Desing a AVR based DC Power Supply(Thinking Stage!)

Log in or register to post comments
Go To Last Post
47 posts / 0 new
Author
Message
rodney_alm
rodney_alm's picture
Level: Hangaround
Joined: Thu. Apr 17, 2008
Posts: 162 View posts
Location: Goa, INDIA
#1
Posted by rodney_alm: Tue. May 26, 2009 - 05:50 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Im Planning of building a AVR based DC power supply.

for the sake of discussions let us assume that a output range of 5v-30v is required(lets call it V_Desired), Current not exceding 1 Amp(Imax), Input voltage after rectification about 50V(V_in) and ofcourse the V_out, which should be equal to the V_Desired

The circuit i plan to build will output a pwm signal which will switch on a Mosfet to drive the load.

Now, what should be the strategy to control V_out

1) Read V_in, change pwm to match V_desired by calculation
or
2) Read V_in, Read V_out, accordingly increase or decrease pwm to match V_desired

I have no idea how would i limit the current by varrying the pwm. Can someone explain this relation

Regards Rodney

Tags:

AVR Microcontrollers, megaAVR and tinyAVR
Top
theusch
theusch's picture
Level: 10k+ Postman
Joined: Mon. Feb 19, 2001
Posts: 39455 View posts
Location: Wisconsin USA
Posted by theusch: Tue. May 26, 2009 - 05:59 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

http://www.tuxgraphics.com/elect...

All done for you. ;)

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

Top
ka7ehk
ka7ehk's picture
Level: 10k+ Postman
Joined: Fri. Nov 22, 2002
Posts: 20130 View posts
Location: Tangent, OR, USA
Posted by ka7ehk: Tue. May 26, 2009 - 06:03 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You really don't need to monitor Vin, except for verifying proper operation. Your total control algorithm would be PWM and Vout measurement. The feedback will deal with any changes in Vin (over some allowable range).

Current limiting is harder. IF you have a way to measure load current, then the usual scheme if current exceeds the limit is one of two: 1. Reduce the output voltage to some very low value, or 2, Reduce the output voltage until the current no longer exceeds the set limit. The first is good if the excessive current is likely to come from something in the circuit which will be damaged by the current. The second gives you "more nearly normal voltage" in the case the circuit just draws more current than you had estimated.

There are a number of good current sensor ICs on the market these days. See Analog Devices, Linear Technology, Maxim, Allegro, and others. Some are isolated (you probably don't need), some are bipolar (current both directions, you may not need), some don't go as high as 30V. But you should be able to find something suitable.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Top
rodney_alm
rodney_alm's picture
Level: Hangaround
Joined: Thu. Apr 17, 2008
Posts: 162 View posts
Location: Goa, INDIA
Posted by rodney_alm: Tue. May 26, 2009 - 06:12 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

theusch wrote:
http://www.tuxgraphics.com/elect...

All done for you. ;)

I Had checked that site. but not very happy with the circuit cause he used R2R DAC and Transistors and darlington transistors.

I am thinking more on the lines of PWM and Mosfets

Besides i want to make my own little baby!!.

Just reading about PID... had missed this in university as i come from a phyics background

Top
rodney_alm
rodney_alm's picture
Level: Hangaround
Joined: Thu. Apr 17, 2008
Posts: 162 View posts
Location: Goa, INDIA
Posted by rodney_alm: Tue. May 26, 2009 - 06:17 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
Current limiting is harder. IF you have a way to measure load current, then the usual scheme if current exceeds the limit is one of two: 1. Reduce the output voltage to some very low value, or 2, Reduce the output voltage until the current no longer exceeds the set limit. The first is good if the excessive current is likely to come from something in the circuit which will be damaged by the current. The second gives you "more nearly normal voltage" in the case the circuit just draws more current than you had estimated.

If i understand correctly.. if i fold back the voltage i will be able to reduce the load current.. Yes?

For measuring the load current, I was planning of just reading the voltage across a very small resistance.

