Variable PSU stability problem

[tempray]

Hello to everybody,

I have encoughtered some instability problems with this schematics :


I build them two, current limit works, @ 30V 2.5A load works also (100mV drop)

Then i have come up here and reed lots of audioguru posts. I decided to do some modifications:

1) Added emitter 0.22ohm resistors to each toshiba, 1 unit has now two 2SC5200 output npn-s
---!!--- With emitter resistors, i got little more V drop then without emitter resistors---!!---
2) Replaced R7 to 0.22ohm 5W(its just for testing, later i will add 10W or more).
3) I added tantalums near LM regulators, cuz they are ~10CM away from filter capacitors.
4) Finally i rised output voltage on LM317 from 32.64 to 33.73V, Total voltage across the TL082 is now 38V(under current sinking its 36.8V)
5) The only big differents from the posted schematics is that i have 1.3W BZV85C3v3 zener and 1.3W BZX85C6V2-TAP zener. The schematics states that both zeners must be 0.5W BZV55 type-s
6) Transformer has exactly 30VAC 150VA when there is no load. After the rectification and filtering, there is 40,50 VDC at no load.

After the four first modification i was able to suck 3.4A @ 30V with still 100mV drop

REAL problems:

A) As in schematics, i measure with multimeter 8.5V after the zener diodes... but it must be 9.5V And only If i apply some 3A load at 30V, then i get 9.5V at diodes as in the schematics... but thats wrong right ?
B) Why i cant see -3V on the LM337 output ? I see there -1.8V.. but how than i see across TLo82 pin 4 and 8 38V when there is no current sinking(load ath the output) ?
C)I have 14mF filter bank, and adding more 2200uF panasonic and nichicon capacitors does really help resolving that (100mV drop)
I tried with 22mF and so far i got the best results: 80mV drop at 30V 3.4A output.
That means if i can get there to about 100mF filter capacity, it will be around ~40mV drop... lol  @3A @ 30V Expensive and pointless.
D) According to other ppl modifications i tought i was able to sink 5A but no, i get 2V drop @ 5A of current from 30V to 28V

Is there someone who can help me with these problems and explain why :S And the most important part of reference voltage... why do i have to buy the Exactly the same Diodes used in Schematics ? I tought bigger diode is easier to install and more releable?

Thank you very much for reading this thread.

I have attached 22mF filterbank that archieved best results but is absurdly big and not ergonomic... i use this to power AB class amplifiers.

Dear ppl, my goal is to archive 5A at low ripple 30V, make it symmetrical power supply after and use current limiter for safety reasons.

With regards,
Rene

 

[audioguru2]

Your project has poor voltage regulation when its voltage and current are high because the voltage regulating opamp has a low supply voltage of only 33V. Its maximum output is about 31.5V if you are lucky, the base-emitter voltage of the BD139 could be 1V and the Vbe of the output transistors and their emitter resistor voltage drops could be 2V.

That is why we use opamps that have a maximum allowed supply of 44V so that they do not need a voltage regulator.

 

[tempray]

But, the zeners... what about them ? Is there a point to get 0.5W zeners as showen on the schematic ? Or leave 1.3W different zeners as is ? Its strange that some of the ppl from diyfun blog managed to get it 3A 30V with 30mV drop....

Most of the questions are still unanswered and i myself dont know what to do next.... But if i change that dual-opamp with NE5xxx one that is 40V max supply voltage... would that change regulation in a better way?

 

[liquibyte]

I just don't get why people keep pulling out the parts they do.  Why can't we all cooperate and fix the regulation issues the redesign has and then go from there?  I can understand wanting to put in a couple of regulators for separate outputs in addition to the variable but if you're going to use regulators, you might as well just design a power supply around those and leave this design as is.  I've yet to see a variation on this supply that fixes the issues that it does have and quite possibly are introducing others.  One of the interesting side effects of doing these is that you eventually get several designs floating around the web with nary a real solution in sight and many a newbie asking which one of these is the best.  I'd say that doing a redesign on top of this is futile until the issues it has are fixed or the omissions/additions are ones that fix the issues already present.  Just my 2 cents.

 

[audioguru2]

But, the zeners... what about them ? Is there a point to get 0.5W zeners as showen on the schematic ? Or leave 1.3W different zeners as is?

The BZX55 zener diodes have their voltage spec'd when their current is only 5mA. Then the 6.2V zener diode has a power of only 5mA x 6.2V= 31mW, not anywhere near its maximum of 500mW. The 3.3V zener diode has a power of 16.5mW which is also nowhere near 500mW.

Why don't you look at the datasheets yourself? the BZX85C3V3 zener diode is 3.3V only when its current is 80mA! In this circuit the current is only about 7.8mA.

But if i change that dual-opamp with NE5xxx one that is 40V max supply voltage... would that change regulation in a better way?

The NE5xxx opamp has a slightly higher maximum output voltage than the TL081 but its 33V positive supply voltage from the LM317 is much too low so it will not make much difference in that circuit.

 

[tempray]

I just don't get why people keep pulling out the parts they do.  Why can't we all cooperate and fix the regulation issues the redesign has and then go from there?

My opinion also, i just didn't do a complete research of available PSU schematic problems... and really liked the idea using only 1 chip(dual-opamp).

Audioguru, I will change the zener diodes in the future as stated in schematic, maybe this will tighten some things up when LOAD is present.

The progress is better now: Changed emitter resistors to 150mOhm each on Toshiba output transistors, then changed R7 to 2xparallel 330mOhm resistor = 165mOhm at R7 now.
Current regulation works still and precision is still nice, At 3.4A i get now 60mV drop and voltage seems to be more stable.
Thats with 25mF capacitor filterbank(2x10mF + 2x2.2mF).

