EEVblog #814 - Keysight N8762A 600V 5100W PSU Teardown

[bktemp]

Why 3 bridge rectifiers?

Alexander.

Because it's 3-phase?

3 phase input has 3 wires. 3x bridge rectifiers have 6 inputs. For a 3-phase bridge rectifier only 6 diodes are needed, but her are 3x 4 diodes used.

It would be interesting to know what the difference between the 208V and  the 400V version is. The rectifier sections looks totally different on the 400V version.
I had expected 3 PFC circuits, but it looks like only 1 unit is fed by the rectified output from all 3 phases. I have know idea how this can work.

[Monkeh]

Why 3 bridge rectifiers?

Alexander.

Because it's 3-phase?

3 phase input has 3 wires. 3x bridge rectifiers have 6 inputs. For a 3-phase bridge rectifier only 6 diodes are needed, but her are 3x 4 diodes used.

Because purpose built bridge rectifiers are cheap, durable, easily heatsinked, and in stock in huge quantities. It's entirely normal.

[Yansi]

You just connect the ~ pins together and get a double diode in one easily heatsinkable package. Then take three of these and make cheap 3f bridge.

[bigsky]

Why 3 bridge rectifiers?

Alexander.

Because it's 3-phase?

3 phase input has 3 wires. 3x bridge rectifiers have 6 inputs. For a 3-phase bridge rectifier only 6 diodes are needed, but her are 3x 4 diodes used.

Because purpose built bridge rectifiers are cheap, durable, easily heatsinked, and in stock in huge quantities. It's entirely normal.

And you can use the same part for both single phase and three phase boards, which has many benefits.

[station240]

Technically you only need 1.5 (eg 2) bridge rectifiers as you have 2 inputs per package.
However as the did the output stage with 2 parallel diodes (another bridge rectifier), then no reason not to do the same with the input.

Alternative is 3 twin diode packages, with the correct A-KA-K (common A and K), which is available as IGBT bricks, and I suspect smaller packages as well.

[German_EE]

For all young players (and those not used to three-phase electrical circuits) the correct terminology here is either wye-delta or star delta https://en.wikipedia.org/wiki/Three-phase_electric_power What Dave has is a delta three phase unit with three live connections and no neutral (not very common in an office or home environment). The normal way of doing things is the wye or star circuit where there are three live connections PLUS a neutral lead.

I have seen delta wired units that have been brought to life on a single phase supply by connecting live to L1, neutral to L2 and a motor start capacitor between L1 and L3 but this was an inductive load and not a piece of electronic equipment. A certain amount of care needs to be taken here or the magic smoke may escape again.

[Phoenix]

I'd actually like to see a davecad™ of the PFC section.  I get how single phase active PFC works, but 3 phase wouldnt play so nice with the simple "plop a boost converter after the rectifier" type used in single phase supplies. I'm seeing a few different typologies online, and I'm curious as to what someone like TDK lambda would use.

I'm quite curious too. I can spot 2 magnetics that look like they might be 2 PFC inductors... Can't find a topology that fits for that though.

[max666]

For all young players (and those not used to three-phase electrical circuits) the correct terminology here is either wye-delta or star delta https://en.wikipedia.org/wiki/Three-phase_electric_power What Dave has is a delta three phase unit with three live connections and no neutral (not very common in an office or home environment). The normal way of doing things is the wye or star circuit where there are three live connections PLUS a neutral lead.

I have seen delta wired units that have been brought to life on a single phase supply by connecting live to L1, neutral to L2 and a motor start capacitor between L1 and L3 but this was an inductive load and not a piece of electronic equipment. A certain amount of care needs to be taken here or the magic smoke may escape again.

I'm not claiming to know enough about Three-phase power to make a statement here, but how can you distinguish whether Dave has a star or delta unit? The neutral in a star configuration is optional, so the 3 bridge rectifiers could be wired in a delta as well as star configuration, can they not?

And I venture the motor start capacitor method would be even more dangerous than the suggested VFD method. The capacitance needed for a motor start capacitor is dependent on the inductance of the winding and in turn of the loading, so I guess depending on what the electronic load does you could even get over-voltage at the input by connecting a motor start capacitor. 

And since a lot of people are wondering how the PFC is done, and again I'm just talking out of my ass here, but if I'm not mistaken for single phase PFC you have to make sure that your current draw looks like this:

so I would say all you have to do for three-phase rectifiers, is make sure the current draw after them looks like this:

[rs20]

...so I would say all you have to do for three-phase rectifiers, is make sure the current draw after them looks like this...

