DPS5020 dual power supply build

[Dannyx]

Re: DPS5020 dual power supply build

Why is the on-board FET a cause for concern here exactly ? The way I understand it is when Q4 in your schematic (which I'll kindly ask you to remind me which software you used to create since I keep asking and keep forgetting ) "opens" (gate goes low) it causes a spike on the input side of the module which can exceed the rated 50v input ?

I see TWO sets of resistor+capacitor: one on the input, one on the output. These act as "quenching" devices so to say ?
Also I believe there's an error with the symbols for the X caps across the switch, as they're shown as electrolytics

Like I said in my original post, I already built a similar unit and I now realise there's a LOT of room for improvement and I'm probably running it improperly there: no arc-suppression on the AC switch, no zener diode across the input terminals of the DPS, no caps on the input and output the way you suggested....I'll probably open it up again and improve it as well, especially since this one is a single transformer setup and don't have to worry about keeping anything separate.

[redwire]

Re: DPS5020 dual power supply build

My worry is not damaging the DPS module due to the output on/off mosfets. What is weird about the modules is the low-side ground switch they have.

Stray capacitance on the power transformer means things can float with noise and hum, so I suggest adding a 10nF and 1MEG to give leakage currents a path to earth ground. If things floated say to 50V it would damage the DPS mosfet when the PSU output was switched off. Output switched off -> (-) floats

The DPS5020 has a couple fuses on the PCB. I don't think you need an input fuse - it has F3 on the pcb. F1 I think is for the MCU and display.

This is the circuit I might try.

[megaraider]

Re: DPS5020 dual power supply build

(According to this source)
Not disappointing when pushed to near absolute max output current limit...
Noise stays within acceptable boundaries, and very important no voltage spike.

[megaraider]

Re: DPS5020 dual power supply build

the interface is a serial>usb adapter with a WinChipHead chipset (a damned good chip)
now here is the interesting bit, in the picture it looks like it's already opto-isolated!!!!!

Confirmed:
CH340 « serial to USB conversion (standard application configuration)
EL357N « phototransistor optocouplers (2 pcs: Rx & Tx)

[Dannyx]

Re: DPS5020 dual power supply build

I swiped THIS from "The Key Master" today when he wasn't looking It's the 230v input filter I was talking about.

[stj]

Re: DPS5020 dual power supply build

your boss was almost right.
it would be more accurate to say the tolerances are such that you never know which contact will move first.

the mosfets may be for rectification/reverse protection.

[redwire]

Re: DPS5020 dual power supply build

I would have a cap across each switch contact. But I am not knowledgeable on the European usual method. I think 10nF is minimum and I use 33-47nF for larger transformers to 1500VA.
My boss yelled at me - "the {DPST} switch contacts are never identical. One opens first and another closes first". So the power off arc could occur at either contact.

From the DPS5020 manufacturer:
"if you use AC-DC converter or transformer coil, please add rectifier (rectifier bridge) and filter (more than 4700uf filter capacitor). If input power supply is unstable, please add LC, L is over 47uH, C is over 2200 uf."

So I think 10,000uF is perfect. The LC filter is probably if you are supplying noisy power from a SMPS. It's poorly translated.
You can see two power mosfets at the IN(-) on the PCB so not sure why.

[Dannyx]

Re: DPS5020 dual power supply build

Yes, I believe on just about every device I've seen so far around here that uses a mechanical switch for the mains it's going to be a DPDT to "cut" both wires going to the plug. Makes perfect sense, since you want to isolate BOTH wires. The plugs are not polarized so L and N can be flipped around at the luck of the draw - you have a 50% chance of inserting the L into the N when plugging the cord in

Would THIS be a good way to connect the X-cap ? Or should it go across the switch and not the load ? In that case, I imagine I'd need two: one for each set of contacts. I also have no idea what value it should have - I'll probably just trash-pick one from a drill or some SMPS

[redwire]

Re: DPS5020 dual power supply build

Do you 220V guys always use DPDT power switches?
North America is 120VAC and SPST switches. Safety codes do not generally allow disconnecting neutral connection in a product.

[Dannyx]

Re: DPS5020 dual power supply build

Oh right, forgot about the power switches. I should place an X cap across them like I've seen in other devices which use a physical switch to switch transformers (see audio amps).

So I should keep the resistor/cap idea after all.

[stj]

Re: DPS5020 dual power supply build

no idea about the mosfet.

[redwire]

Re: DPS5020 dual power supply build

If you leave the secondary-side floating, there will be stray voltage - it can be as bad as a SMPS brick Y-cap floating to 60-90VAC but even a 1MEG can lower it so nothing gets zapped. The Y-cap in this case is the transformer's pri-sec capacitance. You can measure it if you don't believe me, it can be a 100nF on toroids.
There is no cap across the on/off switch so a large arc happens and switch-off spike can hit the secondary side, it can kick up 100's of volts.
So I use a 1MEG/10nF cap on the secondary to earth ground just so it can't float or spike up. It's still considered floating for stacking power supplies.

stj, do these have the low-side N-ch mosfet? It was why RdTech said some DPS modules can/cannot charge batteries.

