I have seen a very simple circuit to generate three phases from one:
Line - Capacitor1 - Capacitor 2
Phase 1 - Phase 2 - Phase 3
1][2][3
Phase 1 feeds one cap, which shifts it 120 degrees, that is the second phase tapping at the joint between the capacitors, then the second cap also shifts it another 120 degrees, that gives the third phase at the far terminal of the second cap...Simple.
Enough for small intermittent loads...
Line - Capacitor1 - Capacitor 2
Phase 1 - Phase 2 - Phase 3
1][2][3
Phase 1 feeds one cap, which shifts it 120 degrees, that is the second phase tapping at the joint between the capacitors, then the second cap also shifts it another 120 degrees, that gives the third phase at the far terminal of the second cap...Simple.
Enough for small intermittent loads...
It seems you have not made the effort to read the thread, based on your posts about issues already dealt with.
For example
When I have already stated the load is ~1800 W to 2300 W.
Or your post of a CEEform connector when that has already been dismissed.
Usual car alternators are ~15 V, maybe ~100 V at maximum revs if the load is disconnected.
Do you think this will be safe at ~600 V mains?
But my main concern is that you have missed my point.
I know the usual reason to use 3 phase is for more power but that is not my objective here.
My proposal is only because it may enable a power supply that is simultaneously more reliable, less expensive and of better performance.
Comments that I don't need 3 phase for such a small power supply miss the point.
There is almost no extra effort involved, for the baseline cases just a 3 phase rectifier if I use SMPS, or if I use transformers then three smaller 600 VA units rather than a 1800 VA one.
Best wishes
David
Last edited:
Just about any transformer fellow would make one if you pay him the money. It's just business and they're not bothered about what you do with the transformer.
The practical aspects:
Peak rectifier voltage (nominal) = 230 *√2*√3 = 563V (or 588V for 240V).
Peak voltage during low/high line (200-260V) = 490 to 636V.
Diode voltage rating with safety/reliability margin => 800V or 1000V (better).
Diode RMS current = DC current /√3 = 0.577*Idc.
Diode Avg current = DC current/3 = 0.333*Idc.
Huge margins may not be required in case of the diode current, as it's unlikely that the amplifiers would be operating at full power continuously.
Last edited:
My car alternator had Motorola diodes with 1000 V ratings inn the diode plate, basically diodes push fitted on a heat sink / fixture made of aluminum.
And it was a Lucas 43 A rated unit, you can work out the individual diode ratings.
And most TV horizontal and SMPS transistors too are capable of 900V to 1000V.
Another thing, here it is common to have variation of up to 20 volts in the phase voltages, say I will get 243-260-263 on different phases (to neutral).
What that does in terms of balance is for you to consider.
And it was a Lucas 43 A rated unit, you can work out the individual diode ratings.
And most TV horizontal and SMPS transistors too are capable of 900V to 1000V.
Another thing, here it is common to have variation of up to 20 volts in the phase voltages, say I will get 243-260-263 on different phases (to neutral).
What that does in terms of balance is for you to consider.
Last edited:
Please note that in addition to the rectifier core, one would also need a mechanism to limit the inrush current due to the filter capacitor for proper operation. In the industry, this is usually achieved by way of a series resistor (bypassed by a relay after cap reaches steady-state) but a similar arrangement using an NTC thermistor (and no relay) maybe sufficient for capacities upto a few kW.
The grid imbalance is usually within limits and doesn't cause any serious performance issues unless phase loss occurs. Thus, some mechanism that inhibits the SMPSes (using enable/disable logic) during a phase-loss condition may also be needed.
That is more than usual for car alternator, the numbers I have seen were 400 to 600 V.
400 V would probably fail immediately, 600 V may be worse in some ways, may last a while until you think it's safe then fail unexpectedly.
1000 V should be OK except for...
Lucas are famous in Australia for terrible, unreliable electrics on British cars and motorcycles, part of the reason for the wreck of the British automotive industry. I have never seen anywhere near as bad as that on my power.
Maybe it happens near a factory with extreme loads but the Australian system performance is usually excellent.
It is worth consideration but I don't see it will be a problem.
