Korad KA3005P power supply faulty/repair

[cavlovic]

I got this power supply a couple of days ago and it worked well until today. I tested it by charging almost empty 72Ah car leadacid battery at 14V 4-5A, under supervision a few times. It worked all well.

Today when I try to do the same thing, once output is turned ON voltage readings are way off of actual voltage. Output on Korad reads 6-7 volts (14V set voltage) and my multimeter reads almost 13V (which is fine as battery is not fully charged).

I also measured current and it reads fine on both Korad and my multimeter. Only voltage reading is way off.

I believe it needs to be calibrated, but not sure how it got so messed up in a few days. I tried to find any information on calibration all day long, but had no luck. I also dissembled the unit but didn't find any obvious pots for adjustments, which led me to believe that it should be done through software or some kind of key combination. Does anyone know how to do this? Manual is all in chinese and I doubt it contains information about calibration.

Also, software which comes with this unit is complete garbage. It's half scrambled and unreadable and it also doesn't function at all. I need to change location to some automatic log location, but I can't seem to be able to do it, so I can only start unit from software but not adjust voltages and current.

Is there any other software that can be used with this Korad KA3005P power supply?

Thank you!

[iloveelectronics]

There was a document posted not too long ago here. Try a search for it.

[Bored@Work]

I believe it needs to be calibrated,

No, it doesn't. It needs to be repaired.

[wraper]

I believe it needs to be calibrated,

No, it doesn't. It needs to be repaired.

+1

[Carrington]

I believe it needs to be calibrated,

No, it doesn't. It needs to be repaired.

+1

[cavlovic]

There was a document posted not too long ago here. Try a search for it.

I know I saw something regarding its' calibration, but just couldn't find it. I also tried google search "site:eevblog.com" and didn't have much luck either. I will keep digging

I believe it needs to be calibrated,

No, it doesn't. It needs to be repaired.

Thank you for the input! I will try to repair it tomorrow and report.

[alm]

Charging a battery without series diode is risky, especially if you turn of the power supply with the battery connected. Many power supplies get really unhappy when they have to sink current. I would expect that to kill the pass transistor or its driving circuit, not the voltage monitoring however.

If this problem suddenly appeared after the battery charging, then I agree with B@W that it's unlikely to be calibration and more likely to be a defect.

[van-c]

I agree that the OP's power supply needs to be sent back for repair.

However, for anyone interested, the following is a calibration procedure I received from SRA Soldering Products.  They obtained it from the manufacturer and the English is a little hard to follow in places, but the procedure does seem to work for calibrating the 0.0V and 30.0V endpoints.  (I didn't try changing the current calibration because on my unit it seemed okay.)  I had to play around with the voltage calibration, though, because getting good agreement at the 0.0V and 30.0V endpoints left the interior points a little inaccurate:

Calibration instrument: benchtop multimeter with 6.5 digits display, 5A electronic load with milliampere display.

Calibration condition: Preheat for 2 minutes after switch on .
Being into calibration mode, press continuously button M4, and switch on at the same time until current and voltage displays blink. At this time, C.V indication is on, which means the power supply is in the mode of zero calibration.
1.      Voltage Zero Calibration: connect the positive pole and negative pole of the multimeter leads into the output terminals respectively. Then, watch the voltmeter and adjust the knob to make the multimeter in the range 0v – 5mv. After that, the zero calibration is over. Press M1 to save the calibration value.
2.      Current Zero Calibration: press the button VOLTAGE/CURRENT, and the current display of the power supply blinks. Connect amperemeter to adjust the current value in the range 0MA – 1MA. Press the button M1 to save the calibration value.
3.      Voltage full-range calibration?adjust voltmeter in the shelf of voltage test. Press M4 and then C.C indication lights on, which means full-range calibration. And then the current display blinks, turn the knob to adjust the voltmeter between 30.01 – 30.02. Press the button M1 to save the calibration value.
4.      Current full-range calibration: press the button VOLTAGE/CURRENT, the current display of the power supply blinks, which means it is in the mode of current calibration. At this time, connect the amperemeter(or electronic load), and then adjust the current value to 5.000A±5MA. After that, press the button M1 to save the calibration value.
5.      Switch off, and then restart the power supply. The calibration ends.

Hope this helps,

--Van

[cavlovic]

Just a small update.

Today I did go through calibration process, and it did help with display voltages, but is still not fixed. I tore the hole unit apart to get a feel of it and did some visual checks. Nothing wrong upon inspection. I will have to dig a lot deeper with my multimeter. Too bad I don't have a scope, as it would help a lot.

Regarding software: do you know any third party software that can be used with this unit? Mine is too scrambled:

This is the version:

[mcinque]

Just a small update. Mine is too scrambled:

Sometimes who write a software use a local charset for the GUI and probably you're missing it. Maybe chinese?

[cavlovic]

Just a small update. Mine is too scrambled:

Sometimes who write a software use a local charset for the GUI and probably you're missing it. Maybe chinese?

Charset should be included with windows, but they maybe programmed it in previous versions and some charcode labels have changed in windows 7, that's why I'm getting those characters. I will install software on windows XP and check if it's the same.

On the other note, I poked a little today through my power supply. They went through trouble of cleaning flux after soldering, but left a mess after cleaning:


I cleaned all boards, and this should be the board, located in front part of PSU, where all the logic is happening:


Current comes to the most right bottom connector from PCB on back of the power supply, where it's been rectified and smoothed, and exits to the front terminals from J2 connector.

I checked all resistors with my mutimeter, and some of them, just up from those 4 caps, read much lower then their written values. Some 1Mohm read 10K, some 10K read 6K etc. Some of them seem fine. I believe those cause the problem with reading proper values, but just want to make sure with more experienced guys before I start desoldering and replacing those.

For input protection there is a 1N5408 diode, 1000v 3A, which would not stop current from car battery when power supply is off. I guess I had it connected before turning output of the power supply ON.


Here is back PCB with rectifier and pass transistors:


Am I onto anything, or way off my path?

Thank you all!

[alm]

Charset should be included with windows, but they maybe programmed it in previous versions and some charcode labels have changed in windows 7, that's why I'm getting those characters. I will install software on windows XP and check if it's the same.

Locale settings may make a difference. It looks like the garbled labels are easy enough to guess, so it should be usable, though ugly. Third-party software is unlikely to support this power supply which is not very popular or emulating an established standard. Your choices are either using whatever crap the manufacturer gives you or writing it yourself, unless another user already did the latter. If their software is written in LABView, then they probably shipped LV drivers. If you own and are skilled in LV, then writing a better GUI would be fairly easy.

I checked all resistors with my mutimeter, and some of them, just up from those 4 caps, read much lower then their written values. Some 1Mohm read 10K, some 10K read 6K etc. Some of them seem fine. I believe those cause the problem with reading proper values, but just want to make sure with more experienced guys before I start desoldering and replacing those.

Most likely other resistors in parallel. A 1 Mohm resistor is unlikely to go down in resistance to exactly 10 kohm. If you want to be sure, desolder and measure its resistance. I would rather figure out the circuit and measure voltages. Presumably a resistive voltage divider (possibly switchable if it has multiple ranges, those 74HC part might be analog switches), possibly a buffer amplifier (one of those SO-8 packages?) and then an ADC (probably built into that MCU with its part number sanded off). Ideally you would have values from a known good power supply to compare. I would be more suspicious of active devices, since resistors are less likely to be damaged by minor overloads.

For input protection there is a 1N5408 diode, 1000v 3A, which would not stop current from car battery when power supply is off. I guess I had it connected before turning output of the power supply ON.

That's most likely an anti-parallel diode across the input, that does not protect against current flowing from the battery to the power supply. I was talking about a series diode between power supply and battery.

[mcinque]

Charset should be included with windows

Afaik, european, u.s. or australian Windows does not have asians charsets installed by default. Maybe I'm wrong.

Locale settings may make a difference.

I agree.

[alm]

I think there was another piece of crappy software from one of the Chinese test equipment manufacturers (Owon? Hantek? Rigol?) that had a similar issue.

Edit: this may have been the post I was thinking about.

[mos6502]

Is there another PCB in there with opamps? Because it doesn't look like there's analog regulation circuits on there.

Could it be that the KA3005 uses the microcontroller to do the actual regulation?

I looked at the Korad website, and the specs say that the transient response is 100ms. Yes, milliseconds! That's around 1000 times slower than a proper linear supply. Unless that's an error, that would indeed mean that the micro does the actual regulation. Which would be insane, and horrible, and make this supply all but useless except maybe for battery charging.

I watched Dave's review and the steps in the voltage response would also point towards that.

Anyone have a schematic for this piece of ... kit?

[alm]

How about U5-U7 (top right in the second picture)? Can't read the markings, but they look like they might be op-amps to me. Three dual op-amps is plenty for a linear bench supply with an analog regulation loop.

[mos6502]

Right. But they could also be level shifters/buffers for the R2R DAC. IMHO the 100ms transient response (no opamp is THAT slow) and the steps in the voltage after turn-on (watch Dave's video) speak a clear language.

Unfortunately Dave didn't do a transient test. Anyone who owns one of these want to do one? All you need to do is to connect a big resistor to the output and switch it with a MOSFET at ~100 Hz and trace the output voltage on a scope.

I'd love to see a schematic for this thing. These are the cheapest 30V/5A PSUs out there. You can get a KA3005D for 65 Eurobucks here in Germany. But it looks like you'd have to do some major hacking and add your own regulation circuit.

[alm]

I agree with you about the slow transient response, suspicious turn-on behavior and that putting an MCU in the regulation loop would be a bad idea, but without a schematic or someone reverse-engineering the design there is no way to be certain. The slow transient response could also be a really over damped loop response (at least it won't oscillate), for example.

[Electro Fan]

Anyone who owns one of these want to do one? All you need to do is to connect a big resistor to the output and switch it with a MOSFET at ~100 Hz and trace the output voltage on a scope.

+1 on this request/suggestion

[cavlovic]

Is there another PCB in there with opamps? Because it doesn't look like there's analog regulation circuits on there.

Could it be that the KA3005 uses the microcontroller to do the actual regulation?

I looked at the Korad website, and the specs say that the transient response is 100ms. Yes, milliseconds! That's around 1000 times slower than a proper linear supply. Unless that's an error, that would indeed mean that the micro does the actual regulation. Which would be insane, and horrible, and make this supply all but useless except maybe for battery charging.

I watched Dave's review and the steps in the voltage response would also point towards that.

Anyone have a schematic for this piece of ... kit?

Nope, this is the only board with opamps. I would also like to know if anyone has schematics for this power supply. I couldn't find it anywhere.

Once again, the board:


Parts:
U5-U7: dual opamps TL082C:


U12: darlington transistor array ULN2003A:


U13: ATMEL364:look previous at picture with your head or monitor upside down  .

When unit is powered on and my LED lighting pulling 1.4amps at 12V (12V set to power supply), voltage reading jumps around at the unit, but multimeter shows exactly 12v. Current readings on the power supply are fine.

When I put my finger on R10-R17 (right above 4 caps in first picture) and thus heating them up and increasing resistance, voltage drops on power supply, with it the current. From 12V and 1.4amps I managed to bring it down to 11V and sub 1amp. This may tell you a little more about regulation loop.

Any further help is very much appreciated.

Thank you all,
Balsa

[alm]

Like I said, I would trace the voltage measurement connections and go from there. There are likely thin traces coming from J2 (or the input header) going via some resistors to U10. See if there are any active components on this path, check resistors and check if the U10 input gets a stable DC voltage (probably varying between 0-5 V or so, scaling linearly with the output signal).

The fact that voltage regulation appears to be independent of voltage measurement argues against the MCU in the regulation loop in my opinion. Why would you waste two ADC inputs on sensing the output voltage?

[cavlovic]

Like I said, I would trace the voltage measurement connections and go from there. There are likely thin traces coming from J2 (or the input header) going via some resistors to U10. See if there are any active components on this path, check resistors and check if the U10 input gets a stable DC voltage (probably varying between 0-5 V or so, scaling linearly with the output signal).

The fact that voltage regulation appears to be independent of voltage measurement argues against the MCU in the regulation loop in my opinion. Why would you waste two ADC inputs on sensing the output voltage?

I did listen to your advice and traced J2 to U10, through a couple of resistors and caps, but there are no active components in the path.

Then I spent 4 additional hours trying to figure out what is wrong (it was PITA to measure while unit is powered on since there is very little space, even though I have a set of really thin probes), and I may be onto something (my 4 hours equate to maybe a couple of minutes to some of you here, so please bare with me ).

Voltage set and detection part:
Both outputs of U5 opamp are used for voltage regulation and display. Output 1 (PIN 1) goes from 0.04v (0v set output voltage) to 2.84v (31v set output voltage) and this output is only dependent of set voltage, regardless of load. Output 2 of U5 reads 0.7V to 1.16V, also jumping all over the place, and this opamp output is dependent of the load, and this should be the one that causes voltage reading problems. Output 2 of U6 also reads 0.04v to 2.84v and this one is not dependent of the load attached.

Current set and detection part:
Output 1 of the U6 reads 0.04v (0amps set current) to 2.56v (5.1amps set current). Same goes for output 1 of the U7. Both of them are not dependent of the load and represent just set current values. Output 2 of U7 reads -3v to 2.71v and this one is load dependent and probably is used for displaying actual current flow. This reading started acting up today also, so both voltage and current readings are off, even though they are spot on to set values.

I believe that both output for voltage and current when load is present should read almost identical to their set values (0.04v-2.84v for voltage and 0.04v-2.56v for current) if opamps didn't fail.

I guess those opamps need replacement, and they are causing  all the trouble. I also may be off my path here, so any input is much appreciated.

Also, if anyone happens to have Korad KA3005P or D laying around, measuring output 2 (PIN 7) of both U5 and U7 under load, from minimum to maximum voltage and current would help me a lot.

Thank you all,
Balsa

[KORAD TECHNOLOGY]

I got this power supply a couple of days ago and it worked well until today. I tested it by charging almost empty 72Ah car leadacid battery at 14V 4-5A, under supervision a few times. It worked all well.

Today when I try to do the same thing, once output is turned ON voltage readings are way off of actual voltage. Output on Korad reads 6-7 volts (14V set voltage) and my multimeter reads almost 13V (which is fine as battery is not fully charged).

I also measured current and it reads fine on both Korad and my multimeter. Only voltage reading is way off.

I believe it needs to be calibrated, but not sure how it got so messed up in a few days. I tried to find any information on calibration all day long, but had no luck. I also dissembled the unit but didn't find any obvious pots for adjustments, which led me to believe that it should be done through software or some kind of key combination. Does anyone know how to do this? Manual is all in chinese and I doubt it contains information about calibration.

Also, software which comes with this unit is complete garbage. It's half scrambled and unreadable and it also doesn't function at all. I need to change location to some automatic log location, but I can't seem to be able to do it, so I can only start unit from software but not adjust voltages and current.

Is there any other software that can be used with this Korad KA3005P power supply?

Thank you!

This is Krissy from KORAD TECHNOLOGY. Would you please kindly send me your e-mail address? I want to solve it soonest. Thank you in advance!

[cavlovic]

I got this power supply a couple of days ago and it worked well until today. I tested it by charging almost empty 72Ah car leadacid battery at 14V 4-5A, under supervision a few times. It worked all well.

Today when I try to do the same thing, once output is turned ON voltage readings are way off of actual voltage. Output on Korad reads 6-7 volts (14V set voltage) and my multimeter reads almost 13V (which is fine as battery is not fully charged).

I also measured current and it reads fine on both Korad and my multimeter. Only voltage reading is way off.

I believe it needs to be calibrated, but not sure how it got so messed up in a few days. I tried to find any information on calibration all day long, but had no luck. I also dissembled the unit but didn't find any obvious pots for adjustments, which led me to believe that it should be done through software or some kind of key combination. Does anyone know how to do this? Manual is all in chinese and I doubt it contains information about calibration.

Also, software which comes with this unit is complete garbage. It's half scrambled and unreadable and it also doesn't function at all. I need to change location to some automatic log location, but I can't seem to be able to do it, so I can only start unit from software but not adjust voltages and current.

Is there any other software that can be used with this Korad KA3005P power supply?

Thank you!

This is Krissy from KORAD TECHNOLOGY. Would you please kindly send me your e-mail address? I want to solve it soonest. Thank you in advance!

PM sent.

[mos6502]

This is Krissy from KORAD TECHNOLOGY. Would you please kindly send me your e-mail address? I want to solve it soonest. Thank you in advance!

What is your policy regarding schematics? Can you give them out or not?

[cavlovic]

This is Krissy from KORAD TECHNOLOGY. Would you please kindly send me your e-mail address? I want to solve it soonest. Thank you in advance!

What is your policy regarding schematics? Can you give them out or not?

She was very prompt with email and very kind. She asked for pictures of the display fault, but in return will provide and ship new board to me. I just made a short video which I will share here also, when done uploading.

The support is really good and means a lot to me, especially for future purchases.

They also sent me latest version of software in English (currently downloading - will report later), without me even asking! I guess they took time to read through all posts. I will share it with others once I check with them that it is OK thing to do.

I'm very satisfied with response so far!

[Flump]

This is Krissy from KORAD TECHNOLOGY. Would you please kindly send me your e-mail address? I want to solve it soonest. Thank you in advance!

This is really nice to hear, I need to get a new power supply soon.
I may well consider a Korad unit now.

big thumbs up from me  O0

[cavlovic]

This is Krissy from KORAD TECHNOLOGY. Would you please kindly send me your e-mail address? I want to solve it soonest. Thank you in advance!

This is really nice to hear, I need to get a new power supply soon.
I may well consider a Korad unit now.

big thumbs up from me  O0

Yes, very nice to hear they take care of customers after sales, even though they sell cheap power supplies.

In last email I asked her about some firmware upgrades, that could improve usability, and will let you all know about the response. I also asked her about schematics and publishing of it.

[van-c]

Back in June I purchased a KA3005P from SRA Soldering.  It has been a reliable power supply that is easily calibrated to within Korad's published specs using the procedure I posted in reply #7.  However, there is a bit of an issue that I've seen on mine (and observed on others in videos) that causes the outputs to produce a very low negative voltage when the output switch (not the power switch) is set to the OFF position.  With no load on the output terminals I measure between -107 mV and -115 mV until the output switch is pressed ON, at which point the measured output becomes about equal to the displayed output voltage.

I've discussed this with SRA soldering, who have swapped out an entire unit for me, and recently sent a replacement board, to attempt to fix the issue.  After I replaced the microcontroller board, the no-load output changed slightly, from -108 mV to -115 mV.  But basically, the issue still remains.  Under loaded conditions, I measured the following voltages : 2200 Ohms: -73 mV; 1000 Ohms: -53 mV; 100 Ohms: -10 mV.  This is obviously not a serious problem, but I would expect a properly designed switched output to produce zero volts across the output terminals whenever the output switch is OFF.

Although the issue is still present, I must say that once again, SRA Soldering has done a superb job of customer support.  They paid for shipping my original unit back to them, sent me a new replacement unit without charge, and finally, sent a replacement board all in an effort to fix the issue.  I told them I am willing to just consider this issue a quirk of this particular supply, but they have continued to show an interest in resolving it.

Now that Korad has popped up on this thread, I thought they may want to take the opportunity to chime in on this.  Is Korad aware that the KA3005P produces over 100mV on the output terminals even when the output switch is turned off?  Is this considered normal behavior for this particular model?

