That's about what I usually use and had in mind.
The usual electrical stuff, Cat III multimeter and VDE insulated tools, even face shield and Nomex arc flash protective clothes (used for bush fire defence but electrical linesman's clothes were on special).
On the electronic side mostly older stuff, 20 MHz dual trace CRO with 10x divider probe. current probe etc.
10x Probe is 240 V rated but probably not Cat III.
That's not adequate, if I decide to build this myself I will need to invest in at least a better HV probe, HV insulation tester, probably more.
There's a DIY SMPS forum with a recommended equipment list that I need to check.
I have had a bit of a rethink about PFC input.
I was inclined to dismiss this because I don't really care much about a minor benefit to the electricity company.
But I now think the improved stability of the supply to the inverter is worth the effort, the actual cost will be small, I even have spare boost inductors.
Best wishes
The multimeter is fine if it's indeed 1000V, and wearing the face shield/ protective clothing shouldn't hurt either. However, a battery-operated differential-input scope and a 100x probe would be necessary for measuring the different SMPS voltages.
The current probe would be fine as long as it's non-contact (hall effect loop/clamp sensor) and has a decent bandwidth (say 100kHz), whereas the insulation tester may not be required altogether, if you correctly follow all the safety standards (post #48).
Since input-side devices (IGBT/MOSFET & driver) rated at 1200V are already available in the market, winding the HF transformer is likely to be your main difficulty while making the SMPS.
Further, it would be wiser to have a conventional single-layer through-hole (vs. multilayer) PCB, due to the magnitude of the voltages involved. The secondary low-voltage side routing may also need some extra copper thickness/solder coating to support the high output currents without heating the PCB traces.
The current probe would be fine as long as it's non-contact (hall effect loop/clamp sensor) and has a decent bandwidth (say 100kHz), whereas the insulation tester may not be required altogether, if you correctly follow all the safety standards (post #48).
Since input-side devices (IGBT/MOSFET & driver) rated at 1200V are already available in the market, winding the HF transformer is likely to be your main difficulty while making the SMPS.
Further, it would be wiser to have a conventional single-layer through-hole (vs. multilayer) PCB, due to the magnitude of the voltages involved. The secondary low-voltage side routing may also need some extra copper thickness/solder coating to support the high output currents without heating the PCB traces.
only cheap source for 3 phase transformer EI core I know is 24V for constructions sites (safety 24V lighting) you can find sometimes on ebay.
One nice test you could do to evaluate the effectiveness of 3 Phase PSU would be to buy 3 same class AB transformer powered amplifiers, connect the +V -V and GND of the 3 amps together with heavy wire. plug each amp on one different phase. You should get the 3 amps powered by a 3 phase PSU.
I hope to try this with my 3X JBL UREI 6290 amps.... will report here
One nice test you could do to evaluate the effectiveness of 3 Phase PSU would be to buy 3 same class AB transformer powered amplifiers, connect the +V -V and GND of the 3 amps together with heavy wire. plug each amp on one different phase. You should get the 3 amps powered by a 3 phase PSU.
I hope to try this with my 3X JBL UREI 6290 amps.... will report here
Thanks for the information but I do not plan to use a 3 phase transformer.
As I stated in an earlier post, if I use line frequency transformer(s) then I see little benefit in a 3 phase transformer, 3 individual 240 V toroids connected in Y is easy to rewire for 1 phase if I move house, and probably about the same cost.
Yes indeed
I am fairly competent with LTspice, I wouldn't even think about a project like this without careful simulation.I checked the DIY SMPS site and their recommendations were no surprise, 100x probe, differential probe..
Best wishes
David
With a ~650 V worst case input that would require 800 or 900 V rated FETs.
At that level Silicon Carbide (SiC) starts to look attractive, despite the more complicated drive requirements.
You (or anyone else) have experience with these?
I have had a think about the system architecture.
If there is a PFC boost input then the LLC runs at almost constant input so there is little need for the controller to vary the frequency.
That makes the LLC basically a filter to convert the square wave switches to a sine wave to drive the hi frequency transformer.
This is usually analysed in the Fourier domain with First Harmonic Approximation but my intuition is to see it in the time domain, that we hit a resonant circuit with a pulse at the appropriate times.