The Bluetooth add on module might be better, to go wireless? But I think the firmware is different or something makes it a hassle to add.

[stj]

Re: DPS5020 dual power supply build

look at the end with the 4pin connector - below are 2 4pin packages.

i know the chipset because of the driver they are using.

[Dannyx]

Re: DPS5020 dual power supply build

the interface is a serial>usb adapter with a WinChipHead chipset (a damned good chip)
now here is the interesting bit, in the picture it looks like it's already opto-isolated!!!!!

Yes, there is a serial>USB adapter included in the box. It's got a CH340 chip on it which I'm familiar with since it's present on my SPI programmer, though it's a CH341 there - same thing.
How did you determine what it is and that it's isolated ?

[stj]

Re: DPS5020 dual power supply build

you could - and hit it by mistake!!

btw, i was looking at the pdf pictures.
the unit uses a serial port.
the interface is a serial>usb adapter with a WinChipHead chipset (a damned good chip)
now here is the interesting bit, in the picture it looks like it's already opto-isolated!!!!!

[Dannyx]

Re: DPS5020 dual power supply build

I could use a switch which would toggle between tying the GNDs together in one orientation and tying the + of one to the - of the other in the other position, how's that ? That's more of a convenience, since a piece of wire would work just as well

[stj]

Re: DPS5020 dual power supply build

earthing, earth the rack case and the transformer bodies - although that will happen through the bolts anyway.

split-rail,
you connect the - of one channel to the + of the other channel.

so instead of [0v +15v] [ 0v +15v]
you have [-15v 0v-0v +15v]

if you did that and there was common ground you would short out the first channel.

[Dannyx]

Re: DPS5020 dual power supply build

dont link anything after the transformers to ground.
it will cause problems if you want to tie the 2 channels together as a split-rail output.

So you're saying the Y capacitor idea wouldn't work in this case ? I should just leave the case itself floating, only connected to earth and nothing else then ?
Split-rail refers to +15v and -15v for instance ? If so, wouldn't the GNDs of the two supplies be the 0v point and the output of each module represent each of the +/- output ? I forgot how this worked

why not put a dirt-cheap hub in too - so you only need one external connector.

I considered that as well, given that there's no speed concern here. Still, it would have to be used in conjunction with the isolators you suggested.

btw, how does the usb know which module your addressing? do they have configurable i.d.'s??

Yes, if you check out the manual I posted, you'll see further down that they are addressable. The software itself has a drop-down list in which you select which one you wish to "talk" to. Here's the link to the whole archive if someone's interested and doesn't feel like hunting it down.

[stj]

Re: DPS5020 dual power supply build

dont link anything after the transformers to ground.
it will cause problems if you want to tie the 2 channels together as a split-rail output.

you can get usb isolators from china for beer money.
why not put a dirt-cheap hub in too - so you only need one external connector.

btw, how does the usb know which module your addressing? do they have configurable i.d.'s??

[Dannyx]

Re: DPS5020 dual power supply build

Thanks for the great tips. That's the sort of pointers I like to hear

With the DPS modules I add (Y-caps 2,200pF or more) from (-) and (+) to chassis GND, to limit HF noise.

Should those caps be on the input or output of the DPS module ? I assume the output. So one cap on the negative leg and the cap on positive, then the remaining legs of the caps tied to chassis GND. Trouble is, although I'll use an IEC 13 connector to power the whole box, which has an earth terminal which I shall tie to the metal case, most buildings aren't actually earthed here
Keeping the two modules separate kinda goes out the window, since they now share a common GND.

A clamp-on ferrite could work.

Something like this ? I actually have these after ordering some a while back. Again, should these go on the output or input ? Or perhaps both ?

Add a resistor from the bridge rectifier (-) to earth ground to limit the stray voltage due to the power transformer's leakage.

What value should this have ? Just like with the Y caps, this will just sit there probably doing nothing, since there's technically no earth Also, it removes any chance of the two supplies floating...which may not actually be a bad thing in the long run now that I think of it.

Serial comms will be an issue if both DPS modules are floating.

Yes, I just realized that myself. As soon as you try to connect both USBs to a PC, the ground becomes common. One workaround would be to use the BT functionality. The trade-off of course is that you can't have both connected at the same time, unless you use two PCs, each running an instance of the software. I doubt it can run multi-instance on the same machine to allow controlling two or more supplies in real time. Haven't tried that yet. It would be interesting to find out.