--Van

[pomonabill221]

I did rev eng the power supply and have drawn (hand drawn) the schematic.
I can scan it and create a pdf for your dining and dancing pleasure, but I want to make sure that I won't get in trouble with Korad for publishing it here.
If Korad reads this please let me know if I can post, OR better yet, post the official schematic for us!

[AllanMN]

I'd like to see the schematic. And fast. Because I have to buy a supply soon. I was about decided on this one, but using the microcontroller as the feedback regulation loop.... the transient response. Lack of repairability that entails. Sort of. Have me looking elsewhere.

[pomonabill221]

I'd like to see the schematic. And fast. Because I have to buy a supply soon. I was about decided on this one, but using the microcontroller as the feedback regulation loop.... the transient response. Lack of repairability that entails. Sort of. Have me looking elsewhere.

Well I will scan and upload my hand drawn schematic, but for right now I can tell you that the uP is NOT used in the feedback loop.
All it is used for is setting the R/2R latches which sets the reference levels for voltage and current.
There ARE opamps that actually compare the V/I and control the output stage.
Give me a few minutes and I will scan the drawings.....
Bill

[pomonabill221]

OK here we go!
Notice that this is a PARTIAL schematic where I tried to draw ONLY the important parts... I left out decoupling (not that there was alot) and I tried to note the connectors.

My notations are:

J1-1 means connector 1, pin 1

Components with only a number are the component number on the board, like a resistor with a 7 means R7 and the value like 5.1 is 5.1 ohms.  A diode/transistor should be obvious.

An op amp I have noted the component number (like U6) with the number inside the symbol and the pin numbers are on the outside of the symbol.

I did mod my supply to try and reduce the turn on overshoot by paralleling a 1uF cap across the cap labled 20 on page 3.
  This effectively slows down the Vref turn on ramp.  It works ok but the response of the loop seems to be rather slow on turn on possibly due to an amp being at it's rail and the recovery time of the amp is slow, so the output peaks before the loop can "catch it".  IE, an opamp is saturated and typically this results in long recovery times.
A larger cap makes things worse and a 1uF seemed to be the best value.

What I am going to try, rather than use a relay on the output, is to delay the turn on of the shift registers in the Vref by disabling the registers for a short time.  What I think is happening is that when the supply is turned on, the uP load the registers, but the outputs are active so the Vref is not stable because of clocking glitches on their outputs, and the voltage loop is trying to follow the stepping/glitching.

There is another trick that I have used in the past to slow down turn on, but not affect operating response time, and I will post my findings when I do (try) it.

If you have questions about my drawings... ASK!
Remember, this took HOURS to do because of all the plate throughs/traces under components.  I have reversed engineered many things before, so I have experience doing this.... hence if you don't like my schematics... DO IT YOURSELF!
It ain't real easy!
Enjoy!

[pomonabill221]

Another thing...
Note that the POWER -V is NOT the same as the CONTROL -V!!!!!!

The control -V is connected to the POWER +V  this is NOT a mistake!!!  That is why I used different symbols for the two supply returns.
Make SURE that you pay attention when metering/scoping.

[commongrounder]

@pomonabill221    Thanks so much for your efforts!  This is very useful information for us Korad KA3005 owners.  Will look forward to hearing about further results of your experimenting.  It seems strange that the firmware could not be written to load the registers before turning on the outputs, to give things time to settle.  Probably is some other tradeoff.

[pomonabill221]

Yes, they could have controlled the latch's output enable pin rather than just enable it all the time...
I will keep updates coming as I find out more about the turn on overshoot... I still believe it is due to an op amp sitting at it's rail when off, and the recovery time when turned on, as these overshoots aren't present when changing voltage while the supply is on.

I really hope the schematics help others, and others share what they find as well.  I like my supply and want to improve on a few things.

It ain't an HP/Agilent, but it didn't cost like on either!

[AllanMN]

@pomonabill221 Yes, thanks indeed. I ordered one.

[pomonabill221]

Now I just need to find out why the software doesn't work on win7 64 bit.
It will start and kind of controls the supply but keep complaining about labview which I do not have.
Why does it need labview and where do I get it (free)?
also, I cannot change the path in the box where it is indicating d:\koradka3005p save data... I don't have a D drive (well I do but it is a cd drive).
VERY strange!

[commongrounder]

@pomonabill221:  I have the D version (non-USB/serial) of the power supply, so I can't check this for you, but I do know that some interface software uses the National Instruments drivers to talk to hardware.  My Rigol scope would not communicate properly with the capture software until I installed the driver specific to that model.  There is a Third Party driver repository on the NI web site.  I don't believe you need to install the Labview software itself (many $$$$).  The problem with the default save directory sounds like lousy software coding to me....
Hopefully, someone else using Win7 64bit can relate their experience with the Korad software/hardware.

[pomonabill221]

I know what you mean about the Rigol and the NI drivers... I went through that and got the Rigol to talk just fine.
The Korad is a different story and is frustrating to use, AND is pretty lousy as well.
I just wanted to try the new version, and it sucks! (just like others have found).
The graphics are pretty, but also pretty useless!
Thanks though....

[AllanMN]

The KA3005P arrived today from SRA Soldering. QUick.

Some testing... and.... the negative isn't floating. It surely IS suppose to float on this supply, right? Because if I connect negative to ground it draws over the CC setting (i.e. set to 0.1 it draws 0.2) with nothing on the positive. I noticed it when I tried to scope the output with scope ground to negative.

Allan

[TopLoser]

I've recently heard the phrase 'Single use only' being applied to these power supplies. Korad, Tenma, whatever the name that's stuck on them.

[AllanMN]

'Single use only.... lol.

Yes, I think of them as more of a kit/project. You start with what you get and go from there. So, this thread, and a partial schematic makes for a just more interesting project. I guess.

Allan

[baljemmett]

Some testing... and.... the negative isn't floating. It surely IS suppose to float on this supply, right? Because if I connect negative to ground it draws over the CC setting (i.e. set to 0.1 it draws 0.2) with nothing on the positive. I noticed it when I tried to scope the output with scope ground to negative.

Now, that rings a bell; something about a chassis screw in a really stupid position...  *rummages*

Ah, unfortunately that was on a Mastech supply, rather than one of the Korad threads -- so it's not an already-documented issue with this model.  But just in case it's of any use, https://www.eevblog.com/forum/reviews/bench-supply-ground/ has the gory details.

[pomonabill221]

The KA3005P arrived today from SRA Soldering. QUick.

Some testing... and.... the negative isn't floating. It surely IS suppose to float on this supply, right? Because if I connect negative to ground it draws over the CC setting (i.e. set to 0.1 it draws 0.2) with nothing on the positive. I noticed it when I tried to scope the output with scope ground to negative.

Allan

I checked both the positive output and the negative relative to chassis ground and they ARE floating.... Your supply may have a problem.

[tony3d]

The KA3005P arrived today from SRA Soldering. QUick.

Some testing... and.... the negative isn't floating. It surely IS suppose to float on this supply, right? Because if I connect negative to ground it draws over the CC setting (i.e. set to 0.1 it draws 0.2) with nothing on the positive. I noticed it when I tried to scope the output with scope ground to negative.

Allan

I checked both the positive output and the negative relative to chassis ground and they ARE floating.... Your supply may have a problem.

Yup! Mines floating as well. Call SRA Solder, speak with Doug, and tell him you want another one sent out! They are really good to deal with.

[AllanMN]

Thanks Tony. I'm in touch with Doug. But this is perhaps an easy fix. I think something is touching ground and since the supply floats, that signal on ground is pushing plus and minus around from there. Here's what I see on the scope on the positive output lug (attached image):

5ms/div, 5v/div vertical, 0v is at top of scope screen so you're looking at -10v to -40v about.

I think that's the unregulated plus rail (about -43v) with pulses to the regulated voltage (set at 10, so -10v). It's inverted since I'm looking at it reversed if that makes sense (plus rail is grounded so positive out is pushed negative).

There isn't a lot that's at ground in the supply. I was about the disassemble the front panel assembly but it's plastic (not grounded metal). There's just the ground from the front panel lug coming though.

My question is where would such a signal be on the schematic, and then, circuit board (Pomona Bill). There's where to look for a short to ground I think. Careful inspection didn't reveal any solder bridges or physical damage.

An aside, learning to repair this supply would seem to be a handy trick for the future.

Allan

[tony3d]

Thanks Tony. I'm in touch with Doug. But this is perhaps an easy fix. I think something is touching ground and since the supply floats, that signal on ground is pushing plus and minus around from there. Here's what I see on the scope on the positive output lug (attached image):

5ms/div, 5v/div vertical, 0v is at top of scope screen so you're looking at -10v to -40v about.

I think that's the unregulated plus rail (about -43v) with pulses to the regulated voltage (set at 10, so -10v). It's inverted since I'm looking at it reversed if that makes sense (plus rail is grounded so positive out is pushed negative).

There isn't a lot that's at ground in the supply. I was about the disassemble the front panel assembly but it's plastic (not grounded metal). There's just the ground from the front panel lug coming though.

My question is where would such a signal be on the schematic, and then, circuit board (Pomona Bill). There's where to look for a short to ground I think. Careful inspection didn't reveal any solder bridges or physical damage.

An aside, learning to repair this supply would seem to be a handy trick for the future.

Allan

Let me know how you make out. When the supply is working properly, I really like it. I like the size. Fits perfect on my small bench. Was SRA willing to exchange it?

[AllanMN]

Ah. Found it. The fan wire wasn't routed through the slot in the fan housing and was then crushed when assembled, shorting to case/ground. What I saw on the scope was the PWM of the fan return,  pulled low by U12, an On Semiconductor  ULN2003A High Voltage, High Current Darlington Transistor Array. Splice or replace to repair.