Incorrect, the three phase rectifier will always draw all current from the phase with the highest instantaneous voltage; which means a naive implementation of PFC after a three-phase rectifier will draw dramatically non-sinusoidal currents from each phase. Example:



Here we see a Y input going through a three-way rectifier, and then into R1, which is perfectly phase-correct by virtue of being a resistor (put another way, it implements your suggestion of keeping current in proportion to voltage). As you can see, the current waveform (red) is dramatically non-sinusoidal, as compared to the corresponding phase voltage waveform (green).

EDIT: Fixed broken image link.

[Phoenix]

so I would say all you have to do for three-phase rectifiers, is make sure the current draw after them looks like this:

Not very sinusoidal once you unravel it, but yes that would be the simplest method and only requires a single inductor (video appears to have 2 inductors).

There are many better topologies, check out this guys lit review:
http://scholar.lib.vt.edu/theses/available/etd-08142002-075617/unrestricted/Barbosa_ETD.pdf
Something like Fig. 1-6 is feasible for the Agilent unit as it has 2 inductors (visible in Dave's video at 13:06) and a whole bunch of to220 t0247's on those heatsinks labelled P 200 5k.

Some more good material:
https://www.pes.ee.ethz.ch/uploads/tx_ethstudies/Kolar_Essence_Three-Phase_PFC_02.pdf
https://www.pes.ee.ethz.ch/uploads/tx_ethpublications/08_Towards_a_99__Efficient_ThreePhase.pdf

[bktemp]

so I would say all you have to do for three-phase rectifiers, is make sure the current draw after them looks like this:

Not very sinusoidal once you unravel it, but yes that would be the simplest method and only requires a single inductor (video appears to have 2 inductors).

There are many better topologies, check out this guys lit review:
http://scholar.lib.vt.edu/theses/available/etd-08142002-075617/unrestricted/Barbosa_ETD.pdf
Something like Fig. 1-6 is feasible for the Agilent unit as it has 2 inductors (visible in Dave's video at 13:06) and a whole bunch of to220 t0247's on those heatsinks labelled P 200 5k.

I can see only 2 thick wires between the rectifier board and the PFC, so they are using either the simple PFC with 2 channels interleaved in parallel, or they are using some other clever circuit, but I can not see how this could work without a connection to each phase.

[rs20]

Not very sinusoidal once you unravel it, but yes that would be the simplest method and only requires a single inductor (video appears to have 2 inductors).

What? No, it's not a viable method at all, as I've already demonstrated...

[alxpo]

And I venture the motor start capacitor method would be even more dangerous than the suggested VFD method. The capacitance needed for a motor start capacitor is dependent on the inductance of the winding and in turn of the loading, so I guess depending on what the electronic load does you could even get over-voltage at the input by connecting a motor start capacitor. 

There is more funny way to provide 3 phases for testing - connect 3-phase motor with appropriate phase-shift capacitor in parallel to PSU. Sure, power of motor should be bigger than idle power of PSU and power of test load (I suppose quite small test load).

[German_EE]

As I said, the only time I have seen the start capacitor trick used was on an inductive load (which was of course an electric motor) although I do not recall how the value of the capacitor was worked out as this took place over thirty years ago. As for the Wye including a neutral connection, it is of course possible to connect a three-phase motor in Wye configuration and ignore the neutral center tap but all Wye supplies that I have seen include a neutral return.

Connecting this unit in Dave's lab environment could be a problem.

[drago]

I bet Dave has 3 phases distributed in his lab across different points of load. Extension cords and little bit of fiddling could do the job    

[Monkeh]

I bet Dave has 3 phases distributed in his lab across different points of load. Extension cords and little bit of fiddling could do the job    

I doubt it, it's a total waste of wire and effort running multiple phases to a small office like that.

[max666]

Incorrect, the three phase rectifier will always draw all current from the phase with the highest instantaneous voltage; which means a naive implementation of PFC after a three-phase rectifier will draw dramatically non-sinusoidal currents from each phase. Example:



Here we see a Y input going through a three-way rectifier, and then into R1, which is perfectly phase-correct by virtue of being a resistor (put another way, it implements your suggestion of keeping current in proportion to voltage). As you can see, the current waveform (red) is dramatically non-sinusoidal, as compared to the corresponding phase voltage waveform (green).

Ahh yes, I can see now where my naive picture went wrong. Thanks for the simulation 
But I have to say, compared to a rectifier going straight into bulk capacitance, the above PF doesn't look that shabby to me. I guess around 0.85 maybe?
Can LTspice calculate the PF? Otherwise I'll see if I can take a shot at it, after I get some sleep.

Edit: I really should go to sleep. My sleepy state math gets me a PF of 0.83 

[SeanB]

I bet Dave has 3 phases distributed in his lab across different points of load. Extension cords and little bit of fiddling could do the job    

I doubt it, it's a total waste of wire and effort running multiple phases to a small office like that.

