Banggood PSU Enhancements

[soligen]

I bought a Banggood 0 - 30V PSU kit so that I could at least have something better than an old ATX psu to use.  Looking at the components, the 3A rating is rather generous, so I am looking to beef it up a bit to at least get the 3A safely, and maybe get 5A - 6A. Maybe this is folly and I should just accept that 2A is a more reasonable spec, but where is the fun in that   

Here is the item: http://www.banggood.com/0-30V-2mA-3A-Adjustable-DC-Regulated-Power-Supply-DIY-Kit-p-958308.html

Schematics and parts list are here: http://img.banggood.com/file/products/201505080459530-30Vinstall.pdf

So here are my thoughts for changes. I'm looking for things I may have missed, traps for new players, or any reason this whole idea is shear folly.  I have my eye on a 24VAC 80VA transformer, but if I can beef up the current capabilities, I may go larger.

1) The rectifier diodes have a max rating of 3A (1N5408) so upgrade these to 6A or 8A diodes (recommendations? I'm looking at this http://www.allelectronics.com/make-a-store/item/p600m/6-amp-1000-volt-rectifier/1.html)
2) Q4 (2SD1047) move off board to a good heat sink.  It has a max of 12A, so should be OK at 6 if I can keep it cool. This is an area of concern about beefing this thing up.  Can I dissipate the heat?  I'm thinking of using a CPU cooler with a fan
3) The shunt resistor R7 (0R47 5W) is near its max rating at 3A (~4.5 watts). I can replace this with an aluminum case resistor, and drop the value to get more current limit adjustment range.  0R25 I think should get me in the ball park of 6A adjustment range. Also move off-board.
4)Beef up the traces for the main current path by piggy backing some wire along the traces
5) Add a fuse

Would I need to increase C1 and/or C7? (main input and output caps, respectively). If C1 needs to be doubled (second cap) I think I would need to move them off-board for the space.

I am thinking 16 guage wire for off-board connections in the main current path.

Any thoughts, concerns, tips are welcome.  Even if I stay with a 3A max, I'm thinking a diode and shunt resistor upgrade would be a good thing.

Just for full disclosure, I am also planning to add an Arduino with LCD readouts for set amps/actual amps/volt/watts, and of fan for the heat sink, so I am conservatively allocating about 500mA for internal load. I have already done this on my DC Load project, so I think I know how to do this part already.

Thanks

[Audioguru]

You bought a Chinese copy of a Greek kit that has almost every item overloaded. It was a project at www.electronics-lab.com when I saw it about 12 years ago and many people complained that it was unreliable. Its transformer is overloaded at 24VAC/3A because its maximum current is supposed to be 4.24A instead. The 24V transformer produces maybe 26V or more when it has a low current load then its peak voltage is 36.8V and the rectifiers drop the positive supply to +35.4V. The opamps also use its negative supply of -5.1V so their total supply is 35.4V + 5.1V= 40.5V but the absolute maximum supply for the TL081 opamps is only 36V. The 24V transformer voltage is too low anyway for the project to produce 30VDC at 3A.

I fixed it and I used a 28VAC 4.3A 120VA transformer, TLE2141 opamps rated at 44V and two output transistors each with an emitter resistor. Then TLE2141 opamps do not need many volts for a negative supply so I used -1.3V. Of course the value of C1 is much too low at only 3300uF, I used 12000uF. Unfortunately many of my modified parts will not fit on the Chinese pcb.
Here are my parts list and schematic: 

[electr_peter]

Hello, soligen
I saw this PSU kit in real life, so I can share my insights.

PSU kit works properly only with AC input from transformer. DC input is not enough because negative rail (derived from sine wave) is needed for proper operation.
I am not sure why you would want to go over 3A with this PSU - it barely achieves 3A output in factory condition. I would say it is 2A rated supply because many components overheat at ~2-3A output. Main transistor has to dissipate ~90W worst case - very serious heatsink with fan is required. Kit does not provide any heatsink by default - try not to burn it in testing. Also, two big resistors near main transistor heats up too much at 3A - additional cooling would help as well.

