Problem with Hypex SMPS400/67 (OEM V4)

[tatus]

I have built a stereo amp with 2xUcD400, one SMPS400/67, plus a total of 40mF inbetween them (20mF from + to GND, 20mF from GND to -, so the SMPS "sees" 10mF). I had (probably falsely) assumed that the SMPS would use current limiting when charging the caps at startup. Instead it seems to "hick them up". When looking at the bus voltage with a scope, the voltage develops in steps of about 20V. It takes 5 to 6 steps with a pause of approx 1 sec inbetween until the voltage stabilizes. Luckily the UcD400's have little ildle losses but they switch on/off/on during this process (the SMPS controls the shutdown pins of the amps).

Does anyone see the same behavior?

I would now like to modifiy the SMPS to either shorten the hick-up recovery delay, or if possible to implement a continuous current limit. Any hints are appreciated, it would be best to receive the schematic diagram in order not to have to reverse engineer the entire circuit.

Frank

[luka]

I don't understand why you even put that much capacitance there, this is not 50Hz trafo and analog amp. Remove all extra caps and try that if it works and it should

[tatus]

@luka: there are several reasons why I put this much extra capacitance:
- that SMPS acts in fact like a 50hz trafo. It's output voltage is proportional to the voltage across the primary bus capacitors. And that capacitor's voltage creates (50hz) ripple as the amplifier starts drawing current.
- the SMPS has medium sized primary bus capacitors, but only very small secondary ones. Speaking in terms of ESR, I have to take the primary bridge, the entire transformer and the secondary rectifier into account. This would not result in a "stiff" supply for the amp.
- the SMPS' total capacitance is just too small (around "virtual" 2.5mF in total) for 600W peak output power for the amps. I would lose too much drive voltage due to excessive ripple.

[luka]

then better option would be to put PFC in front of it, you won't ever have ripple problem again, also output will be 10x more stiff as is now... isn't this a regulated smps (from what you said, its not), which would make output voltage stiff in any case, also 600w peak is not much...
But I will ask, did you test this or you think it will behave like that, coz it should be able to drive that amp with no problem.

I know that if smps is good for the power, and doesn't have much primary capacitance (I have 2x 470u for half bridge), PFC will make it look like its several times better and coz it will be...

AND you will have good PF and able to drive your system from low or high AC lines

[tatus]

Yes, the SMPS is unregulated, it is a module from Hypex intended for use as a transformer replacement (smaller, lighter). They follow a different approach there, as normal regulated SMPS, with or without PFC, can be subject to power pumping. This module has the disadvantage of being unregulated, but at the same time it is 2-quadrant capable, taking back energy back from the amp in case.

Regarding the capacitor dimensioning, I rely on calculations and simulations. For example, 600W = 130V * 4.6A. (130V is the typical SMPS output voltage.) For a 2.5mF capacitor, this results in approx. 30V ripple. This can be handled well by the Hypex UcD modules as they have a good PSRR, but this limits the peak power handling capability of the entire amp.

PFC would certainly help, but only when followed by a regulated converter. This in turn brings the power pumping problem back on the table. Hypes has made their experience with this topology as well, and they moved away from it for good reasons I think.

As a bonus, this unregulated concept results in a very simple and also very efficient design. I have tested it, the SMPS can run at 300W continuosly at the table, and you have to look at it. It has no massive heat sink, just two really small aluminium sheets.

[Jan-Peter]

Tatus,

our SMPS is with synchronized rectifying, the secondary caps 'sees' also the caps on the primary. This in totals gives you a much stiffer PSU as with a 400VA linear power transformer.

It has really no sense at all to use so much extra secondary capacity.

[tatus]

Hi Jan-Peter,

I have taken this into account. Correct me if I am wrong: the primary capacitance is C1=410uF (two 820uF in series, 230V operation). Charged at V1=325VDC, the stored energy is W=0.5*C1*V1^2 = 21.6 joules. Translating this to the secondary side: a capacitor with the same energy storage capacity at the output voltage of V2=134VDC would have the value C2=W/0.5/V2^2 = 2412uF. Adding the "real" output capacitance of 150uF gives a total of the above mentioned 2.5mF.

The simplified circuit of the entire SMPS should then be a 134V 50Hz sine voltage source, followed by a full bridge rectifier, plus that 2.5mF capacitance at the output. Plus some ESR that I also consider not being neglectable.

The ripple voltage at 600W (or 4.5A - I am driving two UcD400's with one SMPS) load can then roughly be estimated to Vripple <= I*t/C2 = 4.5*1/100*2.5m = 18V. (I see that I made a mistake in the calculation of my previous post...) A quick spice simulation tells me Vripple=14.8V.

This is more than 10% of the output voltage, and this means to me that when excited with deep bass, the amp will run into clipping too early in this situation. This is why I added that extra output capacitance. Having added that 10mF, the ripple is lowered to 3.2V, plus I get a lot better ESR from this setup.

Anyway, if someone can tell me that this is all nonsense I would be very happy, because I want to fit everything into a quite small Hammond steel/wood case, and the only parts that make me headache are these capacitors. I managed to find some with only 51mm height, but this is still too much and I will need to mount them horizontally in the case...

Until then, I would be happy to get a solution how to prevent the SMPS from running into hiccup mode.

[tatus]

@Jan-Peter,

I hope that I can start a discussion on capacitor selection for UcD/SMPS. What do you think of my considerations above, does that make sense?

[Kv3audio]

A good, and well designed SMPS, should easily handle 2 x 6800uF per rail as extra capacitors at the output.

Good Luck