Can I series buck converters to get higher stepdown ratio?

[sodoyle]

TL;DR Summary

I would like to know if I can connect buck converters in series. I am looking at the LT1170HV specifically

I have two capacitors in series across my bus. I have a some series resistors across the capacitors for voltage balancing. I would like to power some low voltage low voltage components tapping off of the voltage divider using that as the voltage supply.

The bus voltage will be approximately 2 kV. The DC/DC converter I'm planning to use is the LT1170HV which has a maximum input voltage of 60 V. I would like to operate this near 20 V and need a load current of approximately 20 mA at 5 V. The application note for this chip recommended a minimum load of 100 mA so I guess I'll have to add a dummy load on the low voltage side to draw more current.

To find the input current to the converter, I used D=Vo/Vin and Iin=D*Io. With Vin=20 and Vo=5, I get D=0.25 and Iin= 25 mA.

Since this converter will be in parallel with a resistor from the voltage divider, I'll assume the input input impedance of the converter is equal to the resistance of the parallel resistor used as its source. In that case, the current through the voltage divider must be 2*Iin or 50 mA.

To get current of 50 mA through the voltage divider, Req=2000/(.05)=40 kOhm

Using Zin = Vin/Iin, and Iin= 25 mA, I find Zin=800 Ohm.

The voltage divider would then look like some number of resistors totaling approximately 39.2 kOhm with a 1.6 kOhm resistor at the bottom. With the converter connected and drawing power, the equivalent resistance of the would then be approximately 40 kOhm (in the ideal world) i think right?

I'm sure there are a lot of problems with this approach so please berate this approach and help me find a better solution. The first problem that I see is the resistors would be dissipating roughly 100 W. Actually, that piece alone stops me from going further with this. Is my math and approach correct from all of the steps above?

I was thinking maybe I could use several converters in series so I could tap off a higher voltage requiring a lower input current. I would have to series several bucks to get down to 5 V because the converters that do like 700/5 V conversion are WAY outside of my price range.

Any recommendations are greatly appreciated.

 

[berkeman]

One problem with series-connecting buck DC-DC converters is stability. Switching power supplies often do not react well to load currents that are varying near their switching frequency. You might be able to synchronize the stages to get around this, but it will still take some analysis and simulation and experimentation.

There might be a better way to do this -- have you looked at switched capacitor converters, for example?

 

[DaveE]

I would ditch the LT1170HV and do a more discrete design. If you use a 1KV MOSFET you can easily operate at 400V instead of 20V at the input which will drop the power in the series resistor by a factor of 20x or so. You will not get the requisite ~1% duty cycle from most control ICs, but this isn't such a difficult thing to solve. Since the load current/power is small you could design a crude SMPS, perhaps with a hysteretic (bang-bang), or fixed on-time controller, and then follow it with a linear regulator.

This is the sort of design that will benefit from some system engineering up front, before you start choosing parts. I think there are a few different approaches possible, but none will be "off the shelf" applications.

Pay extra attention to protection due to transients, boundary cases, abnormal operation, start-up/shut-down, etc. Operating a 60V, or 1KV circuit from a 2KV source is an easy way to accidentally destroy things.

 

[sodoyle]

There might be a better way to do this -- have you looked at switched capacitor converters, for example?

I actually hadn't thought about a switched capacitor converter. I just looked at a few and I think they look a lot better since they're suitable for low power with only a few mA. I'll look into these more and hopefully have more luck.

 

[sodoyle]

You will not get the requisite ~1% duty cycle from most control ICs, but this isn't such a difficult thing to solve. Since the load current/power is small you could design a crude SMPS, perhaps with a hysteretic (bang-bang), or fixed on-time controller, and then follow it with a linear regulator.

I was hoping to stay away from having to make my own design partially because I'm not too sure how to tackle control of and driving the MOSFET. I searched buck converter controllers but I'm only seeing results come back for the buck IC with the FET already integrated. Am I able to get just a small chip for the controller and gate driver separate?

 

[Baluncore]

I have two capacitors in series across my bus. I have a some series resistors across the capacitors for voltage balancing. I would like to power some low voltage low voltage components tapping off of the voltage divider using that as the voltage supply.

You have 2 kV, in two series capacitors. That is 1 kV each.
Change that to 8 capacitors in series, to get 250 volts across each.
Quarter voltage, four times the capacitance = same physical volume.

Get a miniature 250V AC input switching converter off-the-shelf, to produce 5 V @ 20 mA = 100 mW. Internally the AC will be rectified to charge a capacitor to 350 V. Feed it instead with 250 VDC from across the lowest capacitor, closest to the 5 V supply ground.