Your ideas about modding PSU are in the right direction. I would focus on increased cooling capacity and PSU case first, then think about better components.

[kolonelkadat]

I know its not the perfect answer for you, BUT if it were me, I'm not sure I would bother. Only because once you start buying transformers, massive great heatsinks, enclosures, components, the price starts to go way way up.

I dont know what country youre in, but for the hundred or so usd that it would cost me to make that one youve linked as good as it should be, id much rather start with something like this bad boy here
http://www.ebay.com/itm/Lambda-Electronics-LES-F-04-OV-0-60V-Digital-1-Output-Adjustable-DC-Power-Supply-/152017186599?hash=item2364ee2f27:g:EpcAAOSwAuNW6Mvb

real top quality 0-60volts up to like 10-15 amps. CC/CV OVP/OCP
https://www.us.tdk-lambda.com/ftp/catalog/le_fall85.pdf

since its a rack mount unit, theres still plenty of tinkering to be done to make it perfect for a bench setup, so you can still have some fun and learn something along the way.

[Kleinstein]

The more useful modifications would be increasing R18 by about a factor of 1.5 and choose a lower voltage transformer, e.g. 18 V and 3.5 A rating: The first change would bring the maximum current down to about 2 A and thus to a value more suitable for the resistor, filter cap and the output transistor. The lower voltage give the TL081 a chance t survive the voltage. Otherwise a an upgrade of the OP is needed and the current should be even lower.

It does not make much sense to use this circuit to much higher power. With emitter resistors to parallel two or more transistors about an extra 0.5-1 V are lost.  With higher power a different design with less heat generation is usually more suitable, e.g. using two or more transformer taps and a different kind of circuit that could give up to something like 2-3 V more output from the same transformer.

When you by a cheap Chinese supply you may end up with a similar circuit in a reasonable case and with a transformer that is slightly to small for the rated current. So it may also be a good idea to reduce the maximum current there. But it can still be cheaper than buying a case, the heatsink and the transformer.

[soligen]

Thank you all for commenting.  Between looking at the original schematic and the one from Audioguru, I am understanding this circuit much better now.  There are just a few things that I still don't quite get from the schematic, if someone could help me understand, I would appreciate it.

Referring to the original Banggood schematic:

What is the purpose of R1?  Looks like it is just a dummy load on the raw DC voltage.  Is this needed for stability, or is it just to discharge C1?

Is R2 just to limit the current to the negative rail? Wouldn't R3 already be limiting the current?

What is the purpose of D10?

What is the purpose of Q1?

Thanks

[Kleinstein]

R1 is just to discharge the main filter cap C1.
R2 is to limit the current used to charge C2. This reduces current spikes. The voltage is rather high anyway.
Q1 is there to turn off the output as long as the negative auxiliary supply is not there, as a kind of delay during turn on.

D10 is there to protect the output transistors from a to negative base-Emitter voltage, that other wise might happen with a large cap as a load and the voltage turned down fast. It can also help to bring down the output it happens to be to high.

[soligen]

Thanks for the help. I think I have a plan now.

I really do appreciate Audioguru's design, and If I were going from scratch, it is what I would do.  I definitely learned from it, but it's too big a deviation from the kit for me right now.

My goal now is to not try to increase it's amperage, but to bring the parts into reasonable spec/reliability.

As Kleinstein suggests, I am going to adjust R18 to limit this to 2A.  I want to try to keep the higher voltages (even if not at full amperage), so I would prefer not to go down to an 18V transformer (but I will if I need to).  Here is my alternate plan:

One of the things not on the schematic is that they added a 7824 regulator to drive a fan (odd since most fans are 12V). Q3, R19, U3, and U1 are all connected to the rectified DC, but none of these really need this much voltage.  I plan to add an output cap to the 7824 and drive all these parts from 24V.  I think this should also be a much more stable voltage, which should be a good thing. R22 will need to be adjusted accordingly.

