Power Supply Case Grounding

[jesseh1]

(My first post, yay!)

I am currently working on converting an old (~10 years old) Dell psu into a lab unit that I can use to test my other electronics projects.
(see http://web2.murraystate.edu/andy.batts/ps/powersupply.htm)

I am mounting the unit in a cardboard shoebox so that I can have more room to work with. I have cut slots in the shoebox for the fan, wall plug, and side vent. I have mounted a toggle switch in the front of the box and connected the gray wire (ps_on) to one lead and black ground wire to the other lead (I know that Dell didn't stick to the ATX standard back in the day and thus I used pinouts.ru to determine which wire is which).

Now for my question:

When testing for continuity (to make sure that I didn't short anything), I discovered that all the wires (and I mean ALL the wires) running from the psu appear to connect at some point (I suspect the unit's metal casing). Suspicious, I checked another ~10-year-old Dell psu and found the same results. What I found really odd was that the ps_on wire, of all wires, was grounded. Startled, I compared this to a newer BesTec psu (not sure of the age, but it had the 20-pin mb connector and the 12V 4-pin connector) and discovered that most if not all of the wires, EXCEPT the green ps_on, were connected to the case in someway.

Can someone please explain what is going on here? Why are all of the Dell connections shorted together, but not all of the BesTec ones? Why are most the connections meeting at the metal casing? Shouldn't that cause a short circuit?

Also, with regards to my project, what safety precautions should I take to reduce the risk of burning the shoebox?

Thanks,
Jesse

 

[jsc]

Was your ohmmeter reporting ) ohms or a low resistance that was not quite zero ohms. There is a difference. If the green wire were really grounded in an itherwise perfectly functioning PSU, the PSU would not turn on.

 

[jesseh1]

Was your ohmmeter reporting ) ohms or a low resistance that was not quite zero ohms. There is a difference. If the green wire were really grounded in an itherwise perfectly functioning PSU, the PSU would not turn on.

I just tested the psu that I'm using with another multimeter (for which I have the documentation, lol). This time, the audible signal did not sound, but the multimeter displayed a resistance of several hundred ohms, for the connections I tested. According to the its documentation regarding the continuity test, the audible signal will sound only if the resistance is less than 30 ohms.

I don't know what resistance triggers the other multimeter, but it must be something a lot higher than 30 ohms.

I take it that the black ground wires are supposed to connect to the psu's metal casing? (Testing that triggers the audio signal).

Why is the multimeter still reporting a resistance (albeit several hundred ohms, as stated above) between all the connections and the psu's metal casing? What purpose does having the wires connect to the casing serve?

 

[hunter315]

Flip the leads on the multimeter around and see what it reads, you might actually be reading a resistance from the reverse diode drop across the main output MOSFETs and not actual resistance. If it measures a dozen kilo ohms or more in the other direction its fine. All of the wires should not connect to the metal casing as that is directly connected to earth ground, only the black ground wires should, also a resistance of a few hundred ohms tells me that its not tied to the case, there is something in the middle otherwise it should be under about 10 ohms.

 

[pjmelect]

These power supplies have a low resistance connected across the outputs (to ground) to provide a minimum load. Often this is not enough and it is a good idea to provide more of a load to the power supply, particularly the 5V supply, as if this rail is not sufficiently loaded the 12V rail will be low.

 

[jesseh1]

Sorry about my delay in response, as the last week has been rather busy. Fortunately, during that time, I was able to make a lot of progress on the lab power supply.

I re-checked the resistance as per your recommendations and realized that I didn't short anything (yay!), but the power supply was designed that way.

Hunter315: I'm not sure what you meant by reverse diode, but when I reversed the multimeter's leads, I did not notice a change in the resistance.

pjmlect: I knew about the minimum load requirement, so I purchased a 10 ohm, 10W resistor (as per many recommendations I found on the internet) and soldered that between a 5V wire and ground. Would a higher ohm resistor work? (I've heard that too much resistance will not draw enough current, and thus will not keep the power supply on).

You guys have been extremely helpful.


As of now, the following is soldered into place and works:

- The on/off toggle switch
- Red 'Standby' LED
- Green 'Power_good' LED
- 'Ground' binding post / banana jack
- 5V banana jack and circuit breaker (guard against short-circuit)

I will try to post pictures when it's done. I still need to make another run to the local hobby electronics shop (NOT RadioShack).

Do any of you guys know what 'protection' features (overcurrent, short-circuit) the psu I'm using (~10-year-old Dell 200W) would definitely have? I checked the model number online and have not found any relevant results.

Also, will selecting the best answer close the thread or can members still reply to it?

 

[pjmelect]

pjmlect: I knew about the minimum load requirement, so I purchased a 10 ohm, 10W resistor (as per many recommendations I found on the internet) and soldered that between a 5V wire and ground. Would a higher ohm resistor work? (I've heard that too much resistance will not draw enough current, and thus will not keep the power supply on).

The resistance built in to the power supply is just enough to stop it from blowing its self up, but not enough to keep the 12V rail from dropping in value. If the 12V rail is high enough then the resistance across the 5V is OK.

Do any of you guys know what 'protection' features (overcurrent, short-circuit) the psu I'm using (~10-year-old Dell 200W) would definitely have? I checked the model number online and have not found any relevant results.

It should have short circuit protection and over voltage protection (apart from the 5V standby supply).

Also, will selecting the best answer close the thread or can members still reply to it?

Giving a best answer does not close the thread and members can still reply to it, however the motivation for some members to reply is reduced. A best answer is always appreciated.

 

[delluser1]

Giving a best answer does not close the thread and members can still reply to it, however the motivation for some members to reply is reduced. A best answer is always appreciated.

Giving a best answer provides mods with a reason to lock a thread , then there are no more responses.

Link in first post doesn't work for me. I'm anxious to see pics.
Dell had gone to a standard ATX pinout around 10.5 - 11 years ago.
Pics of the innards would be nice too
Maybe just the model #

 

[jesseh1]

Best answer selected by jesseh1.

 

[jesseh1]

Giving a best answer provides mods with a reason to lock a thread , then there are no more responses.

Link in first post doesn't work for me. I'm anxious to see pics.
Dell had gone to a standard ATX pinout around 10.5 - 11 years ago.
Pics of the innards would be nice too
Maybe just the model #

I just fixed the link. Try it now.
Model number of the psu I'm using: NPS-200PB-73M.

As I previously said, once it's done, I will upload pictures (can they be stored on tomshardware.com or do they have to be stored elsewhere?).

Because I have soooo many extra 5V and ground wires, I'm tempted to find some use for them (like adding USB connections to charge stuff), especially considering that the 5V rail is rated for up to 22A. I know new iPods won't charge without voltage dividers on the USB data wires (gosh darn you Apple!).

Any other ideas for the superfluous 5V wires?