Protect the switcher and lower capacitor with a 300 V chain of zener diodes.
Use voltage balancing resistors across the remaining caps in the chain.
Assume 250 mW maximum switcher input. 0.25 W / 250 V = 1 mA.
250 V / 1 mA = 250 k maximum balancing resistance per capacitor.
The voltage balancing chain now dissipates 2 watt total.

Normal resistors are only rated 100 V. Balancing resistors must be rated to 250 V or built up from series chains.

 

[DaveE]

I actually hadn't thought about a switched capacitor converter. I just looked at a few and I think they look a lot better since they're suitable for low power with only a few mA. I'll look into these more and hopefully have more luck.

Yes, the switched capacitor idea sounds good. I haven't thought about it in this context though, I'm afraid you may need a >2KV switch for it. Unless of course you keep the dropping resistors. If you look for this, they are also called "charge pumps".

I was hoping to stay away from having to make my own design partially because I'm not too sure how to tackle control of and driving the MOSFET. I searched buck converter controllers but I'm only seeing results come back for the buck IC with the FET already integrated. Am I able to get just a small chip for the controller and gate driver separate?

There are many, many, many SMPS control ICs available. You can also make your own control circuit IF YOU HAVE TO. Once you've done the system design, and know what you want, the controller isn't the hard part.

The issue with SMPS controllers and gate drive ICs is that there is a huge number of options. If you haven't found too many choices, you aren't searching hard enough. Go to the TI and Analog Devices web pages to start, they have a bunch of parts and good design support. The Digikey site lists >12,000 DC-DC control ICs with about 3,500 active and in stock. They list almost 7,000 gate drive ICs with over 2,000 active and in stock.

This isn't an easy design, you will have to do some work to figure it out. Advice from social media can help, but it's not the answer.

 

[sodoyle]

Get a miniature 250V AC input switching converter off-the-shelf, to produce 5 V @ 20 mA = 100 mW. Internally the AC will be rectified to charge a capacitor to 350 V. Feed it instead with 250 VDC from across the lowest capacitor, closest to the 5 V supply ground.

Protect the switcher and lower capacitor with a 300 V chain of zener diodes.
Use voltage balancing resistors across the remaining caps in the chain.
Assume 250 mW maximum switcher input. 0.25 W / 250 V = 1 mA.
250 V / 1 mA = 250 k maximum balancing resistance per capacitor.
The voltage balancing chain now dissipates 2 watt total.

Normal resistors are only rated 100 V. Balancing resistors must be rated to 250 V or built up from series chains.

This may be "simple" but sounds pretty complicated for me at the moment. I'll research this so I have my options for future designs. Thanks for the suggestions.

 

[sodoyle]

Yes, the switched capacitor idea sounds good. I haven't thought about it in this context though, I'm afraid you may need a >2KV switch for it. Unless of course you keep the dropping resistors. If you look for this, they are also called "charge pumps".

Yeah I would still have to use the resistors to drop the voltage. Ideally, I will be able to connect the converter to the balancing resistors to reduce component count. That will all depend on how how well I can maintain voltage balancing on the caps while pulling current from one of the resistors though.

If you haven't found too many choices, you aren't searching hard enough.

I guess I'm not really sure what exactly i should be looking for then. I'll work out the other design details, then move to this part.

This isn't an easy design, you will have to do some work to figure it out. Advice from social media can help, but it's not the answer.

Thanks for your input. I'm definitely not expecting to get all of the solutions to my problem from here. It has been helpful just getting everyone's input though. Heck I've made switched capacitor circuits (open loop) before and that didn't even cross my mind! So Just the discussion has been helpful. Thanks again.

 

[Baluncore]

This may be "simple" but sounds pretty complicated for me at the moment. I'll research this so I have my options for future designs.

 

[DaveE]

Get a miniature 250V AC input switching converter off-the-shelf, to produce 5 V @ 20 mA = 100 mW. Internally the AC will be rectified to charge a capacitor to 350 V. Feed it instead with 250 VDC from across the lowest capacitor, closest to the 5 V supply ground.

This will work, but derate the AC-DC PS some. There are occasionally some differences in running the input circuits on DC. For example input rectifiers don't heat evenly.

It's probably irrelevant in this case, but this will often violate the safety approvals for an AC-DC PS. Because of component stresses, or, really, they just didn't write it up in the paperwork or do the testing for DC. Also connecting to a 2KV source will violate the safety approvals unless you demonstrate fail safe circuitry, probably with double or reinforced insulation, to reduce the voltage.