Author Topic: Hybrid PSU  (Read 3932 times)

0 Members and 1 Guest are viewing this topic.

Offline raptor1956Topic starter

  • Frequent Contributor
  • **
  • Posts: 868
  • Country: us
Hybrid PSU
« on: July 31, 2016, 02:09:10 am »
It seems like pretty much all bench PSU's are linear and as such they waste a lot of power and therefore require more massive components like transformers.  One of the big problems is that even with multi-tap transformers there are voltage settings that are going to result in the linear supply dropping many volts to output as required.  If the tap voltage is, say, 18VDC and the desired output voltage is, say, 8VDC then it has to eat 10V at whatever the current is.

What if the PSU was instead a hybrid with a switcher front end that outputs a couple volts above setpoint to the linear section thereby reducing the power loss.  Yeah, the switcher will be noisy and that's not a good thing, but if that section were in a Faraday cage and decent filtration were done to the output before going into the linear section then the final output should be as clean as a full linear but do so much more efficiently.  Would also make for a reduced cooling requirement and a quieter fan system.


Brian
 

Offline ovnr

  • Frequent Contributor
  • **
  • Posts: 658
  • Country: no
  • Lurker
Re: Hybrid PSU
« Reply #1 on: July 31, 2016, 02:31:21 am »
... this is a thing, you know. Have a look at my TTi CPX400A teardown for an example.


And filtering switchers sufficiently well is a surprisingly hard problem.
 

Offline prasimix

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: Hybrid PSU
« Reply #2 on: July 31, 2016, 06:49:04 am »
Hybrid topology is present for quite some time. Starting from this forum you have Dave's uSupply that has step-up pre-regulator stage. Another frequently cited design is one that comes from LTC (DC2132A). Power efficiency could be further improved if you replace transformer with AC/DC adapter preferably with one that has PFC.
Take note that switching pre-regulation will result with increased ripple and noise on the power supply output regardless the linear post-regulator stage. Four layer PCB and additional filtering could improve situation and there is at least one design that claims that its hybrid topology output is not just comparable but even exceed pure linear design.


Offline raptor1956Topic starter

  • Frequent Contributor
  • **
  • Posts: 868
  • Country: us
Re: Hybrid PSU
« Reply #3 on: July 31, 2016, 07:14:27 pm »
Thanks prasamix, I had no doubt that hybrids had already been made I just haven't seen any of the bench PSU's made with that design. 

One way I can imagine to compensate for the noise the switching section makes would be to characterize the noise with testing then inject a 180 degree out of phase signal as is used in noise cancellation systems.  Another method might use a final output stage of the switcher that has dynamically adjustable LC components as the load/voltage changes.

I think it's just a matter of time before hybrids supplant linear's in most all applications.


Brian
 

Offline jpb

  • Super Contributor
  • ***
  • Posts: 1771
  • Country: gb
Re: Hybrid PSU
« Reply #4 on: July 31, 2016, 07:17:39 pm »
As the previous posters have said there are a lot of hybrid designs (others include the Hameg supplies as well as several TTi ones).

For an example see the R&S HMC8043 (Hameg):

https://www.elektormagazine.com/articles/RS-HMC8043-PSU/6745

My own preference is for pure linear because for a bench supply efficiency is not that important - or at least I put it lower priority than having a very low noise/ripple level and good transient response.

Size is perhaps a bigger negative on a crowded bench - certainly the hybrid designs are much smaller and lighter (though for a bench supply I don't think weight matters much either).
 

Offline raptor1956Topic starter

  • Frequent Contributor
  • **
  • Posts: 868
  • Country: us
Re: Hybrid PSU
« Reply #5 on: August 01, 2016, 06:24:34 am »
Yeah, for a bench PSU power efficiency isn't perhaps the most important thing but it's not unimportant either.  The Rohde & Shawarz PSU is a bit surprising both in the cost and relatively low power handling.  I can appreciate that a somewhat newer design is likely to cost more at first but given the fact that total power used is less and the ability to use transformers with fewer taps should ultimately reduce cost and bench space. 

I have a Rigol DP832 on the way which will be my first full function PSU -- amazing the work that goes into something that most people take for granted.  But, a good PSU like this is also a piece of test equipment given the ability to monitor and control both voltage and current.


Brian
 

Offline prasimix

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: Hybrid PSU
« Reply #6 on: August 01, 2016, 06:27:10 am »
Maybe use active filtering on switching node output as is it described in this paper.

You can check my project that include hybrid topology but also offers MCU managed "low noise" mode when switching node is simply by-passed by setting duty cycle to 100%. See results here. That can be used for supplying more sensitive circuits. The output power and current are limited but could be more then enough for testing e.g. audio and low power RF circuits. When not in "low noise" mode it can deliver continuously up to 160W or 5A per channel.

Anyway I learned that designing power supply is far more then simply select the right topology. There is bunch of details especially when you comes to programmability and usability (of course if you are not satisfied with simple power supply's "hello world" equivalent that just offer U and I displaying and programming).

If you have some circuit to propose for active filtering please let me know and at least I can try to add it in some of the future PCB revision for testing.

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: Hybrid PSU
« Reply #7 on: August 01, 2016, 09:04:31 am »
For low power versions the classical linear PSU is still OK and not that heavy. Here it is only the possibly lower price that might promote the hybrid versions, as they can save on the transformer and filter caps. However price driven units tend to have poor filtering and thus high noise. Even it the output is clean, noise injected to the grid could be a problem. There can also be more problem with common mode capacitive coupling - this is really hard to get rid of.

However for higher power (e.g. more than 200-500 W) the hybrid design has it's advantages. It gets a little crazy when a linear supply needs water cooling and gets larger than a washing machine. Good filtering is possible, it just adds a little to the weight and costs.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: Hybrid PSU
« Reply #8 on: August 01, 2016, 10:43:28 am »
There can also be more problem with common mode capacitive coupling - this is really hard to get rid of.

I'm not sure if that's fair, it's just not a common design requirement for switching power supplies so COTS transformers designed for it pretty much don't exist. Whereas shielded mains transformers are all over the place.
« Last Edit: August 01, 2016, 10:45:45 am by Marco »
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Hybrid PSU
« Reply #9 on: August 01, 2016, 11:02:50 am »
The principle of linear power supply with pre-regulator has been known for a long time.

There are different solutions for the pre-regulator:
- Phase control by thyristor or triac
- Taps switching made by relays or triacs
- few secondary transformer taps combined with phase control by triacs
- Switching power supply

The existence of a pre-regulator may however affect the dynamic behavior of the power supply in case of large output variations in voltage or current.
 

Offline prasimix

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: Hybrid PSU
« Reply #10 on: August 01, 2016, 11:25:56 am »
The existence of a pre-regulator may however affect the dynamic behavior of the power supply in case of large output variations in voltage or current.

This is true but question is how dynamic it is need to be. Here is the output from my PSU with SMPS pre-regulator. Output voltage is programmed from 0 to 20 V (half scale) with load over 1 A:


Offline Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: Hybrid PSU
« Reply #11 on: August 01, 2016, 11:28:46 am »
It would be hard to have a pre-regulator slower than the 10 msec you have to design around, ie. lots of capacitance, if you use rectified mains with low headroom. Nothing stops you from still having lots of capacitance at the headroom voltage to deal with load changes with the switching power supply.
« Last Edit: August 01, 2016, 11:34:05 am by Marco »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf