/02/06 (1)
- Re: Or, just regulate the daylights out of it :) - BBeck 20:09:39 11/02/06 (0)
- What did you use to do this analysis? - Russ57 06:06:20 11/02/06 (12)
- The "Flywheel" power supply - Rehy Pants Nacker 09:09:42 11/02/06 (11)
- Wow! - Nickel Core 13:06:47 11/02/06 (0)
- Fascinating ... - mikeyb 10:27:47 11/02/06 (9)
- Further thoughts - Ratspar Cheenk 13:21:34 11/02/06 (8)
- Re: Further thoughts - Steve O 17:22:05 11/02/06 (1)
- Re: Further thoughts - Urael Cnfgreanpx 19:13:05 11/02/06 (0)
- Turbine - Mark Kelly 14:54:33 11/02/06 (1)
- Re: Turbine - Urael Cnfgreanpx 19:06:32 11/02/06 (0)
- Re: Further thoughts - Nickel Core 13:57:27 11/02/06 (3)
- Re: Further thoughts - Cranky Thrensap 14:05:05 11/02/06 (2)
- No sorry... - Nickel Core 14:47:13 11/02/06 (0)
- Fair enough. Thanks! (nt) - Nickel Core 14:10:32 11/02/06 (0)
- Thank you Henry - Mark Kelly 19:15:28 11/01/06 (0)
- Re: Regulation over long time periods for high DCR high L supply - C.Y 16:58:00 11/01/06 (0)
- Well said ... - Naz 16:13:14 11/01/06 (0)
- Re: Regulation over long time periods for high DCR high L supply - mikeyb 16:10:32 11/01/06 (0)