I think this only leaves U2 to deal with. From Audioguru's numbers it isn't out of spec, but is very borderline.  I am going to experiment and see how close I am, and if an extra diode drop will give it a little more breathing room. If not, then I can either use a lower voltage transformer, or get a single TLE2141 like Audioguru used.

And one more question: Can two of these (isolated) be connected in series for positive and negative rails?

[Kleinstein]

There are quite a few OPs with a 44 V supply rating available for U2.  Already having only a minimal negative supply helps.  Another option is to use something like a 34-35 V zener diode and a resistor to limit the positive supply of U2. The zener diode mainly limits the supply for high mains voltage. Though a +34 V supply will hardly be enough to get 30 v out. So without regulation for the supply for U2, the TL081 specified at 36 V max will hardly get 30 V out while staying in specs.

One could use 2 separate diodes and a separate filter cap for the supply of U2 - so the voltage drop due to ripple will be lower there and thus a higher output (e.g. 2-3 V more) is possible with a given transformer.

One can have two isolated supplies in series to get a +- supply. In this case the diode at the output (e.g. D11) should be strong enough for the full current so more like a 1N540x or even P600.   

[soligen]

I see I had made a mistake in my thinking.  I was thinking U2 was connected to ground, not the negative rail. but looking again, i see I am wrong, so U2 is way out of spec, not borderline.  So, its either use an 18V transformer, or get the 44v op amp.

The 30V is not that important to me, but I would really like at least 24 at the full 2A. Will a 100VA 18V transformer achieve this?  If not, what will?

[Audioguru]

Referring to the original Banggood schematic:

What is the purpose of R1?  Looks like it is just a dummy load on the raw DC voltage.  Is this needed for stability, or is it just to discharge C1?

Its current is so low that it does not do much. It reduces the unloaded DC voltage a little.

Is R2 just to limit the current to the negative rail? Wouldn't R3 already be limiting the current?

R2 limits the high current in C2 and C3. R3 limits the low current In D7.

What is the purpose of D10?

D10 prevents emitter-base reverse voltage breakdown of the driver and output transistors if the project is turned off but a capacitor supplies a positive voltage to the output.

What is the purpose of Q1?

Q1 shorts the output of U2 to ground when the main power is turned off and the negative raw voltage drops before the positive raw voltage. It was used with the original TL081 opamps that had a problem called Opamp Phase Inversion that caused its output to suddenly go as high as it can when its negative supply dropped (then the input voltage becomes too close to its negative supply voltage). With the new opamps it helps prevent a voltage spike when the power is turned off.

[Kleinstein]

With an 18 v transformer you will not reach 24 V under all conditions, not even under light load.
To get 24 V out the raw DC voltage would need to be at least about 27 V (depending on the OP). With some reserve for ripple this would be an 30V unloaded. So a 22 V transformer might just work if line voltage is not to low. To be sure it's more like a 24 transformer is needed, and thus a little to much for a 36 V OP, even with only a small 2 V negative supply.

You can get a way with a little less if U2 gets a separate filter cap and diodes, as this could save on the ripple and get just away with a 22 V transformer.

2 A DC needs a transformer rated for 3.2-3.6 A AC, so the 18 V ,100 VA would be OK from the current.

[sosokruashvili]

Hi everyone, I am glad I found the topic about this psu in here. I purchased this kit and what I see is that everything works well when psu is in constant voltage mode but in current limit mode I have output ripple for above 250 mv peak to peak and it is proportional to load. I checked every part and is OK. I also tried to power current limit opamp with clean current from my bench but same problem. If anyone has the same problem please give me some advises  

See the iamge:

[Kleinstein]

Current limiting work with very low voltages and it thus a little more sensitive to disturbance / hum. The signal looks like some residual ripple. The capacitor C5 could couple some of the raw supply ripple over to the measured current.

