Maker Pro
Maker Pro

My new toy

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I am now the proud owner of a Keithley 228 Voltage/Current source.

Not exactly the newest piece of equipment on the block, but very useful for a lot of stuff I want to do.

I've also picked up an Advantest R6441A bench multimeter.

Perhaps I should be looking for an HP-IB (IEEE-488) interface for my lab pc too.

These are significantly complex pieces of equipment, even if they are long obsoleted...
 

GreenGiant

Feb 9, 2012
842
Joined
Feb 9, 2012
Messages
842
we used a couple variations of these at my last job, they are great once they are set up.

Labview had libraries for them to interface fairly easily with, I know there were good interfaces with C# as well.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
we used a couple variations of these at my last job, they are great once they are set up.

Labview had libraries for them to interface fairly easily with, I know there were good interfaces with C# as well.

Excellent information, thanks.

Nice, bit of a beast by the looks of it.

In more ways than one. Fortunately the (easily available) manual for the Keithley 228 has full schematics, parts lists (albeit with some propriety part numbers), maintenance, and disassembly instructions.

Whilst it is marked as 110V on the outside, there is a switch on the inside which allows you to change from nominally 115V to nominally 230V. Mine is still fused for 110V, but I took the chance powering it up and it all came up without problems.

SHMBO is getting me a set of fuses of the correct capacity this morning, so I should be able to run some of the basic tests this evening. The more advanced tests require a 1 ohm 100W resistor which I may have to build up using a series/parallel array of smaller resistors.

The Advantest meter seemed to be within spec on all the ranges I could test. What was initially worrying (meter not going to zero on voltage ranges) turned out to be due to the very high input impedance. Shorting out the meter probes (or connecting it to an actual voltage source) made the display stable.

The only range I have concerns about is the AC current range which seems to zero at 0.3mA rather than zero.

I feel like a kid in a candy store :D
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Don't we all at times :)

I try :)

Last night I swapped the fuses over to the ones recommended for 220-250V.

The recommended fuse for the switchmode supply is 2.5A. SWMBO grabbed me a couple of fast blow 3A fuses as 2.5A was unavailable (same thing with the 375mA slow blow fuse, she got 500mA).

The 2.5A fuse replaces the 5A one for 110V.

It promptly blew (on power-on).

It's a fast blow ceramic style 3AG fuse. Since there's a significant switch on transient for the SMPS, I guess that's what did it. I'm also wondering if that was a faster fuse than I expected? So, I'm back on the 5A fuse. This is a little bit of a worry.

I'm going to check the schematic to see if there's a slow start for the SMPS, I'd bet there's not.

Initial testing suggests that the voltage output is within 1% (probably within 0.25%) of the current range. That's probably close enough, but since the manual tells you how to calibrate it, I might do that...

I haven't tried the current accuracy yet.
 

shumifan50

Jan 16, 2014
579
Joined
Jan 16, 2014
Messages
579
Please don't take exception, I am trying to learn something here:

Why is the fuse amperage halved for 110V. I would expect the amps to go up for the same energy if the voltage is lower. I assume the 5A is for 230V. So should the fuse not be 10A for 110V. Again sorry for the stupid question.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Please don't take exception, I am trying to learn something here:

Why is the fuse amperage halved for 110V. I would expect the amps to go up for the same energy if the voltage is lower. I assume the 5A is for 230V. So should the fuse not be 10A for 110V. Again sorry for the stupid question.

Good question.

The answer is that the power supply draws a certain power from the mains. When the voltage doubles, the current halves.

In the case of the linear part of the power supply, switching the big internal switch I think I mentioned changes the primary windings of a transformer from parallel to series.

For the switchmode power supply I don't think anything changes (but I haven't studied the schematic). presumably the capacitor is charged to a higher voltage, the switching transistor is on for approximately half the time, thus the average current is halved also.

For anyone playing along at home, here is a handy copy of the user manual.

Next time I have it open I'll take a few pictures. There's not a lot to see because the airflow management hides a lot of stuff. What you can see is very old school and very nicely constructed.

Oh, one interesting detail is that the internal fan is 110V and is wired across one of the primary windings of the linear power supply transformer. It used the two primary windings as an autotransformer when configured for 220-250V
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I have just run some tests using an 8 ohm 50W resistor as a load.

I started off using the flimsy wires that were attached to the unit and discovered quite quickly what less than 0.2 of an ohm of resistance in each wire can do, even at relatively low currents.

I've done a quick check by cutting them really short, but I'll have to replace them with far heavier cable, and preferably use the remote voltage sensing.

Off-load, the voltage seems to be within +0.02% and -0.16% between 1V and 10V. The spec is 0.1% + 10mV and it's well within that.

The largest percentage error is at 2.5V where it measures 2.496V.

The largest absolute error is at 6.5V where the measured output is 6.495V

Mathematically compensating for the resistance of the leads gives an accuracy of within +/- 0.25% which exceeds the spec, but uses a current measurement that only has 3 significant digits.

Interestingly, both methods yield a systematic (i.e. average) error of 0.4%

Even more interestingly, if one looks at the standard deviation, it is 0.04%, so more than 99% of readings are likely to be within +/- 0.12% of the central value.

This is a good indication that even if it's out of calibration, it's likely not out by a great deal. :)
 
Last edited:
Top