3

I have a power supply board that supplies a DSP, FPGA, CPLD and other components of a system. When I measure the resistance between power rails and GND using a multimeter I get the following readings:

  1. Between 12V and GND: Infinite
  2. Between 5V and GND: Infinite
  3. Between 3.3V and GND: 250 Ohms
  4. Between 1.2V and GND: 40 Ohms

The 3.3V and 1.2V rails have many decoupling caps around power supply pins of the DSP and FPGA, are they the only cause for this much reduced isolation between these voltages and ground?

Is it safe for the long run to work with these specs? The system is now running perfectly.

Also how can I avoid this issue in future designs?

Edit #1

In a project I'm currently working on, one of the tests that will be done on the product will be isolation tests. The customer will measure the resistance between supply rails and Chassis ground and want it to be greater than 10 Mega Ohms, that's why I want to avoid these low resistance values. Is there a device used to measure isolation? or it's done with a normal ohmmeter?

Edit #2

These measurements are done when the system is powered OFF.

Abdella
  • 2,605
  • 4
  • 30
  • 38
  • 1
    What voltage is the DVM using to measure those resistances? It's probably too high for the 3.3V and 1.2V inputs. – Leon Heller Aug 24 '12 at 21:34
  • I don't know the voltage used to measure resistances. Should it be mentioned in the DMM datasheet? – Abdella Aug 24 '12 at 21:54
  • Since you mentioned decoupling capacitors... Are you sure you waited long enough for the voltages to settle? I too have a circuit which reads 40 ohm resistance when connected to a multimeter, but after several minutes, it starts increasing and settles at infinite. – AndrejaKo Aug 25 '12 at 15:57
  • Yes I thought of this, but the reading stays at around 40 ohms no matter how long I wait. Is there something wrong with my measurement method?! – Abdella Aug 25 '12 at 20:55

3 Answers3

3

When I've heard about power supply "isolation testing" before, the context has been a test done on the mains inputs to a system or device. I've also heard these tests called "insulation resistance" tests or "hipot" tests, though those could be slightly different tests. The test confirms that voltage applied to the mains input won't connect to the case of the UUT and cause a shock hazard.

The test involves applying a fairly high voltage (1 to 5 kV or so) to verify there's no spark gaps or insulation breakdown paths between the mains and user-accessible surfaces.

According to this, isolation testing tests

The maximimum voltage (ac or dc) that can be continuously applied from the input to output or input to case of an isolated power supply. ... Testing power supply isolation requires specialized equipment (Hipot tester) and can be destructive.

The main point is that this is a test you would do on the mains input of an isolated power supply. It is not a test you should be doing on the inputs where your operating circuit expects to receive its local regulated input power.

The Photon
  • 126,425
  • 3
  • 159
  • 304
  • Mmmm, your answer made some points clearer. I reread the isolation test in my customer specs and found that he will measure between voltage rails and the "Chassis" ground not the circuit GND which makes more sense with your answer. Does that mean that I should completely isolate the Chassis GND from other grounds in the circuit? This will answer my other question here too. http://electronics.stackexchange.com/questions/37662/should-shielding-gnd-be-isolated-from-circuit-gnd Thanks – Abdella Aug 25 '12 at 21:03
  • Also does that mean the I should use isolated DC-DC converters? – Abdella Aug 25 '12 at 21:05
  • This test is more relevant to your ac-dc converter than to any dc-dc converter in your system. The test is done on the mains input to your system. – The Photon Sep 02 '12 at 16:19
1

A multimeter measures the resistance by applying a test voltage and measuring how much current flows. When you measure something that isn't a resistor, the measurement that you get doesn't make much sense. Instead, you should use an ammeter to measure the current consumed on each of the rails while your circuit is functioning normally, and determine if that current is within the range you'd expect.

Jim Paris
  • 2,685
  • 1
  • 12
  • 12
  • In a project I'm currently working on, one of the tests that will be done on the product will be isolation tests. The customer will measure the resistance between supply rails and ground and want it to be greater than 10 Mega Ohms, that's why I want to avoid these low resistance values. Is there a device used to measure isolation? or it's done with a normal ohmmeter? – Abdella Aug 24 '12 at 22:23
  • The multimeters I'm familiar with apply a test current and measure the voltage. – markrages Aug 25 '12 at 04:11
  • @Abdella, please edit that information into your question. I think you are misunderstanding the requirement. Or the requirement is insane. – markrages Aug 25 '12 at 04:12
  • Btw, these measurements are done when the system is powered OFF. – Abdella Aug 25 '12 at 08:20
1

There is no problem with having an open circuit output resistance of 40ohms. Possibly, it's a fixed divider (internally). Without having knowledge of the circuit exactly, it's hard to say exactly how much power is going to be absorbed, but just use I=1.2/40=30mA as a start point -- nothing to be worried about. If you want a better gauge of current being drawn, measure the current through an external load with power on... from there, you can derive current being absorbed by the internal resistance.

Lower voltages typically have more decoupling capacitance, because the signal to noise ratio is lower by virtue of a lower supply rail.

pat
  • 2,075
  • 12
  • 11