The use of C5 could be for decoupling the supply of U1, the OP for the reference circuit. If needed one could have a resistor in the positive supply of U1 and C5 only after this to reduce ripple current.

Is there a fuse between the transformer and the rectifier ? This would reduce current peaks and thus ripple component due the filter caps ESR. This could be important with a low impedance (e.g. ring core) transformer.

Which circuit is used ? The "original" one gives way to much supply to U3 and thus U3 has every right to behave funny or release some of its magic smoke. So the modification with an extra zener (D13) diode in series is a good idea. However U2 still can have a problem !.

[sosokruashvili]

Fixed   Thank you Kleinstei  , for your suggestion. Yes, the main source of this ripple is current ripple generated on PCB ground trace. I just made wire connection between R17 ground side pin to current sense resistor pin(low side) and ripplce is almost gone. It makes sense, as this PCB has very thin traces for rated kind of loads.
This is the mod image:

[soligen]

Your image didn't come through.

I had ripple when using a constant current load to test, but I don't know if the cause is my DIY constant current load or the PSU (I assumed the load).  More capacitance between PSU and load fixed it.  Thanks for posting your solution.  I will add the mod to mine as well.

[soligen]

Just to round out this thread for future people who find it, here is the list of all the mods I made.  Target max is 2A at 15V to keep things in spec.

Used a 18VAC 100VA transformer
Replaced C1 with 2 4700uF 50V capacitors (Nichicon) that I had salvaged
R22 changed to 3K (2 1K5 in series) due to reduced rectified voltage
R18 changed to 94k (2 47k in series) This gives me just over 2A
P1 upgraded to 10 turn pot
D11 upgraded to 1N5402
Added .1uF bypass cap on all op amps
Fed power to U2 via a shotky 1n5819 diode and Added 330uF 50V bypass cap.  This is to reduce ripple
Added 10k resistor between P1 and U1 - this reduces the voltage range to just over 15V
Added a 20 ga wire between R7 and R17 to beef up the trace (per sosokruashvili above)
Moved Q4 off board - mounted on old CPU heat sink with fan.  Fan also blows accross the board

I also added a second board with an arduino to monitor and display volts, amps, current limit, and watts.  This board contains:

The new C1 capacitors
Inexpensive ebay buck/boost converter to get about 9V for the fan and arduino VIN
Arduino Nano
ADS1115 16 bit ADC
16 x 2 LCD with I2C backpack

The 4 channels of the ADC are connected to V-out, Both sides of R7 (current shunt) and the center tap of P2 (for the current limit setting).  Display calibration is done in software.

This project has been on the shelf for a while, but I hope to finish it up in the next couple weeks.

[soligen]

I am hoping one of you experienced guys are still following this thread.

I have a question about what kind of ripple/noise is generally considered acceptable.  Using my scope at 100MHz bandwidth I get some high frequency spikes.  They are not uniformly distributed and when they happen it looks like 1 - 3 waves of ringing, then it goes back to normal noise.  Each "ring" can be 100 - 200 mV peak to peak with a frequency from 10 to 80 Mhz.  Is it reasonable to be seeing these?

I also have had some stability issues, but I think I have that solved.I would get high frequency ripple that would come and go in a repeatable pattern.  Things got worse with more load.  It made the scope trace look a solid line that got thicker and thinner.  When I zoomed in on the thick part it was comprised a of a 1Mhz sine-ish looking wave.  I didn't nee the issue until I tested with a resistor because I had a 470u cap on my electronic load, which masked this issue, so I figured I need more output capacitance.  Not wanting to go too high and foil the current limiting, I reduced the extra capacitance to a 10u ceramic.  Things looked good and I think this is an acceptable level of extra capacitance.  I had put the cap on the binding posts to test, but when I soldered it onto the circuit the instability returned (although a little bit improved).  The only thing I could come up with was it having a lower ESR being soldered in, so I added a 0R62 resistor in series with the cap (lowest I had on hand).  This did the trick.  I also added a .1u output cap, not in series with the resistor.