Not sure if that is useful, just find it helpful to write it down while I think about it some more.
Best wishes
David
Dave Zan said:
I haven't used any SiCs myself, but at such bus voltages, even regular Si devices require a robust and preferably bipolar gate driver (for obvious reasons). An isolated gate-drive with under-voltage and de-saturation protection is invariably used in such applications.
If I understand correctly, the significant differences between Si and SiC are within the devices themselves (and body diodes as well), while the gate-drive requirements appear to be largely similar.
Dave Zan said:
Well, that's about line regulation alone, but there's load regulation to be considered as well. Amplifiers could draw large amounts of charge from the output capacitors (e.g. during a helicopter or dinosaur scene) that then needs to be replenished by the LLC (by reducing its switching frequency), in order to minimise output voltage error. Yes, it's mostly a filter but still an adjustable one.
SiC seems to need up to 20 V or so, some of the common driver ICs don't look suitable.
Also there is some discussion of reverse bias but it seems to be mainly to prevent turn on from Dv/Dt so not a problem for an LLC with ZVS, at least as far as I know.
Yes, there is load variation but I suspect it's not such a problem.
Many amps, after all, work just fine on a simple mains transformer, diode and capacitor supply with no load feedback at all.
Since I don't yet understand the LLC well I even started to wonder if it would make sense to take the load feedback to the PFC front end.
That made me look more closely at various alternate PFC front ends.
There were so many other possibilities than the obvious rectifier and boost section.
The Cuk version I posted earlier still looks worth more study too.
Luckily, even the simplest solution will still work well because the rectifier diode drops are a proportionally smaller fraction of a 415 V input that a 120 V US style supply.
But some kind of active rectifier could do PFC, soft start and deal with line variation too.
So I now I think I understand the Vienna rectifier better, I don't want to "reinvent the wheel" but I have to work stuff out for myself before I really understand it.
Best wishes
It's easy to half wave rectify each phase relative to neutral and connect in parallel.
But if you add a -ve half to full wave rectify then there is + and - 330 V for a total of ~660 V.
So either I misunderstand your proposal or one of us is mistaken, could you show a circuit schematic?
Best wishes
David
If you use 3 full wave bridges.
Each bridge input ~ ~ is connected to one phase and the other input to the neutral. Then you get 330V at the output of each bridge.
You can then connect the outputs of the bridge in parallel 3 + together and 3 - together and still get 330v nearly DC because of the 3 phase shift.
I don't have drawing tools at my hand to draw a schema right now.
Each bridge input ~ ~ is connected to one phase and the other input to the neutral. Then you get 330V at the output of each bridge.
You can then connect the outputs of the bridge in parallel 3 + together and 3 - together and still get 330v nearly DC because of the 3 phase shift.
I don't have drawing tools at my hand to draw a schema right now.
Thank you for the picture, but there's a problem (not yours!).
The proposed SMPS is driven directly off the mains line.
There are no input transformers and this alters the situation.
I suspect he has in mind a circuit just as you have drawn but hasn't considered what happens when it's driven by sources that are not isolated.
I could be mistaken of course.
Best wishes
David
Many gate-driver ICs like IR2213 are capable of 20V, with an absolute maximum of 25V. However, at such bus voltages, even regular Si devices benefit from bipolar gate-drive, already supported by popular ICs like M57962L etc. The following is a link to some popular devices from IR/Infineon. IXYS, Farichild, TI etc. also make similar products.
https://www.infineon.com/dgdl/Infineon-1200V_Level_Shift_Gate_Driver_Product_Portfolio-ProductBrief-v01_00-EN.pdf?fileId=5546d46266a498f50166b0a0438100c9
The OP appears to be looking for transformerless rectification, please see post #78.
Last edited:
Yes, I had in mind some of the LLC controller ICs with included drivers that don't have as robust drive capability as a dedicated driver IC.
The best implementation is still to be decided - whether an all in one controller can do it simply or whether I need driver ICs.
And there are different possibilities of Si or SiC FETs.
Similarly the PFC implementation is still unclear.
Lots to keep me occupied in the Covid lock-down.
Best wishes
David
Last edited:
