LT app note: "High Performance Portable DC Bench Power Supply"

[ivan747]

I have been interested in both amspire's "General purpose power supply design" and the series of videos Dave has made on power supplies.

If you don't know what I am talking about, it is time for you to read some of the most prolific threads and videos about the topic in this website:
Dave's supply:
http://www.eevblog.com/projects/usupply/
Amspire's supply:
https://www.eevblog.com/forum/projects/general-purpose-power-supply-design-7488/


The common thing here is that both power supplies are designed use what's called a tracking pre-regulator, wich is a DC-DC converter that gives a certain voltage above the voltage output of a linear reulator that follows it. So that when the linear regulator is 5.5V, the pre-regulator stays, for example, 2V above that. This gives lots of advantages including reduced heat, smaller heatsinks, decreased weight and volume and naturally, more efficiency.

Another common aspect of both power supplies is that they are designed to be fed from a wide variety of unregulated sources, including batteries and AC power (Dave's PSU can also be fed from USB if memory serves me well). In principle, Amspire's can be fed from a pre-regulator, DC source, rectified and filtered AC and batteries. If you use a buck-boost DC-DC converter both solutions could be powered from many kinds of battery configurations.

 Amspire's solution is the most generic one, it can be built from some of the most common parts available in the electronics market with the exception of a set of 4 well matched resistor, witch could in principle be trimmed or matched manually from common tolerances. It can be made completely using though hole technology, unlike Dave's and the new design I am going to present.

Dave's solution is compact, makes integral use of a microcontroller, although the circuit can live without it. It needs two voltages to set the output voltage end current, so does amspire's.

It's excellent the work both of these men have made.

So it seems that someone at Linear Technology, Mr. Keith Szolusha realized this is a good idea, so he's presenting his take on the design using some very neat LT parts, and he made an app note about a similar sort of power supply. It shares some of the characteristics with the previously mentioned power supplies. He documented the work brilliantly as well. The size of the board is quite impressive and it competes with Dave's solution in terms of size.



It does have the disadvantage that it was designed to work with potentiometers only, and they should be of high quality according to the author. If an interface with a microcontroller is needed, one can always use a digital pot. Only the voltage output potentiometer is ground referenced, so it could pose a challenge if one want to digitally control the current. The current setting is non-linear in this design because the potentiometer (acting as variable resistor) is in parallel with another resistor, so that has to be compensated in software or corrected in hardware if one wants to make it digital.

A bill of materials is missing from the app note, so the cost of the solution remains to be seen. LT parts have a price above the average and in this design practically all integrated circuits are LT parts, but the performance of LT parts is often hard to match. I do not have intentions of mass manufacturing this, at most, an OSHW board could be laid out from the schematic, as well as the mechanical design needed to fit it in a reproducible enclosure. That means cost is not an important factor to me.

Here is the article:

http://cds.linear.com/docs/en/lt-journal/LTJournal-V24N2-02-df-BenchSupply-Szolusha.pdf
http://www.linear.com/solutions/5086
In case these links break in the future, 10 years from now, the code for the article is DC2132A

I hope you enjoy the app note and share your implementations when you get to build them 


P.S. it's available as a demo board for $US200. That price is probably inflated and does not reflect the cost of the BOM, I have to confirm that.
http://www.linear.com/purchase/DC2132A
Obviously, the better route here is that we make an OSHW circuit board from the reference design.

[Bored@Work]

So it seems that someone at Linear Technology, Mr. Keith Szolusha realized this is a good idea,

Linear does the pre-regulator thing since decades. The famous Jim Williams published a few circuits in Linear application notes, e.g. AN-32 in 1989. There should be a few more.

Dave even took his preregulator out of an LT datasheet and made a video of where he played with and measured that preregulator and regulator.

[ivan747]

I have seen Jim Williams pre-regulator app notes, he made them using SCRs and op-amps. Well, you're right, I could have been implying too much.
But Dave's design shares does some similarities. The LT app note uses almost the same LDO Dave uses, but in a different way. Both are flexible and can be powered from many different voltage sources.

One thing that I find with this is that it is a bit noisy, more than I wish it was. Look at the waveform in the attachement.


Jim William's pre-regulator is practically noise-less.

[David Hess]

Linear does the pre-regulator thing since decades. The famous Jim Williams published a few circuits in Linear application notes, e.g. AN-32 in 1989. There should be a few more.

Linear Technology AN-2 from 1984 has a few and I am sure there are older examples in other application notes.  The idea behind using a switching preregulator goes back a long time in the form of SCR based preregulators but even transistor based high frequency switching preregulators were used in designs predating 1989 and 1984 including the Tektronix 22xx series oscilloscopes.

The noise would be considerably improved by using a lower noise switching preregulator and/or adding differential and common mode filtering between the switching preregulator and the linear regulator.

[chickenHeadKnob]

BravoV started a thread about this demo board back in september: https://www.eevblog.com/forum/projects/linear-technology-dc2132a-cvcc-adj-bench-power-supply-board/

member DutchGert bought one (they are expen$hiv) and gives us a review.

[ivan747]

Thank you very much for the heads up!

[szolusha]

Hi folks,

The $200 price tag represents the cost associated with getting a prototype sample of this demonstration circuit.
We developed this circuit to show off how LT3081 can be used in combo with LT8612 to make a nice bench top supply.
We do not intend to mass-manufacture them. If somebody wanted to mass-manufacture with this bill of materials, that person can copy the bill of materials and layout and produce the boards for $35-$50 probably.

Please let me know if you have any more questions.

K Szolusha

[dom0]

Linear does the pre-regulator thing since decades. The famous Jim Williams published a few circuits in Linear application notes, e.g. AN-32 in 1989. There should be a few more.

Linear Technology AN-2 from 1984 has a few and I am sure there are older examples in other application notes.  The idea behind using a switching preregulator goes back a long time in the form of SCR based preregulators but even transistor based high frequency switching preregulators were used in designs predating 1989 and 1984 including the Tektronix 22xx series oscilloscopes.

The noise would be considerably improved by using a lower noise switching preregulator and/or adding differential and common mode filtering between the switching preregulator and the linear regulator.

Tek used switching regulators since the 60s. They are, I think, almost exclusively resonance converters, which have inherently lower noise and HF content than other types of SMPS. I think the first was the 547 which AFAIK used a resonant Royer converter to generate the HV.