Top
Microsuffer
Microsuffer's picture
Level: Hangaround
Joined: Fri. Jul 21, 2006
Posts: 195 View posts
Location: Silicon Valley
Posted by Microsuffer: Tue. May 26, 2009 - 06:57 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Sensing VIN and adjusting PWM to get the correct VOUT is called "feed forward compensation" and is done on some switching supplies. It is a faster way to react to changes to output level caused by changes to VIN. It adds complexity and I wouldn't bother with it in your first design.

Inserting a sense resistor after the inductor, or between the output and the load will work, but you get power and efficiency losses that way. Commercial designs sense current by measuring the IR drop across the inductor. They avoid sampling at the phase switch points. This approach, of course, takes a differential measurement and the phase (switching) side of the inductor will ring to well above your input voltage. Also keep in mind the high Tc for copper or your current reading will walk up as the inductor warms.

A buck regulator uses PWM changes to compensate for changes induced by input voltage change, internal losses and load variations. Most importantly, PWM increases in response to load increases and losses. You MUST limit any such increases to avoid thermal runaway. If PWM period is too large, you have an overload and must shut down. Where to limit is dependent upon your hardware - look out for very slow thermal creep.

Plan on adding an R/C circuit at the phase node to limit the ringing or you can get some very interesting failures with magic exiting the switching transistors.

Lots of fun to be had exploring this kind of circuit.

:-)

"It's easier to ask forgiveness than it is to get permission" - Admiral "Amazing" Grace Hopper.

Top
theusch
theusch's picture
Level: 10k+ Postman
Joined: Mon. Feb 19, 2001
Posts: 39455 View posts
Location: Wisconsin USA
Posted by theusch: Tue. May 26, 2009 - 08:03 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As many of your questions related to current-limiting, I'm surprised that you dismissed Tuxgraphics so readily as there is much discussion about current-limiting.

Quote:
And if thine eye offend thee, pluck it out, and cast it from thee:

If you don't like R2R DAC, then replace it with something more to your liking. The fact remains that the design does current limiting, and is AVR-based (which I assume you want as you are posting here).

Have you explored the power-supply circuits at www.discovercircuits.com ?

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

Top
Microsuffer
Microsuffer's picture
Level: Hangaround
Joined: Fri. Jul 21, 2006
Posts: 195 View posts
Location: Silicon Valley
Posted by Microsuffer: Tue. May 26, 2009 - 08:08 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Reproducing a design that works makes for an excellent test bed for making changes and getting to know the basic circuitry. Any failure is likely to be a build error, not an application of theory, or the result of your last change.

"It's easier to ask forgiveness than it is to get permission" - Admiral "Amazing" Grace Hopper.

Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Tue. May 26, 2009 - 08:32 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How about JFET's for current limiting?

Top
steve17
steve17's picture
Level: Raving Lunatic
Joined: Tue. Sep 7, 2004
Posts: 6189 View posts
Location: New York State
Posted by steve17: Tue. May 26, 2009 - 10:14 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I looked at the tuxgraphics stuff but I was never able to find the schematic. It seems to be good, but I would have to modify it. I wanted 1.5 to 5.5 volts with current limited to a few dozen ma. and the ability to set the voltage to within a couple of mv.

I ended up modifying an old power supply I built 20 years ago. It uses a plain old LM317T and has a ten turn pot. I still don't have a digital voltage readout on it, but using a DMM fixes that and it's a whole lot easier to do. :)

I still don't have any current limiting except the limiting in the LM317T itself, but some day I'll use a simple voltage regulator circuit for that. One LM317 or whatever and one resistor. Seems like a plan to me.

Mine doesn't have an AVR in it, but it could be worse. I was thinking of using an MCP182 LDO regulator which is made by, err well, ahem, I believe it's made my Microchip. Would that get me kicked off this site? :)

Top
jayjay1974
jayjay1974's picture
Level: Raving Lunatic
Joined: Wed. Oct 30, 2002
Posts: 5770 View posts
Location: The Netherlands
Posted by jayjay1974: Tue. May 26, 2009 - 10:36 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There are loads where reducing the voltage increases the current.

For some reason I don't like a MCU to control a power supply unit. The transient response will be bad and resolution will be limited. And when something goes wrong with the MCU (glitch, latent bug that suddenly surfaces etc) it can blow up your DUT or the PS unit itself, which might be an expensive exercise.

I'd use the MCU to enter voltages, current limits etc leaving the hard work to dedicated analog electronics.

Last Edited: Tue. May 26, 2009 - 10:41 PM
Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Tue. May 26, 2009 - 10:36 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well hum, since you decided to go with the simple/less fun way of doing things, how about the rest of us design an inexpensive low power (say 1A max) programmable power supply with programmable current limiting from one of the smaller Tiny's?

This kind of topic comes up often, I myself know enough to get started, but I am sure a lot of the brilliant people here will add considerably to my knowledge and that of others on this forum.

I propose a programmable current limited power supply based from an AVR and some common discrete parts only, that anyone can build for themselves.

Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Tue. May 26, 2009 - 10:44 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Some initial target specs might be:

- Vrange = 0v to 5.5v
- Irange = 0mA to 1A
- RS232/USB interface to PC?
- Low ripple
- Fast transient response
- Isolated

What else?

Last Edited: Tue. May 26, 2009 - 10:47 PM
Top
jayjay1974
jayjay1974's picture
Level: Raving Lunatic
Joined: Wed. Oct 30, 2002
Posts: 5770 View posts
Location: The Netherlands
Posted by jayjay1974: Tue. May 26, 2009 - 10:47 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

- Sink/source capability

Top
ka7ehk
ka7ehk's picture
Level: 10k+ Postman
Joined: Fri. Nov 22, 2002
Posts: 20130 View posts
Location: Tangent, OR, USA
Posted by ka7ehk: Tue. May 26, 2009 - 10:47 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jfet as current limiter are usually limited to a few 10s of ma.

Ordinary jfets are not well characterized for this, nor are they very temperature stable. A very few manufacturers make jfet "constant current sources" which is what you may be thinking about. These are somewhat better characterized, but are still limited to well under 100ma. jfets are also pretty "soft" where they change from proportional current to constant current. In my mind, this would be a very undesirable attribute for a power supply.

I would opt for either one of the hall-based current sensors or one of the active current sensors. Most of these operate at under 100mV (some a lot less) across the current sense resistor. At 100mV full scale, this only wastes 100mW at 1amp compared to 1A*5V = 5W to a load. That only represents 98% efficiency! With 10V out and a 1A load, you are at 99% efficiency, At 100mW, you hardy need any special heat dissipation.

The tuxgraphics supply does the current sensing in the "ground leg" of the supply. That removes the need for high common-mode rejection required for high-side sensing. I will bet that it works pretty well and is worth considering.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Tue. May 26, 2009 - 10:53 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok that should be doable without much problem, sinking 5.5v @ 1A is 5.5W though, might be a bit much to sink. We could limit sink capabilities to 500mA or so and reduce costs a bit.

Should this design be a switcher or linear regulator?

Top
ka7ehk
ka7ehk's picture
Level: 10k+ Postman
Joined: Fri. Nov 22, 2002
Posts: 20130 View posts
Location: Tangent, OR, USA
Posted by ka7ehk: Tue. May 26, 2009 - 11:04 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The required adjustable output from 5-30V with 50V input forces a switcher. If it were linear, the pass transistor (of ANY technology) would have 45V across it with 5V out. A 1A load gives you 45W in the series pass. Waaaay too hot.

Making a switcher adjustable adds a whole set of compromises that can be difficult, but they are a whole lot easier than getting rid of 45W.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Tue. May 26, 2009 - 11:15 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well yes you are right if the required output is in fact 5v-30v, it's a bit out of range for most casual hobbyists anyways. I was thinking of a logic range simple device, 0v-5.5v. I know I could use it myself more often than I ever need a 30v PS anyways.