Another change I made, although it didn't fix anything, is a 10u cap across the outside pins of P2.  Since the voltage reference from U1 will fluctuate in relations to U3's ground as current draw varies, my thought was to slow down that change.  I left the change in place because smoothing fluctuations across P2 seems like a good idea.  I will note that at one time I had some current limit ripple similar to what  sosokruashvili showed.  I dont know for sure if this has anything to do with it, but I don't see any of that kind of ripple now,

I considered an LC low pass filter to help with the 10 - 80 Mhz spikes, but my initial try was inconclusive, so I thought I would ask here about this to see if it is something I should even worry about.

Thanks for all the help I recieved on tnis thread. I know this thread spanned a long time.  I appreciate the learning it has provided,

[Audioguru]

I wonder why Banggood did not test the original faulty circuit they copied and also did not test the output of their kit.

[Audioguru]

It looks like Banggood is finished selling this faulty power supply kit (its price is reduced) and are replacing it with a similar kit but with its output voltage and current reduced to 28V/2A so it does not damage the opamps and its transistors do not overheat. Amazon and ebay are also selling the same kit. The new kit might have an LCD meter on it.

[soligen]

Well, I found Dave's video an evaluating power supplies, and, of course, I was doing it wrong.  I dont have the right probes to really do it right, but I was able to figure out that most of the noise spiking is common mode noise.

The best I could do to get readings is set bandwidth to 20 Mhz, earth reference the PSU, and use 2 10x probes with ground lead removed with a-b math function.  I got about 9 mV peak to peak not loaded and 12 loaded to 1.5 amps. Everything seems very stable, not even a hint any more of the oscillations like what sosokruashvili was getting.

I am going to declare this good enough for this PSU.  Considering I am running a fan, powered by a little SMPS inside the case, the design could probably do better if I used a linear supply for the arduino and convection cooling.  Attached is a schematic with my mods marked up.  I am compensating in software for the current limit reference voltage varying with load.

If this isn't going to be sold anymore, not sure others will find this useful, but I sure learned a lot.

Thanks to all those who contributed to this thread.

[soligen]

I ended up having to add a separate voltage reference for my current limit display to behave properly.  I removed R18 and added in it's place a TL431 voltage reference at 2.5V  I used a 3.7k + 3.3k resistor in series to feed to current from the 10ish voltage reference, then used 3 4.7k resistors in series to drop the voltage to approx 1V going into P2.

All is working good now.  The arduino code is at https://github.com/soligen2010/PowerSupplyMonitor

[Acanti]

Hello!
Why not doing it the hard way by adding a separate power supply for the op amps? a 2x10volts filtered and regulated indecently by a different transfo windings? that can eliminate problems to the opamps and simplify the power winding choice, dont you?

[Kleinstein]

This old Bangood circuit is using an output stage in emitter follower configuration. Thus is needs at least one OP to have the full voltage swing at the output. So for a 30 V output things get a little tricky.

Using a separate, floating +-10 V or similar supply for the OPs makes a completely different circuit. This is the classical version with the output stage in common emitter configuration. The circuit has different properties though, usually a high output capacitance. So it would be building a different regulator from scratch. It has some benefits when digital control is wanted, as the supply for the OPs can be combined with the control part.

[Zero999]

Some interesting op-amps have been suggested to replace the under-rated TL081: the MC34071 & TLE2141. There's nothing wrong with them but there are other options. It just needs to be able to handle a total supply voltage of 44V.

How about trying the OP07 which will have a much lower offset voltage, so much so the trimming circuit can probably be eliminated, but will be slower so might need a larger capacitance on the output to ensure good transient response.

Another possible option is the NE5534 which will be faster than the OP07 but will have a much greater offset and will need a compensation capacitor and a different offset trimming circuit.

The OP27 will give good transient response and offset but it's more expensive than aforementioned. It could be used for U2, with cheaper op-amps for U1 & U3.