Author Topic: lm723 based linear power supply  (Read 7094 times)

0 Members and 1 Guest are viewing this topic.

Offline sahko123Topic starter

  • Frequent Contributor
  • **
  • Posts: 318
  • Country: ie
lm723 based linear power supply
« on: September 27, 2016, 05:42:42 pm »
hello

this post is for people who could help with the design of my linear power supply based around the lm723 chip. I wanted to ask if there are any things I should look out for in the creation of this project. the lm723 chip will drive a small pnp transistor pass element into a larger npn transistor pass element. the sense terminals will be at the output with current limiting at the un regulated side of the supply. the supply will go up to at least 25 volts at 3-4 amps.
« Last Edit: September 28, 2016, 04:15:42 pm by sahko123 »
Asking for a friend
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14016
  • Country: de
Re: lm723 based linear power supply
« Reply #1 on: September 27, 2016, 06:12:11 pm »
An output stage with PNP / NPN combination will have gain by itself. So you have to check the "local" stability of the combination. Also the output transistor in the LM723 is likely part of that local loop. Anyway it's better to show a drawing of the supposed output stage. I have an idea of the circuit, but there are more possible combinations of 3 transistors.

With 25 V and 5-6 A, there could be a lot of heat dissipation. So likely a single transistor will not be enough, but more like a parallel combination of 2-4 large NPNs with emitter resistors. So check the SOA of the transistors and don't get fooled from P_tot at 25 case.

Having the shunt for current limiting on the low side could make things more complicated. The simple current limit works well on the high side as well. One might be able to reuse the emitter resistors needed for current sharing for the current limit as well.
 

Offline sahko123Topic starter

  • Frequent Contributor
  • **
  • Posts: 318
  • Country: ie
Re: lm723 based linear power supply
« Reply #2 on: September 27, 2016, 07:02:46 pm »
right thanks for the response but for the current limiting I want a variable potentiometer with a transistor at the other end of the potentiometer. I was thinking that at the regulated side the current limiter would affect the output voltage displayed on the analog voltage panel meter. for the output stage I would imagine 3-4 to-3 style npns but two should be enough with well designed thermodynamics.

thanks for the response
Asking for a friend
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14016
  • Country: de
Re: lm723 based linear power supply
« Reply #3 on: September 27, 2016, 07:47:57 pm »
For the LM723 it is easy to add a simple current limiting on the high side. This can be adjustable, but there is some temperature dependence, as the "ref." voltage used in this case is just a BE junction in the chip. On the high side one could reuse the emitter resistors as a shunt - current regulation is not very accurate anyway.

It is slightly more difficult to add current limiting on the low side, as the voltage reference is towards ground. The simple version is than to have the supply current for the chip also flowing through the shunt. Otherwise there is not much difference. Having the current shunt on the low side might be an advantage if you want to add a digital display of the measured current.
As the current limiting is not very accurate, the useful range for adjustment with just a pot is not very large. So for small current one might consider a switch to change the shunt (e.g. 2 ranges for maybe 0.5-6 A and 0.05-1 A)

Another point is how fart the voltage setting needs to go down. The simple circuits start at something like 1.5 V, but it is possible to go all the way to 0 V.

For a 25 V output, the raw voltage might be well in the 30 V range. This would up to 90 W of heat for a short at 3 A. With transistors in parallel one should include some tolerance in current sharing. So I would not use an 2n3055 at more than 2 A here. The power transistors are not that expensive anymore, so it's better to allow for some extra.

At this power level one usually considers switching the raw voltage (or use a second level) to reduce the power loss. 
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16510
  • Country: us
  • DavidH
Re: lm723 based linear power supply
« Reply #4 on: September 27, 2016, 08:41:48 pm »
I built one like this long ago which used the 723 to drive an LM317 which itself had a pair of PNP power transistors for current amplification and greater power dissipation.  The idea was with the LM317 mounted on the same heat sink as the PNP power transistors, its thermal protection could protect them as well.

Reusing the emitter ballast resistors for the current limit sense can work.  If uneven current sharing prevents good accuracy, then averaging the signal from all of the emitter ballast resistors can help as does matching the transistor Vbe voltages.