Top
jayjay1974
jayjay1974's picture
Level: Raving Lunatic
Joined: Wed. Oct 30, 2002
Posts: 5770 View posts
Location: The Netherlands
Posted by jayjay1974: Tue. May 26, 2009 - 11:23 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I find 5.5V as maximum output quite limited. I'd say 15V is way more useful.

If all you want is a simple PS with adjustable current limiter, the old L200 is probably easier,cheaper and more reliable than an MCU of any kind. Not everything needs to be computerized :D

Top
AndersAnd
AndersAnd's picture
Level: Posting Freak
Joined: Wed. Apr 23, 2003
Posts: 1123 View posts
Location: Denmark
Posted by AndersAnd: Wed. May 27, 2009 - 01:33 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here's a well documented Digital Benchtop Power Supply with RS-232 interface and Windows software:
Schematics and original article in German language: www.haus-am-schueberg.org/Ralf_P...

The article was published in a 3-part article in the Elektor Electronics magazine in 2001, so if you don't understand German you can also buy the 3-part article in English here and download the source code for free:
http://www.elektor.com/magazines...(1).54823.lynkx
http://www.elektor.com/magazines...(2).54867.lynkx
http://www.elektor.com/magazines...(3).55551.lynkx

The digital part of the PSU is based on a PIC16F84 programmed in Basic. It should be possibe to port the cod to an AVR using either Basic, C or ASM.
The control loop is completely analog using op-amps. The PIC is not part of the control loop itself, but just sets the reference voltages on the op-amps via PWM and and low-pass filters to set the desired output voltage and current limit based on button or PC input through RS-232. Therefore the MCU does not slow down the transient response.
The PIC also measure the output voltage and output current, but that's just to display the readings on the built-in LCD screen and transmit to the RS-232 interface for displaying it on a optional connected PC.

Be aware of this correction:
Digital Benchtop Power Supply
November & December 2001 (000166-1/2)
The 10-µF electrolytics in this circuit appear with different voltage ratings in the parts list and the schematic. The following is offered as guidance: C3, C13-17 and C20 should have a minimum working voltage of 16 volts. C19 has to be rated at 35 V or higher. As usual, higher voltage ratings are always possible.

Top
rodney_alm
rodney_alm's picture
Level: Hangaround
Joined: Thu. Apr 17, 2008
Posts: 162 View posts
Location: Goa, INDIA
Posted by rodney_alm: Wed. May 27, 2009 - 04:11 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Oh Man, this project looks bigger than i expected it to be... My initial idea was to use this power supply to charge various lead SMF battries.. Wondering if all this hassel is worth it.

I always wanted to make a switching power supply...

Let me draw some schematics first and then go from there.

Top
ArnoldB
ArnoldB's picture
Level: Raving Lunatic
Joined: Thu. Nov 29, 2007
Posts: 3188 View posts
Posted by ArnoldB: Wed. May 27, 2009 - 06:01 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

UNiXWHoRe wrote:
Some initial target specs might be:
Have a look at the STK500 schematics. There are a few in. It is a rather simple design.

Quote:
- Vrange = 0v to 5.5v
See STK500 schematic
Quote:
- Irange = 0mA to 1A
Would have to change the regulator to a larger on - which is available.
Quote:
- RS232/USB interface to PC?
More a matter of cost then function. More important: isolated RS232 or USB.
Quote:
- Low ripple
STK500: Heavily filtered PWM driving an old fashioned linear regulator.
Quote:
- Fast transient response
STK500: Entirely left to the linear regulator if I am not mistaken.
Quote:
- Isolated
See above.

What else?

Stealing Proteus doesn't make you an engineer.

Top
AndersAnd
AndersAnd's picture
Level: Posting Freak
Joined: Wed. Apr 23, 2003
Posts: 1123 View posts
Location: Denmark
Posted by AndersAnd: Wed. May 27, 2009 - 08:36 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

rodney_alm wrote:
Oh Man, this project looks bigger than i expected it to be... My initial idea was to use this power supply to charge various lead SMF battries.. Wondering if all this hassel is worth it.

I always wanted to make a switching power supply...


Read this application note:
AVR450: Battery Charger for SLA, NiCd, NiMH and Li-Ion Batteries
http://www.atmel.com/dyn/product...
SLA = Sealed Lead Acid

There's both buck converter schematics and source code examples for AVR450.

AVR Battery Charger Reference Design - Schematics https://www.avrfreaks.net/modules...

Last Edited: Wed. May 27, 2009 - 09:01 AM
Top
dak664
dak664's picture
Level: Raving Lunatic
Joined: Sun. Jun 15, 2008
Posts: 2115 View posts
Location: North Carolina USA
Posted by dak664: Wed. May 27, 2009 - 08:39 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Also look at Atmel's BC100 battery charger reference board http://www.atmel.com/dyn/Products/tools_card.asp?tool_id=4219 for examples of filtered buck converters. For $99 that board might even meet your specs, including RS232. The firmware would take a bit of reworking though.

Top
giovannitto
giovannitto's picture
Level: New Member
Joined: Tue. Jul 24, 2007
Posts: 10 View posts
Location: Italy
Posted by giovannitto: Wed. May 27, 2009 - 08:48 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Look at :
http://powerelectronics.com/power_systems/dc_dc_converters/power_buckconverter_design_demystified/ for a very interesting discussion/design of buck converter.

Top
AndersAnd
AndersAnd's picture
Level: Posting Freak
Joined: Wed. Apr 23, 2003
Posts: 1123 View posts
Location: Denmark
Posted by AndersAnd: Wed. May 27, 2009 - 08:54 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Also read this 2-part Circuit Cellar article with source code:

A Better Battery Charger
Part 1: Charging and Termination http://archive.chipcenter.com/ci...
Part 2: Hardware and Software Implementation http://archive.chipcenter.com/ci...

Top
jayjay1974
jayjay1974's picture
Level: Raving Lunatic
Joined: Wed. Oct 30, 2002
Posts: 5770 View posts
Location: The Netherlands
Posted by jayjay1974: Wed. May 27, 2009 - 10:40 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In 1988 or so Elektor published another digitally controlled PSU, but it was bigger and with a 8052AHBASIC IIRC, and LED displays with bargraphs, encoder knobs and a custom membrane front panel foil. I've actually seen the device in the flesh on an electronics show :)

Top
grapeghot
grapeghot's picture
Level: Rookie
Joined: Sun. Feb 1, 2009
Posts: 25 View posts
Posted by grapeghot: Thu. May 28, 2009 - 06:38 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Right now, I am playing around with an AVR based power supply, which involves using a handful of components based around a LM317. I saw it being discussed on another forum and it seems to work quite well. The circuit is shown below. Basically, it replaces the potentiometer that you see in the standard LM317 variable power supply circuit with a n-channel mosfet. It works by varying the mosfet gate voltage, which in turn varies the Rds. I guess you can call it a solid potentiometer. A simple output divider feedback loop is used to monitor the voltage, so that adjustments can be made via one of the AVR pins (PD2 as shown). Using the right combination of R3 and C1 will allow it to reach the target voltage in a reasonable amount of time and precision. Once the target voltage has been reached, the output pin is tri-stated and capacitor C1 holds the voltage at the gate. Only periodic adjustments are made to stabilize the voltage. I have it breadboarded at the moment and it seems to be running quite well. Eventually, I may actually build this into a new bench supply.

Attachment(s): 

Image icon avr317.png
Top
jayjay1974
jayjay1974's picture
Level: Raving Lunatic
Joined: Wed. Oct 30, 2002
Posts: 5770 View posts
Location: The Netherlands
Posted by jayjay1974: Thu. May 28, 2009 - 07:11 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think it would be better to use the reference voltage for feedback, not the output. What happens if the output is (near) short-circuited? Your control loop will try to compensate and pump up the reference voltage. When the short circuit goes away the output voltage will overshoot badly before the control loop regains control again.