Similarly, I have never seen a phase loss, but will check to ensure there is no problem if it ever does occur.
The elimination (or reduction) of this problem is actually one of my reasons to propose 3 phase.
There is always at least one diode pair in conduction so I don't need any "bulk" capacitance, just some smaller filter capacitance.
So the inrush will be much smaller, perhaps I will not need any inrush limitation other than the series resistance and inductance of the wires and diodes.
Will need to check this, maybe some added inductance will help and can be used for filtration too.
I have started to look more seriously at LLC style SMPS.
The simplest solution is to build just one SMPS to run directly on 600 V.
"Simplest" in the sense that it would involve the least number of parts.
Maybe the extra precautions required to work at 600 V spoil the simplicity.
What is your opinion?
Best wishes
David
Last edited:
Lucas TVS, India.
Better than Bosch here, poorer than Denso in performance, but very reliable, with a good service and parts network.
Anyway, Motorola diodes, not their own product.
British and even American quality in car electrics and mechanically seems to be less than Japanese and German, but that is not the discussion topic here...
You intend to use a 3 phase rectifier to get cleaner DC?
Use a NTC or something for inrush current if you feel it is needed.
Enjoy the effort, and tell us how the result was.
Better than Bosch here, poorer than Denso in performance, but very reliable, with a good service and parts network.
Anyway, Motorola diodes, not their own product.
British and even American quality in car electrics and mechanically seems to be less than Japanese and German, but that is not the discussion topic here...
You intend to use a 3 phase rectifier to get cleaner DC?
Use a NTC or something for inrush current if you feel it is needed.
Enjoy the effort, and tell us how the result was.
If the equipment has e-caps, or if increased ripple could damage parts, or load operation could cause damage when a phase loss is not detected then yes the standard industry response is to use a phase-fail relay, and they can come with low and high voltage detection if operation outside of voltage limits also may cause collateral damage. You could also use a frequency selective filter to detect a loss of phase - a few ways to skin that cat.
Whether you really need some in-rush limiting mechanism comes down to how high the in-rush current gets to and whether that level could trip upstream protection or have some other debilitating or flicker consequence - otherwise its just an aesthetic nicety imho.
Well, the more balanced the grid, the better your output would be, but it's not an issue in my opinion. Also, a phase-loss is a rare event (like a solar eclipse) that you might not see unless you work at a sub-station etc. It is usually a result of natural disasters (hurricane/rain/flood), accidents (trees falling etc.) and sometimes human folly.
Dave Zan said:
Even if you choose to not use a capacitor at the rectifier output, the step-down SMPS that follows would still have one at its input, irrespective of converter topology.
A blown rectifier would result in phase-to-phase short circuit faults that could turn into fire hazards, in case the upstream fuses/breakers fail to act in a timely manner. Therefore, it's always better to be safe than sorry, especially since it's a domestic application.
Dave Zan said:
Yes, LLC is a very good topology for step-down application, with sinusoidal transformer current (low EMI) and inherent ZVS (primary) and ZCS (secondary) capabilities. For fewer parts, you'd need a HF transformer that integrates the first 'L' as the primary leakage inductance and the second 'L' as the magnetisation inductance, with only the 'C' (resonant cap) outside. This is usually already done in most commercial units, as it reduces costs and eases assembly.
Not if you get a ready-made item which would be a box that requires some nicely insulated wiring. Most designs already incorporate the required spacing/clearances/creepages etc., according to operating voltage. However, at 600V, the only difference is that it becomes necessary to ensure that the safety standards (CE/EN/IEC) have been followed by the manufacturer. Equivalently, you may need to stay away from an Ebay or Alibaba SMPS when running above 300-400V.Dave Zan said:
Alternatively, you could build one if you have the knowledge, skills and the necessary testing / measurement equipment. However, it is worth noting that a regular oscilloscope rated as a category I (300V max.) device cannot be used to make measurements at 600V or thereabouts.
That must be a really "electric" car.NareshBrd said:
Last edited:
I will be interested to see that.
Ok, these are the ones I want, hence my comment that only an AVR seems to have all the licenses.
But best of luck with your project.
In an AVR it's usually done after the DAC, I think, so there is no decoder/processor option if I use an AVR as the source.