A comment I saw on the video EEVblog #315 - Korad KA3005P Review/FAIL :

" The biggest problem with these 1HL brands is the counterfeit pass transistors. They can barely supply 1/3 their rated current due to a severely reduced SOA. I routinely buy them - they're a steal - and replace the pass transistors with the genuine parts. I never, EVER have problems again. I am very hard on power supplies and do everything I can to blow them up. "

Any thoughts if this mod is warranted? And details? If say, you are inclined to charge SLA batteries at 5 amps for a few hours which I would not be? Cavlovic killed his doing this I believe. Did you ever figure out how, or get a repair/replacement?

I wonder how hot the sink gets at a full dissipation. I don't have a thermocouple to check but will in a few weeks.

The overshoot isn't such a concern to me if it only happens at 3-5A load ranges as in Dave's video. Do you see that too, Bill?

All my voltage outputs are very accurate. I guess I like this PS for the money. Even if I have to fix it!

Allan

[pomonabill221]

Hey Allan... GREAT that you found the problem!  Your scope picture does look like a PWM signal with AC ripple on it, although since the supply for the fan is unregulated, that could be where the ripple is coming from... just glad you found the problem.

Yes, the overshoot does change depending on the load.  I do see overshoot at even low loads (although I don't remember what the loads were and what the overshoot was.. will have to try it again).

I have (I think) the same "version" that you have as far as pass xsistors used and the assembly of the heat sink.  Don't know if they are real or fake though.
I was charging a 24V battery once, and drawing max current from the supply, and it did shut down due to over temp on the heatsink (or thermostat in the xfrmr... really don't know which tripped), but after a while of cool down (about 10 minutes), it powered back up ok.
It was running pretty hot and the thermal overtemp seemed to do it's job.

The heatsink is a little small if you are going to push it, and I may mod it with a larger heatsink (maybe even a larger fan), but for now it is working great for a cheapie!

I know alot of people whine about how cheap it is, and all it's problems, but like I have said, it ain't an HP/Agilent/Lambda/Fluke/etc., and I know that.

One thing on my supply... the FWB was NOT bolted to the heatsink and they relied on the stiff leads of the bridge!  It was NOT laying flat on the heatsink (the lead side was not touching the aluminum), so I desoldered the leads, bolted the bridge on the sink, and resoldered.  How cheap was that!
I would suggest that you check the mounting of the bridge to make sure that at least it is making good mechanical contact to the heatsink, and maybe even bolt it down... it probably runs pretty hot when under high load and will not last too long.

Here are some pics of the bridge (notice the slight space on the lead side), the xsistors and bridge (notice how the goop doesn't look like it was making contact on the bridge), and the contact area on the heatsink (only partial contact on the bridge).

[pomonabill221]

Ah. Found it. The fan wire wasn't routed through the slot in the fan housing and was then crushed when assembled, shorting to case/ground. What I saw on the scope was the PWM of the fan return,  pulled low by U12, an On Semiconductor  ULN2003A High Voltage, High Current Darlington Transistor Array. Splice or replace to repair.

A comment I saw on the video EEVblog #315 - Korad KA3005P Review/FAIL :

" The biggest problem with these 1HL brands is the counterfeit pass transistors. They can barely supply 1/3 their rated current due to a severely reduced SOA. I routinely buy them - they're a steal - and replace the pass transistors with the genuine parts. I never, EVER have problems again. I am very hard on power supplies and do everything I can to blow them up. "

Any thoughts if this mod is warranted? And details? If say, you are inclined to charge SLA batteries at 5 amps for a few hours which I would not be? Cavlovic killed his doing this I believe. Did you ever figure out how, or get a repair/replacement?

I wonder how hot the sink gets at a full dissipation. I don't have a thermocouple to check but will in a few weeks.

The overshoot isn't such a concern to me if it only happens at 3-5A load ranges as in Dave's video. Do you see that too, Bill?

All my voltage outputs are very accurate. I guess I like this PS for the money. Even if I have to fix it!

Allan

Did you look at my PDF of the schematic?  They are using three drivers for the fan, then two more for each relay, and one is spare.

[AllanMN]

Yes, the overshoot does change depending on the load.  I do see overshoot at even low loads (although I don't remember what the loads were and what the overshoot was.. will have to try it again).

Ah. Well, that is a FAIL, and would be a thing to fix if it's very significant. Or at least know about. I haven't checked into that yet.

I was charging a 24V battery once, and drawing max current from the supply, and it did shut down due to over temp on the heatsink (or thermostat in the xfrmr... really don't know which tripped), but after a while of cool down (about 10 minutes), it powered back up ok.
It was running pretty hot and the thermal overtemp seemed to do it's job.

The heatsink is a little small if you are going to push it, and I may mod it with a larger heatsink (maybe even a larger fan), but for now it is working great for a cheapie!

I know alot of people whine about how cheap it is, and all it's problems, but like I have said, it ain't an HP/Agilent/Lambda/Fluke/etc., and I know that.