Top
a_r_khorasani
a_r_khorasani's picture
Level: New Member
Joined: Tue. Feb 6, 2007
Posts: 9 View posts
Posted by a_r_khorasani: Thu. May 28, 2009 - 07:39 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I am thinking more on the lines of PWM and Mosfets
Besides i want to make my own little baby!!.

after filtering your input power with 12uH/1A and then 6800uF/63V parts, its prepared to feed a ferrite core (for example RM10 or EE28x34 with gape)transformer at 50KHz (with about 16:16 for N1:N2 and 0.5mm wire). another side of N1 going to Drain of a n-channel logic level MOSFET (such as IRL2910S/L); Source to Ground, and Gate via a 22 Ohms resistor to PWM pin of microcontroller.
output side of transformer will be rectified using a bridge (of Schotky or fast diodes/ you may use two BYV32E ) and then filtered by 4700uF/63V and then 12uH/1A and then 4700uF/63V parts. also should have a resistor (1K / 3 W) to ground for minimum current.

finally, using a divider (22K 1% and 1K 1%) with a 100n, it is directed to an ADC pin of your AVR.

in a loop with 10 milliseconds delay between each cycle, and a ten bit PWM, write some thing such as
Error=ADCvalue-OCR1Avalue;
Error/=2;
OCR1Avalue+=Error;

Regards
A.R.Khorasani
http://www.instrumentalanalysis.com

Regards
A.R.Khorasani
http://www.instrumentalanalysis.com

Top
avra
avra's picture
Level: Hangaround
Joined: Fri. Nov 9, 2001
Posts: 301 View posts
Location: Serbia
Posted by avra: Thu. May 28, 2009 - 10:49 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@grapeghot:
LM317 with AVR controlled digital pot might also be an option :-)

Top
steve17
steve17's picture
Level: Raving Lunatic
Joined: Tue. Sep 7, 2004
Posts: 6189 View posts
Location: New York State
Posted by steve17: Thu. May 28, 2009 - 12:14 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

grapeghot:
That looks interesting, but I'm only an amateur.

I might put a resistor across the FET (drain to source) to limit the output voltage to 5.5 volts in any case, so my AVRs would be safe. And a pullup on the FET gate so at turnon the output voltage would be at the minimum value.

Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Thu. May 28, 2009 - 01:09 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hehe Actually I have been working on this since a couple of days ago when we started... I too am going with a LM317 (which has hard current limiter btw jayjay) but in a slightly different configuration.

The biggest problem with the LM317 will be dissipating all the energy over the entire range. Let's say we have 18v in after transformer, rectifier, and filter. If you want 10V output @ 1A thats 8W to dissipate. I highly doubt the LM317 will survive long.

What I am trying to accomplish is have 3 or 4 LM317's cascaded in series so that each regulator only has to dissipate about 3V or 4V each. For example first regulator would have it's divider set so that a digipot at R2 would set it between 15V-11V output voltage. Next one would take Vin from that output, and have its divider setup to output between 12V-9V, etc. This would have the effect of equalizing the thermal load across all regulators, and keep things quite cool. I am still working on the maths though... ;)

In the meantime, attached is the 120VAC to 18VDC xformer-rectifier-filter circuit.

Attachment(s): 

Image icon PowerIn.JPG
Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Thu. May 28, 2009 - 02:01 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Does anyone have a reference table for LM317 divider configurations? I remember seeing one a while ago, but I cannot find it anymore...

Top
steve17
steve17's picture
Level: Raving Lunatic
Joined: Tue. Sep 7, 2004
Posts: 6189 View posts
Location: New York State
Posted by steve17: Thu. May 28, 2009 - 03:13 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tables are ancient history. Do you want them on clay tablets or stone slabs? :)

I write programs to design my voltage regulators for me. I might be persuaded to lend you mine, but be forewarned. My gui programs don't look like anyone else's. I invented windows for Windows that are easy to construct long before Microsoft did.

Top
jayjay1974
jayjay1974's picture
Level: Raving Lunatic
Joined: Wed. Oct 30, 2002
Posts: 5770 View posts
Location: The Netherlands
Posted by jayjay1974: Thu. May 28, 2009 - 03:30 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I usually enter all equations that are in the datasheet into an Excel sheet.

