I am building a similar power supply to another thread that was just posted, only I am using a pair of DPS5005 units as the outputs:
These are 50V 5A max output units. Like all standard buck or boost converters (so I read) they are not isolated. I once read in a SMPS handbook that for all non-isolated topologies, one of the two input rails is connected to one of the two output rails.
Because these units require about 1.5 volts higher input voltage than output voltage, I decided to go for a transformer that would deliver about 55V to try and cover the margin. In fact, the transformer I purchased turns out to put out about 58.5 volts with no load.
I didn't know at the time that it came from a
Harmon Kardon 870 amplifier, but I bought a transformer from ebay that promised to have +/- 55V output windings. I was originally hoping to use this to power
4 DPS5005 supplies, but I realized this would leave me unable to isolate them from each other systematically (I
think it would be possible to get two pairs of isolated supplies, but then knowing me I would accidentally short something out in a DUT).
In fact, the full configuration of the transformer looks like:
In other words, it converts 120VAC @ 60 Hz input to two center-tapped +55V to GND to -55V supplies. The seller was nice enough to include the 10,000 uF 63V capacitors (along with the 4700 uF 63V capacitors, too -- and the bridge rectifiers, though I'm using modern ones).
This transformer is
huge. It's
10cm tall and
12.5cm in diameter. The entire inside of the unit is potted.
The configuration I setup is essentially the same, except I did not put in the
22 ohm 1 watt resistors, and I am using more modern bridge rectifiers (rated for 6 amps), which are heatsunk.
In essence, what I have is an
unregulated 52V-58V power supply driving two isolated buck converters, and I have two more unused -55V taps. I originally thought (and am still considering) linear regulation
before the buck converters, but I don't yet have a solid grasp on how much ripple the buck converters themselves are adding. This power supply is supposed to eventually be fully programmable, so it's entirely possible that I could design a linear stage to go
after the buck converter, which is somehow tuned/trimmed by the buck converter's inputs. I am pretty sure that most linear regulators aren't very good at taking out the ripple that SMPS tend to inject (I think the switching frequency is in the 50kHz range, but I'm not 100% confident of this yet).
My first lines of questioning are about filtering, because I've noticed that when both buck converters are powered on (and regardless of them being loaded or not), the 2nd buck converter injects quite a bit of noise onto the first one. I am guessing that this is noise which is injected upward through the transformer, back down to the other converter. I am also pretty sure that I want to do my best to ensure that none of this noise goes back up through onto my AC power because I'm not a monster!
So:
1. Where exactly do I need to put AC mains filtering? In other words, should I
just put a single mains filter on the primary, or should I put a mains filter on
each secondary, or
both?
2. Is whatever mains filtering solution you recommend from (1.) sufficient, or do I also need to construct a DC filtering stage (or stages) using resistors and these giant capacitors? Or should I build some other sort of LC or RLC filter?
3. Is it even worth it to use a linear regulation stage before the buck converters, given your answers for (1.) and (2.)? While this device claims that 55V is the maximum input voltage, it clearly has no problem accepting 58 or 59V. It is possible that this is making the noise worse, but I can't yet be sure. In any case, won't any linear regulation stage I put be somewhat ineffective at reducing the SMPS noise coming back from the other converter?
4. What are my options realistically at filtering the output of the converters? Will a passive filter setup suffice, and what will be the drawback (worse transient performance? worse on/off performance? worse overshoot?)?
5. (added in edit) If I power one of the buck power supplies from a 12V or 24V worth of batteries, is it reasonable to take a ripple measurement in this case as representative of the ripple added by the buck power supply? If so, is it reasonable to assume this is about the same ripple added at 50V?
In case the answer to any questions involve simulation and measurement, I have some parameters about various things, like the inductance of the various coils, and the capacitive coupling of various coils to other coils. I have a DER DE-5000 which I suspect should be capable of measuring such things. I also have a function generator and an oscilloscope, so I can inject signals and FFT them to try and ascertain performance of the transformer. I doubt I can possibly load/saturate the transformer easily.
Thanks!
Home
Help
Search
Login
Register