Distribution board for the floor, or the group of offices, would be 3 phase. The offices might have it supplied, as you can get by with a cheaper thinner cable, you need a 16mm cable and neutral for a 60A single phase supply, but for a 30A 3 phase supply you can get by with 4mm cable, around half the copper. The power delivered is the same, but spread out over 3 conductors, and the neutral only has to handle the imbalance and harmonics. Thinner cable is easier to work, as you can fit it in standard 25mm conduit as well, you do not need to have a 50mm conduit and a termination at the top in the duct.

[Monkeh]

I bet Dave has 3 phases distributed in his lab across different points of load. Extension cords and little bit of fiddling could do the job    

I doubt it, it's a total waste of wire and effort running multiple phases to a small office like that.

Distribution board for the floor, or the group of offices, would be 3 phase. The offices might have it supplied, as you can get by with a cheaper thinner cable, you need a 16mm cable and neutral for a 60A single phase supply, but for a 30A 3 phase supply you can get by with 4mm cable, around half the copper. The power delivered is the same, but spread out over 3 conductors, and the neutral only has to handle the imbalance and harmonics. Thinner cable is easier to work, as you can fit it in standard 25mm conduit as well, you do not need to have a 50mm conduit and a termination at the top in the duct.

I doubt he has a 60A supply, you don't need necessarily need 16mm² for 60A, and the cost of a three-phase board and the breakers is likely greater than the cost of the cable.

You also then have to take into account the presence of multiple phases in an environment occupied by people with no concept of them.

[k2teknik]

I doubt he has a 60A supply, you don't need necessarily need 16mm² for 60A, and the cost of a three-phase board and the breakers is likely greater than the cost of the cable.

Well these kind of stuff is Country specific, very Country specific, no need to claim that on is better, or more right than the other.
These thing is also regulated by local law, which also differ from Country to Country, again it do not give any sense to discus what is right/best, just be aware that this stuff differ from location to location.

[SeanB]

You only need a 3 phase incoming isolator, all others are single phase in a small office. 3 phase is good as you might want an aircon of more than 30 000 BTU which is almost invariably a 3 phase unit.  Even allowing for a 10mm cable you have 20mm of copper, and a PE of 6mm, as opposed to the 3 phase one being 16mm of copper and a PE of 2.5mm. The internal wiring to the outlets and lighting plugs will still be 2.5mm and 1.5mm respectively.

This applies to South Africa, it is very similar in Australia for cable size, as this is dictated by allowable temperature rise in the cable when installed either bare in plaster, in conduit or in free air in a cable way in the building.

[Monkeh]

You only need a 3 phase incoming isolator, all others are single phase in a small office.

You still need a 3-phase board. 3-phase MCBs are no more expensive than single-phase (they are in fact the same MCB with a slightly different case and a link), however going to a 3-phase system requires DIFFERENT MCBs (even single-phase) than a single-phase board, as you have phase-to-phase fault currents. Those are more expensive.

[madires]

I'm not claiming to know enough about Three-phase power to make a statement here, but how can you distinguish whether Dave has a star or delta unit? The neutral in a star configuration is optional, so the 3 bridge rectifiers could be wired in a delta as well as star configuration, can they not?

Not quite! In a star / Y configuration neutral is only optional (electrically, local sparky code might say something else) when all phases are loaded identically, e.g. by a 3 phase motor. The PSU has two power modules for the output and a third one for internal power. Hence it has to be in delta configuration without a neutral terminal. Also the rating of the input voltage suggests this.

[Towger]

I also think Dave may have 3 phase to the panel. But you are forgetting that the power supply is for puny yankee 3 phase, which is why it originally had the arse blown out of it.
Dave would still need to step down the voltage and in any case the Australian rules do not allow Dave to  work on the mains, as he does not have the required paperwork. ..

[Jay_Diddy_B]

Hi,

If this unit was on my bench I would figure out where the auxiliary supply is connected. It has to be connected between one pair of lines, L1-L2, L1-L3 or L2-L3.

The auxiliary supply is required to start the unit. It is probably the small line frequency transformer on the Input board. I would check it first. It could be damaged by being connected to 400V.

I would then run the supply at low power, less than 100W, using two phases. This assuming the power supply is not smart enough to detect a missing phase. If it is I would try and defeat that circuit.

I don't believe that schematics are available for this unit.

I have added some more photographs to this thread:

https://www.eevblog.com/forum/testgear/agilent-n5772a-600v-2-6a-1560w-power-supply-teardown/msg795977/#msg795977

Showing pictures of the boards in the single-phase, lower power, version N577x family.

Regards,

Jay_Diddy_B