That means I would need to do it in the DAC or after.
DAC seems nicer but I could always synchronise 16 motorised potentiometers
Best wishes
David
Dave Zan said:
Cost-effective decoding of HD formats is also possible using a graphics card (for your computer) or an open-source decoder based on a single-board computer (like Raspberry Pi) with HDD/BD interface.
Dave Zan said:
Well, if you're looking at an active multi-way system, then most AVRs are already off the list. The next sensible approach in my opinion would be Bluray/Computer=>HDMI-to-I2S=>processor=> delay / DAC=>amplifier etc. Now, if the Bluray/computer already has internal decoding, then a 7ch PCM output on the HDMI would make things much easier.
Nevertheless, if you have the money, there are complete professional decoder-processor units like CP850 (Dolby) and AP20 (DTS). Both are 16-ch multi-way with the key difference being that the DTS product has a proper Dolby-licenced decoder whereas the opposite is not true, in spite of DTS' popularity.
Dave Zan said:
"After the DAC" seems to be the correct way to do it, as the step-size is also reduced along with the signal (no bit-loss). However, it may not be very practicable in case of multi-channel systems.
I didn't know they existed, I expected it was all proprietary.
Any links?
Why? the plan is to use the AVR as decoder then hack the internal I2S streams into AES/EBU output drivers.
Once I have AES output then can drive the active multi-way system.
I believe license restrictions prohibit this sort of approach too with limitations on decrypted PCM and HDMI.
Any idea what these cost?
Yes, after the DAC is theoretically better but less convenient.
In practice I understand that the best DACs do volume control very well.
Now back to 3 phase power supplies.
It does look like a 3 phase rectified to LLC SMPS is practical.
It would nominally run at <600 V, even worse case is <650 V so silicon FETs should not be a problem nor the rectifier.
I don't know much about SiC, 650 V is about when they start to be the preferred choice, from what I've read.
Best wishes
David
Dave Zan said:
FFmpeg - Wikipedia
Dave Zan said:
Well, that way you still need to carry out the crossover filtering, room/speaker EQ and time-alignment using separate processing unit(s).
Dave Zan said:
In that case, a possible solution for direct I2S is to hack a HD sound card for the decoding. They're usually less expensive when compared to AVRs.
Dave Zan said:
Maybe your car, but definitely not your house. However, there are less expensive multichannel bi-amp and triamp consumer products that decode Dolby Atmos etc. For example:
LS10 | Datasat Digital
Dave Zan said:
An 800V-1200V device would be more reliable, as there could be spikes and ringing (>650V) across the devices due to the energy stored in the parasitic inductance(s) of the devices, PCB and wiring. You may look at the Infineon CoolMOS series for suitable (and affordable) devices. Input current (avg) would be <5A in any case.
Yes, more modular and flexible but not a one box solution.
Thanks for the link, that's a whole new solution architecture I didn't know about.
Depends on one's car.
Yes, I am not sure that 650 V devices have sufficient safety allowance.
Unfortunately, devices over 650 V are less of a commodity item and the cost/performance moves up a level.
I happened to check my broken Solar inverter yesterday and found Infineon CoolMOS devices.
They are rated as 600 V FETs and the inverter as a unit is also rated as 600 V max.
But it's solar panel DC input with no mains spikes/transients, so they can run up to the device limits.
Even so, there are spike filter inductors and MOVs on the solar panel inputs, just in case.
Best wishes
David
Dave Zan said:
For a one-box, I think it would be easier to add decoding to a processor (or find one with decoding) than take the pains to add processing to a decoder (lot of work).
Dave Zan said:
If you would like to try 600V, there's this very common and affordable CoolMOS called SPW47N60C3 (a.k.a. 47N60, affectionately) rated at 600V, that (but) doesn't avalanche until 650-700V.
Yes, that's what in the inverter.
So I hope to find a 650 V rated FET that doesn't avalanche until 700-750 V.
I don't know how much safety allowance is sensible.
I plan make the switch on the power supply a 3 pole circuit breaker, for over-current protection.
Maybe an over-volt protection trip on the breaker would improve safety.
Best wishes
David
Last edited: