Author Topic: Show your favorite and most used benchtop PSU  (Read 230382 times)

0 Members and 2 Guests are viewing this topic.

Offline jaycee

  • Regular Contributor
  • *
  • Posts: 196
  • Country: gb
Re: Show your favorite and most used benchtop PSU
« Reply #475 on: May 13, 2020, 07:10:30 pm »
Not the first PSU I've built, but the first one that was more than an LM317, and built into a proper box :)

It's a fairly common topology.. A TL082 as sense amps, and another as error amps. Voltage reference is a TL431. Transformer is custom wound but an off the shelf 2x9V (plus another small 2x9V for the opamp supplies) would work. The metering is done using an Atmel ATMega324, which also controls switching the load on and off via an illuminated pushbutton

I made a few little mistakes, the metering is not very accurate at the low range because the ADC in the ATMega is not very accurate near ground. I should have added some bias to the sense amps so that the voltages they output doesn't go near ground - this doesnt bother the opamps because they have a -ve rail

edit: Oh yeah, the important bits. It's 0-18V at 0-1A, and 90% of the components are recycled parts - including the ATMega324!
« Last Edit: May 13, 2020, 07:13:54 pm by jaycee »
 
The following users thanked this post: nuno, JustSquareEnough

Offline nuno

  • Frequent Contributor
  • **
  • Posts: 585
  • Country: pt
Re: Show your favorite and most used benchtop PSU
« Reply #476 on: May 15, 2020, 09:10:51 am »
Nice build.

(...) metering is not very accurate at the low range because the ADC in the ATMega is not very accurate near ground.

Can you elaborate?
 

Offline jaycee

  • Regular Contributor
  • *
  • Posts: 196
  • Country: gb
Re: Show your favorite and most used benchtop PSU
« Reply #477 on: May 15, 2020, 06:23:13 pm »
Nice build.

(...) metering is not very accurate at the low range because the ADC in the ATMega is not very accurate near ground.

Can you elaborate?

The ADC in the ATMega, like a lot of ADCs, are not very accurate at the extreme ends of their range. Add in compensation for offset and it becomes an issue. The ammeter on my psu cannot read any lower than 20mA for example, yet it can current limit lower than that so the problem is not in the sense amp. I could have avoided the issue by adding an offset to the sense amps so that they do not go near ground, which would have allowed enough range to both cancel the offset and avoid the nonlinearity.

I attempted to hack it a bit by adding resistors to adjust the range to the front panel display, its not perfect but workable.

I might try a further hack to add the offset I mention. A TL431 should be enough to do the job. At any rate, the power supply works well and it's better than the previous lashups of LM317's etc :)
I'd "Open source" it, but I really dont want some unscrupulous Chinese ebay sellers turning it into a kit.
 

Offline nuno

  • Frequent Contributor
  • **
  • Posts: 585
  • Country: pt
Re: Show your favorite and most used benchtop PSU
« Reply #478 on: May 15, 2020, 10:57:11 pm »
Are you using a current measuring resistor? If so, what value and what amplification (if != 1) before feeding the AVR's ADC?
 

Offline jaycee

  • Regular Contributor
  • *
  • Posts: 196
  • Country: gb
Re: Show your favorite and most used benchtop PSU
« Reply #479 on: May 15, 2020, 11:33:31 pm »
Are you using a current measuring resistor? If so, what value and what amplification (if != 1) before feeding the AVR's ADC?

Yes, 0.1 ohms. There is a sense amp and the output feeds both the ADC and the current control error amp. The sense amp is not the problem.

Full schematic here: http://www.dark-matter.me.uk/files/electronics/psu-18v1a/main_sch.pdf
edit: Actually thats just the main board. The display board is at http://www.dark-matter.me.uk/files/electronics/psu-18v1a/front_sch.pdf

edit 2: My plan is to disconnect AGND from the star ground, and add some offset voltage there. The sense and error amps get shifted away from 0v, but they dont care about that.. the only thing that will see the difference is the ADC which is referenced to DGND
« Last Edit: May 16, 2020, 01:03:55 am by jaycee »
 

Offline nuno

  • Frequent Contributor
  • **
  • Posts: 585
  • Country: pt
Re: Show your favorite and most used benchtop PSU
« Reply #480 on: May 16, 2020, 09:56:34 am »
I have used an AVR to read small values of voltage and I didn't see any problem with it (reading a PT100, only amplification was ATtiny26's internal differential 20x gain amplifier and its internal ~2.7V Vref *). You are using different Vrefs for the pure analog part and for the display/control part, maybe your problem comes from that, what happens if they differ by, say, 40 or 50mV (or maybe not :D ).


* The schematic is here, it's commented in Portuguese but I think it won't be a big problem, there's nothing fancy:
https://troniquices.files.wordpress.com/2008/07/coscov.png

« Last Edit: May 16, 2020, 10:15:49 am by nuno »
 

Offline Pawelr98

  • Contributor
  • Posts: 33
  • Country: pl
Re: Show your favorite and most used benchtop PSU
« Reply #481 on: June 23, 2020, 02:37:55 am »
This is mine.

0-30V 0-6A

Uses uA723 with extra electronics.
Negative 2.5V rail (431 stabilized), allows to go down to 0V
extra overcurrent protection on driver stage
separate 35V supply for the chip that ensures dropout of no more than the Vcesat of power transistors
14V/28V transformer tap automatic switching, 300VA toroid transformer

The output stage is 2xTungsram 2N3055 on an outside radiator.

Case is metal.

On the right there's a tube tester.
Sometimes it also works as a low-current high-ish voltage PSU as it has a primitive 0-100V adjustable voltage source.
Just a 2N60 beeing source follower, no proper regulation.
 
The following users thanked this post: Wolfgang

Offline mariush

  • Super Contributor
  • ***
  • Posts: 3960
  • Country: ro
  • .
Re: Show your favorite and most used benchtop PSU
« Reply #482 on: July 09, 2020, 08:13:12 pm »
Not the first PSU I've built, but the first one that was more than an LM317, and built into a proper box :)


It looks nice. It could be improved quite a bit though, if you want to release the project at some point.

Fuse is in kind of a bad location, hard to shove your fingers in a corner to pull the fuse up.  Maybe should be on bottom left corner under that to-220 with heatsink.
Maybe add connectors for those soldered wires on the left side of pcb.
Plenty of room to lay the four diodes properly .... maybe consider also dual footprint , for regular bridge rectifier (the ones very common used in atx power supplies , gbj , gbu etc, they can be screwed to heatsinks)

Could maybe rotate the to-220 chips and align with that to-220 with heatsink and maybe even with the bridge rectifier, so that a wide heatsink could sink all those chips. Also give them rigidity and whatever.  May have to add insulators.

I'd rotate most through hole resistors 90 degrees , have most traces on top vertical and the ones on bottom horizontal ,  with ground fills or whatever for increased trace thickness.

The DISPLAY connector could probably be on the right corner/edge of the board instead of center, as the ribbon cable could block air flow going through case.

As for parts minimization, you could maybe add footprints for replacing those 47k resistors with 2 x 100k in parallel to get 50k resistors - close enough to 100k.

Not a fan of to-92 transistors, i'd rather make a board surface mount than use those.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf