Inside the Owon P4603/P4305 linear PSU

[Pnoxi]

The manual says:

110 VAC ± 10% or 220 VAC ± 10% (Power is suppliedaccording to the marking on the left side of the powerinput socket on the rear panel of the instrument); ACinput 50/60 Hz

My version is 230V 50Hz. Since it is a linear source with a transformer, it is possible that the transformer will be different in the 110V version, so you have to buy the 110V version (If it exists).

[james38]

I know that this is an old Thread but i found an interesting info on how to calibrate this thing.
I have seen in this video where someone calibrate the current on the PSU.


To enter the special Menu do the folllowing:
Enter Calibration Menu:
System -> Up -> Ok -> Right -> Down

Be aware that the output is automatically ON!

For voltage calibration choose MAN VOLT.
You see a smal chart. Heading volt and then a number
You can navigate with up/down.
The small arrow dedicates which vale you can update.
The value can be updated with the rotary knob.

On the first rown you see under Volt a value which designates the voltage which should be on the Output. Check the voltage with an accurate multimeter.
Adjust the Value with the rotary knob until you get an nearly close value.

The second row  you see also a voltage but this is the measured voltage from the PSU. If you adjust the first row you will see that this value will be show another value.
Adjust the value with the rotary knob until you get an nearly close value to the voltage in the first row.

The P4305 hast in this Menu 4 calibration voltages which should be checked.
To change the calibration voltages got to the header row and
change the other values with the rotary knob.

On Cal-Menu 0 you adjust 50mV
On Cal-Menu 1 you adjust 100mV
On Cal-Menu 2 you adjust 1V
On Cal-Menu 3 you adjust 30V

To save the adjusted values do the following:
Save the Values:
System -> Up -> Ok -> Right -> Down

Thats all.
The same procedure are for MAN CURR.

The Menu VOLT,CURR AND ALL

Writes this on screen:
Start PSU wave from rectify
Meter communication failure
rebuf:

Don't know whats this meant.

Hope this info helps someone.

Regards Chris

[dietert1]

Recently we got 2x P4305 mainly because they are easy to automate using their serial interface. I found a "ODP Series DC Power Supply Programming Manual" on the CD that came with it. There are no manuals for this product, but manuals for similar OWON products.
Fan has temperature control, toroidal mains transformer. Wideband output noise is about 40 mVpp.
The incomplete overvoltage/overcurrent implementation mentioned above can be completed on the host side. I mean the host can clip the set voltage and current to be inside the limits before setting them.
Appended linearity tests are sweeps of measured output voltage deviation from set voltage, 0.1 V steps every 2 seconds. With and without load the average deviation is about -1 mV, for both units out of the box.
Comparison with a Keithley 2700 gave good results, too. I mean the built-in meters of the power supplies are well calibrated and agree to their last digit, e.g. 4,998 V was 4,99797 V on the K2700 with about +/- 20 uV short term stability. On the other unit 30.000 or 29.999 V was displayed and gave 30.0003 V on the K2700.

Best regards, Dieter

[dietert1]

Meanwhile i repeated the linearity test of one unit, using a Keithley 2700 to measure the output voltage (100 V range, Autorange off). And i extended the sweep scanning down after scanning up. Now it gets a little more interesting:

- Looks like they use the same voltage reference for the DAC generating the output voltage and the built-in meter. This way they hide most of the reference drift with temperature. It amounts to about -2.7 mV per 2 °C at 30 V out (45 ppm/K)

- The temperature dependence is also visible when scanning with load. Then the fan starts running at medium output voltages. That apparently affects the voltage reference. There is no such effect when running without load.

- While the measurements are rather smooth up to 17.5 V, above that there is a pattern in steps of 2 V. That DAC is somewhat "hybrid". The other unit doesn't have this, yet it has lost about 10 mV at 30 V.

On a closer look i can't really confirm the pretended "1 mV" quality.

Regards, Dieter

Edit: The voltage reference U4 sits near C8051F350 reference input pin 31. It is a TI REF3025 with device marking: R30C and specified with +/- 50 ppm/K. Chip appeared to be hand soldered.

[dietert1]

Another circuit common to the voltage regulator and the built-in DVM is the output sense amplifier, see appended schematic. In heating tests it appeared more sensitive than the reference. First i suspected leakage in the MLCC caps C152/C153 and replaced them by film caps. This did not help.
Will try with better divider resistors (100K 39K, 10K).
By the way, Agilent 6632B programming accuracy is +/- 10 mV. It may have other advantages, though.

Regards, Dieter

[dietert1]

Thinking about how to proceed, i decided to wire a Keithley 2700 with three channels of a 7706 multiplexer to record the P4305 output voltage, the reference voltage and the output of the output voltage measurement (schematic above). The diagrams once more show an up and then down scan of the P4305 set voltage from 0.01 to 30 V and then back. With a 8.25 Ohm load the fan starts running at about 15 V and doesn's stop until the end of the down scan.
Apparently the voltage reference is rather stable even with the fan running. The voltage measurement stage is the culprit and its deviations explain more than 80 or 90 % of the behaviour i reported above. Maybe the SMD resistors they used for the voltage divider aren't good enough.
Also note the P4305 output voltage deviations are now half of what i saw before when measuring voltage on its output terminals, with the load taking up to 3 A. Of course, in order to have the unit perform to 1 mV under load, it will need to operate in four wire mode. I will probably try to add a 4-pin socket on the front panel to support 2-wire mode with a shorting plug and 4-wire mode using external sense lines.

Regards, Dieter

[dietert1]

When i took out the rear RS232 interface to fit the missing buffer cap, it was already there.
Meanwhile i took out the six sense amplifier gain resistors and characterized them to be 0.1% 50ppm/K type. The 10K resistors were about 36 ppm/K while the 100k+39k gave -18 ppmm/K. So this explains the 50 ppm/K output deviation seen before.
I inserted 10K, 2 ppm/K and 47K, 5 ppm/K SMD resistors (3x 47K = 141K) and they solved the temperature dependence. Using the description above (by james38) i could calibrate the unit with the new resistors, except below 1 V. There remains some deviation of up to 12 mV. Note that 912 mV / 900 mV * 139,2 KOhm = 141 KOhm.
I also added the 4W connector. It's a small 4 pin connector (binder 719 09-9764-70-04), in order to support 2W mode with a shorting plug. Now i can say it's a 1 mV unit.

Regards Dieter

[dietert1]

In order to reduce the residual deviations below 1 V output voltage i ordered some 0.1 % resistors 1M and 57K6. They change a 47K precision resistor to 45K, so one gets closer to the original: 100K + 39K = 2 * 47K + 45K. Now the linearity sweep stays within +/- 1 mV.
The no load case has a minor deviation between upward and downward scan. During the downward scan the 5 sec delay time after each 0.1 V adjustment leaves charge in the output caps to settle.

Regards, Dieter