MeanWell 350 Repair

[bryan]

Hello:

Have a Meanwell 350 Power Supply 2v v model rated to 14.6A output current. Works fine if the load is under 4 amps, but once the current is increased it starts hissing and the output voltage drops to 12v. Some have suggested Q1 and Q2. I note that Q5 gets very hot seconds after connecting a load greater than 4 amps. Q6 stays warm.

Attached is a schematic. Any ideas where I should start looking.

http://www.hordsoffun.com/ebike/pics/Meanwell_S-350-27.png

 

[Nigel Goodwin]

Q5 does nothing except switch the fan ON and OFF, so shouldn't get warm at all.

I would suggest the most likely problem is high-ESR electrolytics, including the main smoothers C5/C6 - SMPSU's often work to a lesser extent with the smoothers seriously high ESR, but fail under higher loads.

However, the circuit seems wrong on many counts, not least all the transistors drawn as PNP.

 

[bryan]

Thanks Nigel, yes looked at the schematic again and it does not even appear to match the board. I suspect this is a counterfeit product. Bought off Ebay. Labelled as a Meanwell supply...yeah right. Many components on the board are not even present. For example none of the fan circuit components are present. Was wondering why the fan was on full power all the time. I guess the DONGBAO capacitors are not quality caps

 

[bryan]

Well checked the caps in the switching side of things and the output regulator. ESR was fine for all, as well the capacitance values within the expected 10-20%

 

[Mosaic]

Scope the output of the Tl494. Hissing usually means audible piezoelectric effect off your caps. I have seen a TL494 'shift' its PWM output oscillation frequency downward in binary steps ( by implementing additional dead time) to control the target current load, then jumps back to the original base frequency when the load is corrected, causing a modulated PWM pulse train which 'sounds' different because of the caps 'buzzing'. If you can detect such unexpected PWM frequency shifting over a few PWM cycles you need to look closely at the error amp circuitry being faulty. This type of unwanted modulation increases heat in the output driver circuitry. Sometimes you can correct this by adding capacitance to the error amp input to act as a simple low pass filter thereby preventing over corrections. Reducing the error amps feedback (amplification) can help a bit as well. Adding additional decoupling close to the TL494 supply pins is also a good idea.

I have found it preferable to use an external, low slew, dual opamp to amplify a precision current sense resistor's V and a difference amp to directly control the DTC input as a means of reliable control when designing with the TL494. I use its internal error amps as thermistor based current limiters.

 

[BGAmodz]

Scope the output of the Tl494. Hissing usually means audible piezoelectric effect off your caps. I have seen a TL494 'shift' its PWM output oscillation frequency downward in binary steps ( by implementing additional dead time) to control the target current load, then jumps back to the original base frequency when the load is corrected, causing a modulated PWM pulse train which 'sounds' different because of the caps 'buzzing'. If you can detect such unexpected PWM frequency shifting over a few PWM cycles you need to look closely at the error amp circuitry being faulty. This type of unwanted modulation increases heat in the output driver circuitry. Sometimes you can correct this by adding capacitance to the error amp input to act as a simple low pass filter thereby preventing over corrections. Reducing the error amps feedback (amplification) can help a bit as well. Adding additional decoupling close to the TL494 supply pins is also a good idea.

I have found it preferable to use an external, low slew, dual opamp to amplify a precision current sense resistor's V and a difference amp to directly control the DTC input as a means of reliable control when designing with the TL494. I use its internal error amps as thermistor based current limiters.

Nice explanation mosaic , so what if the TL494 is outputing a frequency that is lower than expected , for example 15 Khz instead of 60 Khz , with no load ?

 

[Mosaic]

A minimum load on the SMPS is required for the control loop to work, if there is none the control loop can become unstable with unpredictable regulation results. Loads that are too small can cause a single pulse from the TL494 to overshoot the regulation if wide pulse widths are permitted (for heavy loads). This would then sound like the SMPS is sputtering with erratic pulsing.
Reduce the max pulse width via the dead time control voltage (increase V to reduce max. pulse width) for better control of tiny loads. If you have small output transistors then wider pulse widths are ok, beefy output stages require small pulse widths for low loads. higher frequency operation permits overall net smaller pulsewidths with better decimation and control. By just controlling the DTC rather than using the error amps to start with you can have both tiny and large load control.

The meanwell is controlling its voltage regulation and probably uses an error amp as a comparator to shut down if its current or temperature rating is exceeded.

 

[bryan]

Thanks everyone for the replies. Can have a look at the output of the TL494, but looking at this thing it is so cheaply built and no doubt a counterfeit. Think I am better off to just bin it. I am not even sure if it is even designed to properly handle the stated max load of 14.6 amps.