Cheap benchtop PSU upgrade

[ttodorov]

Hi everyone,

(NOTE: This project is not complete yet! I have some questions about the current PSU design, which I need to understand first, hence this early post)

I have had this cheap switching benchtop PSU from MPJA.com for years now. The model is 9615-PS and it can do 0-30V 0-3A (pic below). For some years that was my only PSU, and even after I got a Rigol DP932 a year ago, I can't bring myself to ditch this one. I like the form factor ;-)

Lately the output connectors have gotten loose and I also read some posts on the Internet, which state that the output ripple would be significantly reduced, if the output filter capacitor soldered between the binding posts is replaced with a better quality one. Since the also doesn't display the set current, but only the drawn one, I compiled a wish list of upgrades that I would like to accomplish. Because this is not my only PSU, I thought that would make a nice project. Here is my TODO list:

- Better binding posts
- Add a plug for earth ground
- Replace output filtering caps with high-quality ones
- Replace the interface with a better one (this has multiple points)
  - replace the character display with color OLED
  - display both set and measured voltage and current
  - because the voltage and current setting pots are very coarse, replace them with DACs

The first 3 bulletpoints are no issue. To see if the last major wish with its subpoints is feasible, I opened the case and tried to figure out how the PSU works. It seems that the heart of it is a TL494C PWM controller which handles the 2 error amplifiers for voltage and current. The 2 pots that set voltage and current just feed the control board with a voltage in a certain range and set the Vmax/Imax respectively. After some thought, I decided that I don't need to touch this controller. I believe that using a dual-channel 16bit DAC like the MAX5136 with opamp buffered outputs will let me replace the pots entirely. This would easily let me have 1mv and 1ma resolution on the settings. I could use an ARM chip to drive the DAC, which will also take care of diving a better display.

As to the measurement of the actual values, I reverse-engineered the display board of the PSU (schematic attached). I took some pictures which I blew up on my monitor, and a multimeter to trace all connections. Basically, it uses a LM358 to measure output V and I, which feeds a PIC16 MCU, which puts it on the display through a 7segment LCD driver. The LM358 uses 10V +V and -5V -V, while the rest is all 5V supply and logic. It's on a 2 layer board with a rather bad design.

If anyone cares to have a look at the schematic that I produced, I have these questions:

1) The positive supply of the LM358 goes through a reverse-biased diode. When I took measurements with the multimeter, I got 9.4V on the +V pin of the opamp. When I bread-boarded this circuit I put in a low current schottky diode, but it didn't work - what I measured at the opamp on the bread-board was -4.5V (which is incidentally the -5.1V on the negative rail less the diode forward voltage drop). Is this some special type of diode? Or is this something common and I just have the wrong value in there? What is the purpose of a diode in reverse in that circuit? (On the attached picture, the diode is in the upper right corner of the board)

2)  Why did they add more offset to the opamp inputs? On the schematic (as found on the actual display board) the resistors R4/R5 and R24/R25 just add 50mV additional offset... for what reason? I mean, it's a dual-rail configuration. From all I know (which admittedly is not much), having 0V on one of the inputs, there should be no problem to bring the output to 0V with a negative rail available... Isn't that the desirable state?

I would appreciate any insights - thanks for the help!

Best regards!