Kikusui PAB 18-5.5 - Grounding

[rstor22]

I purchased a used Kikusui PAB 18-5.5 a while ago. I recently noticed the voltage would drift sometimes by several hundred millivolts with no load when adjusting the pot and then eventually kind of settle down. After replacing the course potentiometer with a 10 turn pot it is much better (the fine pot has been removed from circuit). Though the load regulation is not within specifications -- 5.5A load drops about 25mV in about an hour (2A drops about 10mV in about an hour) I can live with it.

I noticed the electrical plug is two prong non polarized. Page four of the manual indicates a three prong plug for the US (I am located in Canada). The way it is wired currently is that one side goes to the fuse and then the front switch, the other side goes to the transformer. Instead of replacing the wiring inside, would the next best thing be to replace the plug with a three prong one such as this:

http://www.homedepot.com/p/Leviton-15-Amp-125-Volt-Double-Pole-3-Wire-Grounding-Plug-Black-R50-3W101-00E/205165472

while ensuring that: (1) the hot wire goes to the fuse/front switch and (2) a separate ground wire is run and connected directly to the case (via a ring terminal screwed from the outside) ?

[alm]

Though the load regulation is not within specifications -- 5.5A load drops about 25mV in about an hour (2A drops about 10mV in about an hour) I can
live with it.

You are not testing load regulation. Load regulation is the change in output voltage from unloaded to some percentage of maximum load on the time scale of a few seconds. You are testing long term stability, something different (and rarely specified for non-precision power supplies).

I noticed the electrical plug is two prong non polarized. Page four of the manual indicates a three prong plug for the US (I am located in Canada). The way it is wired currently is that one side goes to the fuse and then the front switch, the other side goes to the transformer. Instead of replacing the wiring inside, would the next best thing be to replace the plug with a three prong one such as this:

http://www.homedepot.com/p/Leviton-15-Amp-125-Volt-Double-Pole-3-Wire-Grounding-Plug-Black-R50-3W101-00E/205165472

while ensuring that: (1) the hot wire goes to the fuse/front switch and (2) a separate ground wire is run and connected directly to the case (via a ring terminal screwed from the outside) ?

I do not see the question? Connecting the wire to the switch to phase, the wire to the transformer as neutral and a third grounding wire to the case (assuming a steel case) sounds fine by me. Obviously the ground connection should be solid (should use a lock washer and might need to remove some paint around the screw). And for cable routing it might be easier to have the ground connection on the inside of the case. Obviously you would also need to replace the power cord by a three wire mains cord.

However, a single pole switch and fuse holder with a non-polarized plug is not necessarily a problem. The same power supply might have been sold in Germany with a non-polarized Schuko plug. Obviously it should have a ground connection, which implies polarized plug in the US.

[rstor22]

You are not testing load regulation. Load regulation is the change in output voltage from unloaded to some percentage of maximum load on the time scale of a few seconds. You are testing long term stability, something different (and rarely specified for non-precision power supplies).

I was expecting the specified load regulation for a power supply to be similar to its long term stability... thank you for clearifying that these generally won't be the same.

Using the immediate drop after the load was turned on, I get the following results:

For 5.5A the load regulation is 0.365% and for 2A it is 0.0517%
(The manual specifies a load regulation of 0.02% + 3mV)

There is also a note where it states this is "Against 0 - 100% change of output current) -- what does this mean?

I do not see the question? Connecting the wire to the switch to phase, the wire to the transformer as neutral and a third grounding wire to the case (assuming a steel case) sounds fine by me. Obviously the ground connection should be solid (should use a lock washer and might need to remove some paint around the screw). And for cable routing it might be easier to have the ground connection on the inside of the case. Obviously you would also need to replace the power cord by a three wire mains cord.

However, a single pole switch and fuse holder with a non-polarized plug is not necessarily a problem. The same power supply might have been sold in Germany with a non-polarized Schuko plug. Obviously it should have a ground connection, which implies polarized plug in the US.

I was thinking that instead of replacing the power cord with a three wire mains cord that I simply chop off the plug of the existing cable and install a three prong one (while ensuring the wire attached to hot goes to the fuse/switch). I believe the standard practice is that the hot goes to the fuse/switch, correct?. Then I would run a separate ground wire from the three prong plug to the outside of the case to ground it.