Another way to protect the pass transistors is to implement foldback current limiting.
 

Offline setq

  • Frequent Contributor
  • **
  • Posts: 443
  • Country: gb
Re: lm723 based linear power supply
« Reply #5 on: September 27, 2016, 09:21:40 pm »
Read the Art of Electronics Chapter 9, conveniently the sample chapter! http://artofelectronics.net/wp-content/uploads/2016/02/AoE3_chapter9.pdf

There's a whole study of linear power supply design including the LM723.
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: lm723 based linear power supply
« Reply #6 on: September 28, 2016, 07:08:45 am »
You don't give details about what you need: do you want to build a bench power supply?

LM723 is not the best choice for control circuit of a bench power supply.
The current limiter of the LM723  is very fast but not precise.
Use it only for protection of the output transistors, not for precision current limiting.

Look for the Philips PE1542 power supply schematics.
That's a very good application of the LM723 for a bench power supply.

You can download the service manual with schematics  here:
http://elektrotanya.com/philips_pe-1542_double_0-20v-1a_0-7v-3a_triple_laboratory_dc_power_supply_1978_sm.pdf/download.html

Wait for "processing" to became "get manual".
 

Offline ivaylo

  • Frequent Contributor
  • **
  • Posts: 661
  • Country: us
Re: lm723 based linear power supply
« Reply #7 on: September 28, 2016, 07:20:49 am »
Wow, why is lm723 still around, is it that good? I built my first power supply with it in 1986. There are other ones this old, but easy to use lm317, 78xx, etc. lm723 is a finicky beast, with external current shunts, exact passives arround, etc. Isn't there better stuff in 2016?
 

Offline setq

  • Frequent Contributor
  • **
  • Posts: 443
  • Country: gb
Re: lm723 based linear power supply
« Reply #8 on: September 28, 2016, 07:31:46 am »
It's 47 years old and still works fine :)

As for precision current limiting, you're not constrained to just use the LM723 features. It's a Lego kit. Add an external opamp to measure current and stick that and a diode or gate in the feedback loop of the LM723 and you're done and have CC/CV.

It's also dirt cheap. Better Linear Tech and TI options are out there but at 10x the cost. That's why it's still around.

Also my bench PSU doesn't even have any ICs in it and is 40 years old. Fully discrete. Superiority through adding new ICs is a bad route to go down.
« Last Edit: September 28, 2016, 07:33:40 am by setq »
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: lm723 based linear power supply
« Reply #9 on: September 28, 2016, 07:47:13 am »
As Kleinstein wrote, a linear power supply of 25V 6A will have a huge power dissipation of more or less 180W in short circuit condition.
You will have to choose a way to reduce this dissipation:
- pre-regulator
- tap changing
- foldback current limiting
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19284
  • Country: gb
  • 0999
Re: lm723 based linear power supply
« Reply #10 on: September 28, 2016, 07:48:59 am »
I built one like this long ago which used the 723 to drive an LM317 which itself had a pair of PNP power transistors for current amplification and greater power dissipation.
I think you're confusing the internal LM317 transistors with the external pass transistor(s) required to increase the current output. The Lm317 has an NPN Darlington pair for the internal pass transistors.
http://www.onsemi.com/pub_link/Collateral/LM317-D.PDF
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: lm723 based linear power supply
« Reply #11 on: September 28, 2016, 09:20:35 am »
...The idea was with the LM317 mounted on the same heat sink as the PNP power transistors, its thermal protection could protect them as well....
It is only a slow acting protection because they are not directly thermally coupled.
There is a delay by the thermal inertia of the heat sink.