Top
ArnoldB
ArnoldB's picture
Level: Raving Lunatic
Joined: Thu. Nov 29, 2007
Posts: 3188 View posts
Posted by ArnoldB: Thu. May 28, 2009 - 04:12 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

UNiXWHoRe wrote:
What I am trying to accomplish is have 3 or 4 LM317's cascaded in series
Parallel them. Every self-respecting LM317 datasheet ever written (well, copied from manufacturer to manufacturer), has such an application example in it.
UNiXWHoRe wrote:
I am still working on the maths though... ;)
The math gets simple when you parallel them. Done right, n LM317 means P = Ptotal/n power per LM317, and all are driven by one feedback circuit.

Stealing Proteus doesn't make you an engineer.

Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Thu. May 28, 2009 - 04:19 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ah well that DOES sound simpler.. LOL

I wasn't sure if it was safe to parallel LM317's... One advantage of having them in series is that instead of having one 8-bits pot control the whole range, I would have 8N bits resolution with them cascaded. So if I cascade 3 of them and use a quad digipot for the feedback, I end up with 24-bits resolution over the entire range. 800nV/step VS 50mV/step

Does that make sense?

Ok 24 bits might be a bit much, but I think I can certainly settle with 2 of them in series, each with one or two in parallel.

Top
a_r_khorasani
a_r_khorasani's picture
Level: New Member
Joined: Tue. Feb 6, 2007
Posts: 9 View posts
Posted by a_r_khorasani: Thu. May 28, 2009 - 04:40 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

using non switch mode configurations (such as lm317), you will waste input power.
by the way, Lm317 only work till 37 Volts !

Regards
A.R.Khorasani
http://www.instrumentalanalysis.com

Regards
A.R.Khorasani
http://www.instrumentalanalysis.com

Top
andrew99
andrew99's picture
Level: Resident
Joined: Wed. Jan 10, 2007
Posts: 923 View posts
Location: Toronto, Ontario, Canada
Posted by andrew99: Thu. May 28, 2009 - 04:47 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

All,

I have a decent bench supply. The unit is adjustable from 0->60VDC@ 3 amps. The supply switches the input voltage to the regulator components as the o/p voltage increases. So at lower voltages the unregulated supply is lower than when the o/p voltage is at max. The unregulated supply increases in three stages keeping the voltage drop across the regulator to a minimal level.

A

PS: Built lots of bench supplies myself, just never seemed to be able to make one as nice as $300.00 would buy.

AVR Studio 4 Ver. 4.18 684
avr-gcc Ver. 4.3.0
ISIS 7
ELECTRA

Top
grapeghot
grapeghot's picture
Level: Rookie
Joined: Sun. Feb 1, 2009
Posts: 25 View posts
Posted by grapeghot: Thu. May 28, 2009 - 05:05 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974;
Good point, I never thought about short circuits with the output feedback.

avra:
Using a digital pot would be an option, but I am experimenting with parts that are on hand.

steve17:
Good idea about pullup on gate.

Top
ArnoldB
ArnoldB's picture
Level: Raving Lunatic
Joined: Thu. Nov 29, 2007
Posts: 3188 View posts
Posted by ArnoldB: Thu. May 28, 2009 - 05:25 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

UNiXWHoRe wrote:
I wasn't sure if it was safe to parallel LM317's... That's what datasheets are for.
UNiXWHoRe wrote:
One advantage of having them in series is that instead of having one 8-bits pot control the whole range, I would have 8N bits resolution with them cascaded. So if I cascade 3 of them and use a quad digipot for the feedback, I end up with 24-bits resolution over the entire range. 800nV/step VS 50mV/step

Does that make sense?

Not at all. Only the last regulator will control you output voltage. The last regulator hardly cares what the previous regulator delivers, as long as the difference is above the dropout voltage. Your overall resolution will be the resolution of your control of the last regulator, 8 bit.

And digipots are a waste of money here. All you have to do is to drive the ADJ pin 1.25 V below the desired output voltage. An OpAmp or transistor can do that. Well filtered PWM in turn can drive ther OpAmp or transistor. When using bipolar power supply for the OpAmp, and correctly dimensioning it, you can even regulate down to 0 V, instead of the LM317's typical 1.25 V minimum output.

Alternatively, every self respecting LM317 datasheet has a 4 bit digital control application in it. They switch resistors with bipolar transistors. It is easy to extend it to more bits, e.g. using FETs or an IC with FET switches. And that would be kind of a home made DAC, and in fact, you could use a DAC instead, too.

All this has been done a lot of times in the last 30 or so years, because that's roughly how old the LM317 is. It is not that this is a mystery device.

Stealing Proteus doesn't make you an engineer.

Top
steve17
steve17's picture
Level: Raving Lunatic
Joined: Tue. Sep 7, 2004
Posts: 6189 View posts
Location: New York State
Posted by steve17: Thu. May 28, 2009 - 07:17 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a commercial one that costs $95. I think it's pretty good for the price. 0-30V, 0-3A. That's plenty of power to destroy my AVR's in an instant. It makes me nervous just thinking of using it for that purpose. And it's hard to adjust the voltage precisely, to the nearest 10 mv or so. The digital display only shows voltage with 100 mv precision, and current with 10 ma precision.

When I adjust the voltage, I can hear relays clicking. I suppose its changing taps on the power transformer.
http://elexp.com/tst_3003.htm

I needed one that wouldn't blow my electronics, no matter how foolish I was. So I built one that gives 1.5-5.5 volts. Now I can relax.

Top
UNiXWHoRe
UNiXWHoRe's picture
Level: Posting Freak
Joined: Fri. Nov 3, 2006
Posts: 1118 View posts
Posted by UNiXWHoRe: Thu. May 28, 2009 - 07:38 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think you misunderstood my intent. Check attached image... I would have a first stage to handle output voltages from 15V to 7V, and a second stage to handle 8V to 1.25V. Each stage would be controlled by its own 8-bits digipot, making the overall resolution 16-bits. A mixing stage probably composed of a couple transistors would take care of the transition between stage 1 and stage 2. LM317 response being normally logarithmic, I think it would also improve the overall linearity.

Of course now I realize I could simply PWM the ADJ terminal of multiple LM317 in parallel from a 16-bits timer. Duh..

Attachment(s): 

Image icon Stagemix.JPG
Top
dbvanhorn
dbvanhorn's picture
Level: Resident
Joined: Thu. Jan 31, 2008
Posts: 701 View posts
Posted by dbvanhorn: Thu. May 28, 2009 - 07:50 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well.. For the ultimate in overkill, I have this nice Sorensen supply, 0-55V, 0-55A which you could treat as a single component, and control with an analog voltage or dpot driven by an AVR. It needs 220V single or three phase input though.

VBG! :)

Top
a_r_khorasani
a_r_khorasani's picture
Level: New Member
Joined: Tue. Feb 6, 2007
Posts: 9 View posts
Posted by a_r_khorasani: Thu. May 28, 2009 - 08:28 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
for the sake of discussions let us assume that a output range of 5v-30v is required(lets call it V_Desired), Current not exceding 1 Amp(Imax), Input voltage after rectification about 50V(V_in) and ofcourse the V_out, which should be equal to the V_Desired

Regards
A.R.Khorasani
http://www.instrumentalanalysis.com

Top
rodney_alm
rodney_alm's picture
Level: Hangaround
Joined: Thu. Apr 17, 2008
Posts: 162 View posts
Location: Goa, INDIA
Posted by rodney_alm: Sat. May 30, 2009 - 02:49 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have always used the LM317 for most of my power supply projects. But ultimately you have these limits..

1) You need to have atleast a 3v differential between the o/p and i/p

2) Having anything higher than that gives you => (differential voltage)x Load current ===> HEAT also know as power dissipation.

3) trying to get it to work for loads about 1A Heat up the 317. Offcourse a current pass transitor can be used to overcome this.

Top