LM723 PSU with 0V lowest voltage

[earckens]

POST Issue 03 of 2016_12_09

Hy earckens,

Below a complete schematic for the LM723 power supply, with the following features:
(1) Common base transistor to limit the voltage seen by the LM723 (and not increase loop gain).
(2) Output voltage adjustment from 0V to 36V (but note my previous comment about getting down to zero Volts and leakage currents).
(3) Only a simple constant current limit of 1.5 A, which is the maximum current that two 2N3055 transistors will safely provide, as previously stated.

spec

View attachment 102983​

Hi spec, if I am correct your voltage feedback circuit compared to the original one looks like this below.

Only 36V output voltage? Higher not feasible?

Current limit to 1.5A by increasing the current sense from 0.12R to 0.4R? Then I would keep the 0.12R: my current heatsinks are 5x3x12 in size, about 1.4C/W

Lots of trimming pots, the -comp input to pin 4 seems fairly complicated: the original setup was not feasible?

Thanks for the work!
Erik


Edit: oeps, my first upload of a drawing is clearly visible

 

[spec]

POST Issue 03 of 2016_12_11

Hi earckens,

I have now had a look at the output voltage adjustment for your schematic of post #135

With the present resistor values the output voltage range would be, 0V to 32.6V

But the power supply of post #135 should be capable of 0V to 35V.

With a 5K voltage adjustment potentiometer, and to achieve an output range of 0V to 35V, the two 18K resistors should be changed to 47K and the two 82K resistors should be changed to 230K.

I have also simplified the circuit around Q4 as shown in post #137. On your schematic of post #135 please delete R12 and replace with a trace. Also change R15 to 1K Ohms.

spec

(our posts crossed earckens)

Addendum (just for the record)

The relationship between the two voltage controlling resistors and the maximum power supply output voltage is, Voutmax = (7.17 *R2)/R1, where R1 was 18K and R2 was 82K.

 

[earckens]

With a 5K potentiometer, and to achieve an output range of 0V to 35V, the two 18K resistors should be changed to 47K and the two 180K resistors should be changed to 230K.

spec

Addendum (just for the record)

The relationship between the two voltage controlling resistors is R2= Vout * R1/7.15 Where R1 was 18K and R2 was 180K

... and the two 180K resistors ...: you mean 82k ..?

 

[spec]

... and the two 180K resistors ...: you mean 82k ..?

Arrragh- I am getting into a right tangle.

Have a nice Saturday on holiday and I will have it all sorted by the time you come back.

spec

 

[earckens]

... and the two 180K resistors ...: you mean 82k ..?

..in that case Vout = R2 * 7.15 / R1 = 32.5V which is correct for the original design. Very nice find of you, this formula! Thks!
Erik

 

[earckens]

Arrragh- I am getting into a right tangle.

Have a nice Saturday on holiday and I will have it all sorted by the time you come back.

spec

Right, we will, and be thinking of you will sipping our champagne

 

[spec]

Right, we will, and be thinking of you will sipping our champagne

(post #143 now corrected)

spec

 

[spec]

Your voltage feedback circuit compared to the original one looks like this below.
Lots of trimming pots, the -comp input to pin 4 seems fairly complicated: the original setup was not feasible?

Both circuits need trimming, either by changing resistor values or by trim pots, to cater for the various circuit tolerances and obtain exactly 35v and 0V but you can probably get by without trimming.

Only 36V output voltage? Higher not feasible?

I haven't calculated the maximum output voltage, but I am working on 35V at the moment. We can always maximize the output voltage later.

Current limit to 1.5A by increasing the current sense from 0.12R to 0.4R? Then I would keep the 0.12R: my current heatsinks are 5x3x12 in size, about 1.4C/W

The limiting factor with 2N3055s is the safe operating area (SOA). At 46V, when all factors are taken into account (mainly junction temperature), 750mA would be a maximum safe current.

spec

​

https://www.onsemi.com/pub_link/Collateral/2N3055-D.PDF

 

[earckens]

Good morning spec, just a question on your graph: Vce could be max 46Vdc, however, under full load this would drop a bit, say to 40Vdc. So I see on the graph the intersection for the lowest graphline to be at 3A.
Another question: how could I get a LED to turn on when the current limit is active?

Have a great day!
Erik

 

[spec]

Good morning earckens.

Hope you had a good day in Lille.

...just a question on your graph: Vce could be max 46Vdc, however, under full load this would drop a bit, say to 40Vdc. So I see on the graph the intersection for the lowest graph line to be at 3A.

That is true at 25 Deg C junction temperature, but at 150 Deg C, which could be the sort of 2N3055 junction temperature when the the power supply is passing a high current with the output set to 0V, that safe current would drop to around 1.5A. There are also other factors that limit the maximum current. So I would suggest that maybe 1A maximum per transistor may be a good compromise.

But, in addition to the spec sheet limits, from experience with similar power supplies, I would not recommend more than 1A per 2N3055, on reliability grounds.

By the way, you can set the maximum current limit to whatever you like (the LM723 and transistors will source around 6A worst case).

The other aspect of increasing the current are
(1) As the collector current of the 2N3055 goes up so does the VBE and minimum VCE.
(2) The total current flowing out of the reservoir capacitors not only drops the voltage across the reservoir capacitor but also increases the ripple voltage.

All three of these factors limit the maximum output voltage from the power supply.

Another question: how could I get a LED to turn on when the current limit is active?

I will have a look at this. With the LM723 type of current limiting having a LED illuminate during current limit mode is a bit awkward- I think.

spec

 

[earckens]

Good morning earckens.

Hope you had a good day in Lille.
spec

Hi spec, yes we did have; a bit crowdy -the X-mas market was totally overcrowded and maximum security at the gates- but we had a very good time regardless. Nice town, beautifull old city section: if you ever get the chance it is really worth the trip; lots of english speaking visitors too.

Thank you for the explanation on SOA: I learn a lot from you!

If it gets too complicated for the current limit LED I will look for a software solution; feed the input of -comp to a high impedance analog input of the MCU I use to display V and I and do some calculations.

Have a great Sunday,
Erik

 

[Les Jones]

Hi Erik,
I did not follow the start of this thread fully. My feeling at this point is that the specification has evolved from the original and there seems to have been some constraints put on the design such as it must use 2N3055 pass transistors and it must use an LM723. I feel it would have been better to start with the required specification. and to use components that would allow it to meet the specifications. I can see no easy way of adding an overcurrent LED other than having an external (To the LM723) circuit to do the overcurrent detection and then use a signal from that to force the LM723's overcurrent circuit to shut it down. The existing circuit has evolved with one fudge after the other to get around the constraints. (Fudge is sort of a slang term finding a way to get round a problem but not doing it properly.)

Les.

 

[earckens]

Hi Erik,
I did not follow the start of this thread fully. My feeling at this point is that the specification has evolved from the original and there seems to have been some constraints put on the design such as it must use 2N3055 pass transistors and it must use an LM723. I feel it would have been better to start with the required specification. and to use components that would allow it to meet the specifications. I can see no easy way of adding an overcurrent LED other than having an external (To the LM723) circuit to do the overcurrent detection and then use a signal from that to force the LM723's overcurrent circuit to shut it down. The existing circuit has evolved with one fudge after the other to get around the constraints. (Fudge is sort of a slang term finding a way to get round a problem but not doing it properly.)

Les.

Les, I agree with you.

I will start a new thread on this subject: describe the initial subject, describe the "wishlist" of requirements and make a resumé of what has been covered in this thread re. relevant items.

Erik

 

[spec]

Hi Les and Erik,

We have come a long way with this power supply and I think we are reaching a workable design, especially withe Erik's EAGLE schematic which makes the circuit analysis much easier, so it would be a shame to abandon it at this stage.

Incidentally there is a parallel thread about power supplies @ https://www.electro-tech-online.com/threads/2n3055-and-heat.149521/page-7#post-1281612 which covers a 'universal' power supply architecture that can be adapted to any reasonable requirements, including a negative version.

spec

 

[Les Jones]

Hi spec & Erik,
Here is a suggestion for one possible solution for fitting an overcurrent LED.

It would involve moving the current sense resistor to the positive rail.

Les.

 

[earckens]

Hi Les and Erik,

We have come a long way with this power supply and I think we are reaching a workable design, especially withe Erik's EAGLE schematic which makes the circuit analysis much easier, so it would be a shame to abandon it at this stage.

Incidentally there is a parallel thread about power supplies @ https://www.electro-tech-online.com/threads/2n3055-and-heat.149521/page-7#post-1281612 which covers a 'universal' power supply architecture that can be adapted to any reasonable requirements, including a negative version.

spec

spec, I agree with you too, more even now that I read your comment.

I think it would indeed not add more benefit to start up a new thread, since we are almost at the end of our story about this PSU.

And on a personal note: I enjoy immensely the back and forth with spec in this thread

Erik

 

[earckens]

Hi spec & Erik,
Here is a suggestion for one possible solution for fitting an overcurrent LED.
View attachment 103025

It would involve moving the current sense resistor to the positive rail.

Les.

Hi Les, this means an altogether change in design for this unit: I am sure it will work, but when comparing this with the alternative: feeding the -comp into a high impedance MCU analog input and adding about 15 lines of code to an existing program.

Thanks for the thinking about my question; I will print out the solution for later referral.

Erik

 

[spec]

Hi spec & Erik,
Here is a suggestion for one possible solution for fitting an overcurrent LED.
View attachment 103025

It would involve moving the current sense resistor to the positive rail.

Les.

Pretty damn neat Les- I like it.

spec

 

[earckens]

Hi spec, in one of your diagrams you use the BC546. I have a similar transistor: C1815, but is it as good and as reliable as your proposal? Reason Iask is that I read her and there about low reliability of "chinese clones": I am not sure whether mine is genuine or a clone, can't tell. But would be a shame not to use a -in my opinion- a good equivalent?
Erik

 

[spec]

Hi spec, in one of your diagrams you use the BC546. I have a similar transistor: C1815, but is it as good and as reliable as your proposal? Reason Iask is that I read her and there about low reliability of "Chinese clones": I am not sure whether mine is genuine or a clone, can't tell. But would be a shame not to use a -in my opinion- a good equivalent?
Erik

The C1815 should be OK but not advisable: the CEo is only 50V compared to 65V for the BC546.

The function of the BC546 is only voltage level shifting- no current gain is involved. The same current that is sunk from the emitter, essentially flows into the collector.

spec