[alm]

Using the immediate drop after the load was turned on, I get the following results:

For 5.5A the load regulation is 0.365% and for 2A it is 0.0517%
(The manual specifies a load regulation of 0.02% + 3mV)

There is also a note where it states this is "Against 0 - 100% change of output current) -- what does this mean?

Are you measuring at the output binding posts (spade terminals work well for this)? At 5.5A, any length of conductor will drop additional voltage, something the power supply can not correct for. Does it make a difference if you change the load so the current is only 5.4 A. Could be that the current limit is starting to kick in (probably not, since it is already out of spec by 2A). I do not believe this power supply has output terminals on the back or options for remote sensing, so if you are measuring at the binding posts and make sure you are below the current limit, it sounds like the power supply is out of spec. Could be a filter capacitor that is dried out, causing the pass transistor to drop out at high currents.

0% to 100% change of output current refers to the maximum rated current, i.e. from 0 A to 5.5 A for this particular model. It is just written that way so they can use the same spec for multiple power supplies with different voltage/current ranges.

I was thinking that instead of replacing the power cord with a three wire mains cord that I simply chop off the plug of the existing cable and install a three prong one (while ensuring the wire attached to hot goes to the fuse/switch). I believe the standard practice is that the hot goes to the fuse/switch, correct?. Then I would run a separate ground wire from the three prong plug to the outside of the case to ground it.

Yes, hot to switch is correct. I am not sure about what relevant standards say about running the ground connection separately, but in general the design should be that the live/neutral wires should disconnect before ground does. In a plug I leave the ground wire longer so it will not get pulled out first. I do not think running a wire alongside the mains cord would meet this standard. I would just remove the current mains cord and replace it with a three wire cord with molded three-prong plug. You could even cut the IEC 60320 connector off a computer power cord and use that, assuming it fits through the strain relief. But do not use one discussed in this thread .

[rstor22]

Are you measuring at the output binding posts (spade terminals work well for this)? At 5.5A, any length of conductor will drop additional voltage, something the power supply can not correct for. Does it make a difference if you change the load so the current is only 5.4 A. Could be that the current limit is starting to kick in (probably not, since it is already out of spec by 2A). I do not believe this power supply has output terminals on the back or options for remote sensing, so if you are measuring at the binding posts and make sure you are below the current limit, it sounds like the power supply is out of spec.

Yes I am measuring at the output binding posts. The supply is rated at 5.5A however the current limit can be set higher than this. I brought down the maximum to 6.05A per the manual adjustment procedure indicating the maximum should be 110% of the rated current (before it was around 6.15A if I recall).

Could be a filter capacitor that is dried out, causing the pass transistor to drop out at high currents.

When testing the capacitors (on another power supply) I tested capacitance, ESR, leakage, and dielectric absorption. Are there any other tests that are recommended or should this cover the majority of faults?

[alm]

I would just measure the voltage before and after the pass transistor with a scope, and look for sufficient minimum headroom across the pass transistor as you increase the load. If this turns out to be too low, then I would start testing electrolytic caps for capacitance and ESR (or dissipation factor).

[Armadillo]

Check the ventilation fans. Could be raised in temperature or temperature compensation limit.

Open Up the top cover and blow with external table top fan to see if it will exhibit the problems of long term drift.

Just a suggestion.   

[rstor22]

I would just measure the voltage before and after the pass transistor with a scope, and look for sufficient minimum headroom across the pass transistor as you increase the load. If this turns out to be too low, then I would start testing electrolytic caps for capacitance and ESR (or dissipation factor).

If it comes down to testing the electrolytic caps, instead of removing the filter capacitors, is it safe to add capacitance in parallel to see if load regulation improves (see attached pic)? Then if it does, I can then proceed to remove and test individually?

Check the ventilation fans. Could be raised in temperature or temperature compensation limit.

Open Up the top cover and blow with external table top fan to see if it will exhibit the problems of long term drift.

The design is fanless, so perhaps adding a fan will increase stability. Thanks for the suggestion!

[alm]

Yes, paralleling a known good cap and seeing if the results show a substantial improvement is a good troubleshooting method. Just watch voltage ratings, polarity and do not go crazy on capacity to limit stress on upstream components like bridge rectifiers.