If i go lover voltage then i get lower Ripple Voltage also... but then the 2 output transistors heat crazy if i put for example 10V 2A load on PSU

Rene

 

[audioguru2]

The progress is better now: Changed emitter resistors to 150mOhm each on Toshiba output transistors, then changed R7 to 2xparallel 330mOhm resistor = 165mOhm at R7 now.

you reduced the values of the resistors so that the low supply voltage for your voltage amplifier opamp does not cause poor voltage regulation. then if the output transistors are a little different then one will conduct much more current than the other and get too hot. Also the input offset voltage for the current regulating opamp will affect its regulation.

Current regulation works still and precision is still nice, At 3.4A i get now 60mV drop and voltage seems to be more stable.

The regulation should be much better than a 60mV drop. Maybe your wiring has a resistance of 60mV/3.4A= 0.0017 ohms. Use thicker wires.

If i go lower voltage then i get lower Ripple Voltage also

Probably because the low supply voltage for the voltage amplifier opamp causes it to saturate when the output voltage and current are high.

then the 2 output transistors heat crazy if i put for example 10V 2A load on PSU.

It is simple to calculate the dissipation (heating power) in the output transistors then select a heatsink that will cope with it. The heatsink must not be inside an enclosure.

 

[tempray]

Huge heatsink is not problem, got them alot. Audioguru, what happen if i put 3 2N3055 in parallel connection, do i loose output voltage again? There is two right now and they are not so hot like Toshibas were, but i think i will still overheat them if i run 12V ~3A LOAD Over a Night  Toshiba 2SC5200 were hot in seconds and hot-hot in minutes

I have now measured everything, and ended up with upgraded results ;D

Firstofall, changed both Full bridge rectifiers to 10A rectifier. Each channel has now 16mF Capacitor bank(2.2kuF+10KuF+2x2.2KuF)
Ergonimics is still saved.

Then lowered LM317L voltage output from what i had to original 33V. Total supply across TL082 is now 36.42V.

Got to touch some 2N3055 transistors... they are really fast, fast in transferring heat into heatsink (TO-3). Toshibas are now resting, 2N3055 in parallel without EMITTER RESISTORS doing the work now.

With this changes and 1.3W Zeners got 30-40mV drop at 30V 3.44A load. For example at 12V 2.4A load i got voltage drop about 10-15mV.

 

[audioguru2]

The 2SC5200 power transistors in a plastic case are much better than 2N3055 transistors in a TO-3 metal case. They should transfer more heat to the heatsink than the 2N3055 transistors.
The heatsinks in your photos are tiny so no wonder the output transistors get hot. I do not know if you use thermal compound between the transistor and the heatsink to fill in the tiny imperfections and allow much better heat transfer.
Our 5A version of this 30V power supply uses three 2N3055 output transistors in parallel, each with a 0.33 ohm emitter resistor.

Each transistor is different. The Vbe of one transistor might be 0.9V and the Vbe of another might be 0.8V. If they are in parallel without emitter resistors then the one with the lowest Vbe conducts all of the current and burns out while the other transistor does nothing. The value of the emitter resistors (0.33 ohms works well) must be selected to balance worst-case unmatched transistors.

Get rid of the 1.3W zener diodes because they might not even regulate with the low current given to them.
I do not know why you are not using the TLE2141 opamps that we use. Their maximum supply voltage is 44V so the do not need your LM317 low voltage regulator.

 

[tempray]

i will do the modifications as soon as i get hands on different parts.

The heatsink on pictures are used to cool BD139. I use external connectors to connect power transistors on seperate huge radiator.

Btw, one 2N3055 is getting hotter then the other one without emitter transistors Toshibas were equivalent in terms of heat with 150mOhm resistors(silicon pads were used).

This is how big radiator i am going to use to be on the safe side Here, i use nothing, no pads, no paste... and still they are moving hotness quicker then Toshibas

 

[audioguru2]

Without using thermal compound, the transistors are in one city and the heatsink is in another city. Then the heatsink will not be very effective.
If the value for the emitter resistors is too low or the resistors are missing then one transistor will probably heat more than the other.

 

[tempray]

I have made some success, the final result is 60mV drop at 4.84A (30V).

All i did is re-coiling the transformer, not the whole thing... adding some copper to get to 33.8V with no load attached.

Now i get 45V DC, and it drops to 38.9V DC if 6 ohm load is attached to the PSU outputs(30V 4.84A).
0.165OHM is the current sense resistor is used(it doesn't heat up almost at all ), if i raise it, i get much less voltage regulation. But the current limit works very good(Y), 3x2SC5200 toshiba NPN-s with 0.150OHM emitter resistors.

I also have tried the reference voltage across the zener's 8.5V vs 9.5V, there is no difference between them ... it doesn't help if i use 9.5V there(0.5W zener's).

I will post my progress soon as i get some free time.

 

[tempray]

Everything is connected, facing a new wall, cant connect one of the unit's "+" to the second identical unit "-" side to form dual-polarity PSU(series connection).

Transformer has two secondary windings... interesting that i read 30V on multimeter if probe is connected to the first unit output ground and second unit positive output... seems that they share something...

Any tought's someone ?

EDIT, I found the problem and its good to know if u are about to build PSU.

If you buy an ammeter+voltmeter(dual-led light blue/red display) from e-bay and you plan to buy them 2 for split(series=dual-polarity, symmetrical configuration)  then build also 2 identical power supply's to the ammeters/voltmeters from e-bay. I had ONE 5V power supply for TWO ammeters/voltmeters and i couldn't connect power supply boards in series(short-circuits, lots of firework from output pins) because of ammeters having 1 power supply ... So i build 2, each 5V, now totally separated.. everything is in place now and working good.

I will photo the work Tommorow for a preview to give u some motivation to build your own

PSU @ action:


Here are some photos