It is also probably not working.
It may not be use to replace a thermostat of 90° or 95°C  (a klixon is cheap and reliable.)
Shutdown of the LM317 only occur at high junction temperature (140°C ?) and junction temperature of the pass power transistor is always higher than the temperature of the heatsink, so it will protect the pass transistor only if junction temperature of the LM317 is higher than junction temperature of the pass transistor.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16510
  • Country: us
  • DavidH
Re: lm723 based linear power supply
« Reply #12 on: September 28, 2016, 11:02:34 am »
I built one like this long ago which used the 723 to drive an LM317 which itself had a pair of PNP power transistors for current amplification and greater power dissipation.
I think you're confusing the internal LM317 transistors with the external pass transistor(s) required to increase the current output. The Lm317 has an NPN Darlington pair for the internal pass transistors.
http://www.onsemi.com/pub_link/Collateral/LM317-D.PDF

No, not at all.  The LM317 was driving an external pair of power PNP transistors on its input.  Sorry, I did not make that very clear.

...The idea was with the LM317 mounted on the same heat sink as the PNP power transistors, its thermal protection could protect them as well....
It is only a slow acting protection because they are not directly thermally coupled.
There is a delay by the thermal inertia of the heat sink.

It is also probably not working.
It may not be use to replace a thermostat of 90° or 95°C  (a klixon is cheap and reliable.)
Shutdown of the LM317 only occur at high junction temperature (140°C ?) and junction temperature of the pass power transistor is always higher than the temperature of the heatsink, so it will protect the pass transistor only if junction temperature of the LM317 is higher than junction temperature of the pass transistor.

The pass transistors are tightly coupled to the heat sink and the LM317 is not and the LM317's share of the current is deliberately made large with a low value of base-emitter shunt resistance so it operates at a higher than necessary power dissipation.  So with a higher thermal resistance and deliberately high power dissipation, the LM317 operates at a higher junction temperature than the pass transistors and goes into current limit before they fail.

Safe operating area considerations under fast load changes like an output short still require some other non-thermal based protection circuit like foldback current limiting which the LM317 may provide as well.
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: lm723 based linear power supply
« Reply #13 on: September 28, 2016, 12:04:20 pm »
Indeed, if the LM317 is not tightly coupled to the heat sink and the LM317's share of the current is deliberately made large, you could perhaps let the heat sink temperature to be higher than with a thermostat protection.

But i think it does no matter  because what is generally the most limiting specification , that's the SOA (safe operating area) who limit de max. current of each pass transistor.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16510
  • Country: us
  • DavidH
Re: lm723 based linear power supply
« Reply #14 on: September 28, 2016, 01:13:45 pm »
If you do it right, the 317 junction temperature tracks the power transistor junction temperature.  In practice, it is a pretty crude way to do it but it works fine to prevent catastrophic failure.  I think the 317 SOA protection takes junction temperature into account but I usually end up derating the power transistors so much that it would not matter; by the time I have enough worst case power dissipation, the SOA is covered.

As far as the 723, with its zener reference it makes for a lower noise regulator than bandgap designs.  I do not really like it myself but it does make for a less expensive and more integrated solution than one using a separate reference and operational amplifier.  One thing I need to try is using the 723's output transistor as a heater and current sense transistor as a temperature sensor to "ovenize" the 723 zener diode.
 

Offline setq

  • Frequent Contributor
  • **
  • Posts: 443
  • Country: gb
Re: lm723 based linear power supply
« Reply #15 on: September 28, 2016, 02:35:25 pm »
I've got an ST LM723 as a voltage ref that is good to +/-20uV from 5-43oC (freeze spray plus hair dryer). Not bad out of the box! Ovenise could be good.
 

Offline sahko123Topic starter

  • Frequent Contributor
  • **
  • Posts: 318
  • Country: ie
Re: lm723 based linear power supply
« Reply #16 on: September 28, 2016, 04:21:57 pm »
thanks for this large mass of responces ive figured that 5-6 amps would be an overly large number. so ive thought that around 3-4 amps would be a more "proper" number. i wanted to use the lm723 as i have alot of them but nothing to do with them. i also wanted them for the voltage reference
Asking for a friend
 

Offline stj

  • Super Contributor
  • ***
  • Posts: 2153
  • Country: gb
Re: lm723 based linear power supply
« Reply #17 on: September 29, 2016, 12:35:06 am »
real world example on page11
http://arcarc.xmission.com/PDF_Arcade_Williams/Defender_Drawing_Set_(16P-3001-101_R-T)_April_1981.pdf

i fix these, and they are so reliable you would not believe it!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf