Electronics > Projects, Designs, and Technical Stuff
Manson SPS-9602 (30V 30A) lab supply repair and possible upgrade (even SCPI)?
(1/1)
Yansi:
Hello,
today I have scored a Manson SPS-9602 for just 18 bucks, sold in our local "ratshack"-like shop, marked as "faulty, needs repair". It is a 1-30V 30A lab supply, switchmode one.
So I have of course taken it home, this PSU seems to be a quite an expensive one when new and/or working.
What was the fault you ask? Well, shitty cheap ass design, shaving bucks on using a proper NTC thermistor on the mains input, being replaced by just a cheap white ceramic resistors, actually two 15ohm ones in parallel. At least they got a thermal cutout. Not sure whether the cutout shall be temporary or it cuts out for good, anyway both of the resistors blown open.
From my experience, these cheap white ceramic resistors are really shite and never shall be heavily overloaded, even for short periods of time. Defects in the potting may contain bubbles, providing insufficient cooling for parts of the resistive wire inside - where it just splits open when overloaded.
So I will order a set of proper NTC thermistors and put two in series to obtain the most resistance I can get in there. Originally, the charging resistance was 15/2 = 7.5ohm, I will replace it by two at least 6R at least 7A thermistors. (Let's see what I will find).
Meanwhile the thermistors will be on order, I got thinking about modifying the PSU: In its original state, it does not have any means to set the current limit. Which is strange, as the PSU can already do a constant current limit, according to the manual, but fixed at 30A. There is a shunt resistor with OP27 near by, so quite decent sensing to say.
The original control PCB module (based on TL494, booo! Who wouldn't have guessed!) seems to have a couple of connectors unpopulated and some components missing near them too. I'd guess this may be the missing current limit setting option.
I need to get the module desoldered out of the power converter board and start reverse engineering it.
But then I thought: If already reverse engineering it, why not making a better upgrade straight away? Adding a current setting knob on the front panel would become ugly (no good spot to put it on). So then I thought: If I could make a whole new control module, I could even add a small MCU with ADC and DAC and replace the voltage pot on the front panel with an encoder, which would allow for precise control of both the current and voltage. Unlike the original single-turn pot, which I find rather impractical with a 30V range. Too coarse of a setting.
The battle-plan therefore is to reverse engineer the pinout of the module (will be straightforward, just a couple of pins there containing power supply, gate driving signals and some feedback loop pins from current sense). I would preserve the original control circuitry with the TL494 and leave original compensation networks, I would just mod the circuits setting the V and I with a suitable DAC (probably, 12bit would be probably enough).
Adding an MCU in there with the DACs would allow to get rid off of the stupid analog control connector and replace it with at least an RS232 (SCPI anyone?)
Likely the front panel with the meters (I sniff a couple of ICL7106 on there) would need rework in this case too, to be able to display the V I settings. That would not be problem. What is a shame is that there are only three digits, but it would be possible to switch to smaller (narrower) LED displays and fit 4 digits in there. The PSU has even remote sense terminals, so it would make sense (no pun intended) for the increased digits.
What makes stuff easy, from a quick looking at the PSU guts, there is a whole isolated 5V supply available for the front panel meters and then there is a dedicated PSU for the control electronics and I have a feeling it may be even a symmetric supply for the opamps. At least the OP27 seems to be wired for a symmetric supply voltage.
Guessing other people may possess this exact PSU, would there be any interest in engineering such an upgrade? The PSU guts seems to be decent enough to justify this lil upgrade. I am pretty confident this could be done with as little as possible changes to the rest of the PSU. Just replacing the small control board module and likely the front panel display board and optionally hacking in a hole for D-SUB 9 for RS232 on the back side.
//EDIT: The TL494 is hiding behind the wire nest, behind the ferrite choke there.
//EDIT2: Fixed typos and added a photo of temporarily hacked-in NTCs with not enough resistance.
Yansi:
A couple of notes about the PSU itself, just by looking into its guts:
It has an active PFC based on a pretty archaic transition mode controller Motorola MC33262 (datasheet). The PFC output utilizes a pair of 470uF caps, which I find pretty excessive for the power output.
The main DCDC converter is the classic symmetrical half-bridge, controlled by a TL494 through a gate drive transformer, driven by a bunch of TO92 trannies (I do not see any diodes in there, possible design fuckup on Manson side?). Secondary winding is a split type, with a pair of MUR3020WT diodes. A glimpse at the control board dictates the half-bridge may be running at about 35kHz (1nF and 15k oscillator components).
PFC uses a two paralleled SiHG20N50C mosfets, while the main DCDC uses two paralleled pairs of IXTQ22N50, which seem to be quite decent MOSFETs actually.
The noise filtering components seem to be placed and designed sensibly, so the PSU shall have decent output cleanliness. But I have not made any measurements yet, waiting for the thermistors to repair the input section first.
The cooling fan seems to not be temperature controlled, which I find a bit on the annoyance side, but based on what to take the temperature measurement? There are three separate heatsinks and the inductors can also contribute to heat significantly. But looking at the heatsink sizes, the output diodes will make the most heat and would be likely good to base the temperature sensing on.
There is one design fail for sure: The DC link in between the PFC board and the DCDC board is fused by a small 5x20mm glass fuse, rated at just 250V. Stupid! In case of an actual short, it would explode like a small grenade.
Otherwise I do not have any clear problems with the PSU design, it seems to be decent built.
Navigation
[0] Message Index
Go to full version