Is DCDC module case connected to input or output ground?

[treez]

Hi,
Is the metal case of this family of DCDC modules connect to a circuit trace in the module.
Also, is the metal case of aluminium?...as you know, aluminium offers very poor shielding of radiated noise.

https://www.xppower.com/portals/0/pdfs/SF_JTH.pdf

 

[wwfeldman]

according to the data sheet:
isolation is 1500 VDC input to output
1000 VDC for both input to case and output to case
so the case may be connected to some trace on the board, but it is not connected to either the input of the output

i'm guessing the case is aluminum, (maybe an alloy?) because it is convection cooled
aluminum is better than steel and palstic

 

[FvM]

as you know, aluminium offers very poor shielding of radiated noise

Why particularly? Never heard about.

 

[c_mitra]

.as you know, aluminium offers very poor shielding of radiated noise.

I did not know. But why? I expect good conductors to offer good shielding.

 

[treez]

so the case may be connected to some trace on the board, but it is not connected to either the input of the output

Thanks, i suspected this, but the datasheet does not certify it. This is very bad, would you agree? Because if the case is floating, then that can create a noise hazard to nearby circuitry. But since there's no way of knowing to what node (if any), the case is connected to, we are unable to connect it to input or output ground. Do you agree?

--- Updated Jul 28, 2020 ---

as you know, aluminium offers very poor shielding of radiated noise

Why particularly? Never heard about.

Thanks, yes, it is something to do with aluminium being non magnetic.

 

[FvM]

Ferromagnetic shields are effective for magnetic fields below 100 kHz. Electric and magnetic fields of higher frequency can be perfectly shielded by good conductors like aluminium or copper.

Dominant noise of DC/DC converters is conducted differential and common noise. I doubt that capacitive coupled electrical noise from a floating case will be a big problem. If it is though, it can be earthed by an conductive adhesive tape or a spring contact.

 

[treez]

I doubt that capacitive coupled electrical noise from a floating case will be a big problem. If it is though, it can be earthed by an conductive adhesive tape or a spring contact.

Thanks, yes we will do that, since in any electronics system, any "floating" metal whatsoever is going to worsen the electrical noise situation. The metal can act as like a Yagi antenna and re-radiate any ambient noise further into the system. Would you agree?

 

[Easy peasy]

get out a multi-meter ( DVM ) and maesure the capacitance to case on input and output - it is likely there are caps in each place - it is not standard practice to connect the case to other than earth - if it were connected to 0v input for e.g. it should have a big sticker on it to that effect.

Given no earth - yes the case will radiate - if it is sufficiently isolated you can connect the 0v o/p to the case & this may help reduce radiation.

 

[treez]

Thanks, we dont actually need the isolation, and since isolated SMPS's are more electrically noisy than non-isolated SMPS's (this has been kindly confirmed previously on edaboard) we will connect input and output ground. We will then connect this ground to the case.

The DCDC module of the top post also does not say whether or not its underside is metal cased. If it isn't, then i assume we can reduce its electrical noisiness by putting some copper tape underneath it?

 

[Easy peasy]

measure the C first ....

 

[treez]

Thnaks, yes we will, but even if there is capacitance there from case to input/output...we will still short input and output grounds, and connect ground to case, because a direct connection will be less electrically noisy than a capacitive connection. Would you agree?

I mean, with offline flybacks, (as an example to demonstrate the point) its usual to put a y capacitor between pri and sec to reduce noise…but actually connecting pri and sec grounds makes it less noisy (supposing lack of isolation can be tolerated) …because a short is a capacitor of infinite Farads,

 

[c_mitra]

antenna and re-radiate any ambient noise further into the system. Would you agree?

No. Consider total energy balance.

Even an isolated conducting plate in between a source and sink can act as a barrier to EM radiation. Of course the effective reduction will depend on the conductivity, size and shape and magnetic permeability.

If you ground it (same as connecting to a big capacitor), it becomes more effective because you have a larger RC time constant.

If you connect to earth, total energy balance will not work because some energy will be dissipated into the ground. If unconnected to earth, a smaller dissipation will result from eddy and hysteresis losses.

if you assume that you can take up ambient noise and focus that into your circuit so that your system noise temp will increase, you are mistaken.

--- Updated Jul 29, 2020 ---

but actually connecting pri and sec grounds makes it less noisy (supposing lack of isolation can be tolerated) …because a short is a capacitor of infinite Farads...

Consider a simple transformer. The output and input grounds are independent. That makes a perfect isolation.

Consider you are trying to regulate the output (voltage) with respect to the primary circuit. you must have a ground reference for the output sent to the primary side. Else the secondary voltage has a independent reference ground that is unknown to the primary side.

Say you are missing the "Y" capacitor. In other words, you do not have the secondary ground reference.

In that case the primary side cannot figure out the secondary voltage and assumes it all noise. The "Y" capacitor must have reasonably low impedance at the switching frequency. Even if you provide a DC feedback from the secondary side, the secondary voltage (at the transformer) is now a defined ground.

--- Updated Jul 29, 2020 ---

One more point:

1. You have a transformer with two independent secondaries. You can connect them in series, right? That is because these two outputs are independently referenced. They are exactly like two batteries that can be connected in series.

2. Can you connect the two outputs of a dual output DCDC converter? No, because they have the same voltage reference.

3. Is the output of the DCDC converter isolated from the primary side? Yes in a DC sense. No isolation at the operating frequency.

 

[Easy peasy]

you can connect Vin gnd to case to Vout gnd - but if you short out any internal shunts in doing so - there may be some problems ....