One thing on my supply... the FWB was NOT bolted to the heatsink and they relied on the stiff leads of the bridge!  It was NOT laying flat on the heatsink (the lead side was not touching the aluminum), so I desoldered the leads, bolted the bridge on the sink, and resoldered.  How cheap was that!
I would suggest that you check the mounting of the bridge to make sure that at least it is making good mechanical contact to the heatsink, and maybe even bolt it down... it probably runs pretty hot when under high load and will not last too long.

Here are some pics of the bridge (notice the slight space on the lead side), the xsistors and bridge (notice how the goop doesn't look like it was making contact on the bridge), and the contact area on the heatsink (only partial contact on the bridge).

Good to know. Yes, the sink is small, and maybe simple to improve. I might have some 1/8" Al sheet around to cut and bend up.

I'll definitely check the bridge when I replace the fan. Doug is sending me another to avoid splicing leads.

Seems colvinic (?) toasted his charging batteries and we haven't heard any more.

Did you look at my PDF of the schematic?  They are using three drivers for the fan, then two more for each relay, and one is spare.

Most definitely! That is a great contribution and I thank you for it!

[pomonabill221]

Did you think I had to splice the leads on the bridge?  I'm sorry I wasn't clear about that... there was enough lead length on the bridge (at least on mine), that I didn't have to extend or splice onto the leads....

Yes, the overshoot is surely something that I intend on looking into when I have a little more time.  Trying to rely on remembering to disconnect the output when switching on, or having to add a relay to the output (which I still may do for several reasons), doesn't fix the problem and I would like to fix the problem... just a personal "goal" or triumph!  hehe

I am glad that those of you that have downloaded my crude hand drawn schematics have found them useful.  I really tried to be accurate, but if you find something that isn't correct PLEASE let us all know!  OR if you have a question about my drawings, don't hesitate to ask!  (sometimes I have to ask MYSELF "what the heck did you mean by that??")

As crude as the software is, I still can't get it to work on my win7 64 box!  It keeps asking for matlab, and I don't have that!  And WHY would it need it anyway?  Just chalk it up to "crappy software"!  It would be nice though.

[commongrounder]

@pomonabill221:

That's interesting about the software needing Matlab.  I'm wondering if the there is a runtime Matlab application embedded in the Korad software that isn't 64bit compatible.  If that were happening, when the program attempted to start it would not find the runtime app functioning and issue an error message.  Korad programmers may have used a Matlab core to build the graphing and analysis (as much as there is of it) portion of the program.  I spoke with my son, who is a programmer and has used Matlab.  He thinks you could look in the folder that contains the program files for the Korad software, and see if there are any files that are Matlab related.  There might be a 64 bit version available that could be put in place to fix the problem.

[pomonabill221]

HHmmm.. that could be what is going on!
I will take a look and see.
thank You.

[pomonabill221]

WOOPS!!!  I meant labview... sorry
This is the error that pops up when I start the app,or try and run a program.
I can't find any kind of .ini file that would have the save path (it defaults to D:\ka3005p save data\  but I can't change it!)
It is controlling the supply, but I can't do what I want to (run a program).
Frustrating!

[AllanMN]

Did you think I had to splice the leads on the bridge?  I'm sorry I wasn't clear about that... there was enough lead length on the bridge (at least on mine), that I didn't have to extend or splice onto the leads....

Huh? No, sorry, I meant the fan lead which was crushed and shorted to ground/case, and almost severed. It's very fine wire, stranded, and I find those can be hard to splice when they are like a few fine hairs.

[AllanMN]

WOOPS!!!  I meant labview... sorry
This is the error that pops up when I start the app,or try and run a program.
...

Often there is some kind of free runtime software for developed applications you need. Perhaps this?

http://www.ni.com/download/labview-run-time-engine-2013/4061/en/

Might be worth a quick try. I'll try it on XP and see if I can shed light on it sometime here.

Allan

[pomonabill221]

Did you think I had to splice the leads on the bridge?  I'm sorry I wasn't clear about that... there was enough lead length on the bridge (at least on mine), that I didn't have to extend or splice onto the leads....

Huh? No, sorry, I meant the fan lead which was crushed and shorted to ground/case, and almost severed. It's very fine wire, stranded, and I find those can be hard to splice when they are like a few fine hairs.

OH ok... yes I agree the fan wires are small.  Lucky you found the problem and it was easy to fix!
Does the fan speed change with different loads (it will only change from idle with a load).  Normally it just idles.
Just curious to make sure the fan driver isn't cooked.

[AllanMN]

Does the fan speed change with different loads (it will only change from idle with a load).  Normally it just idles.
Just curious to make sure the fan driver isn't cooked.

Yes, all seems well. I only briefly connected negative to ground once I saw a malfunction, fearing doing damage.

[pomonabill221]

IF the supply is floating (which it should be), you should be able to ground (connect to earth) either the positive or negative output.
This is a common way to obtain a negative voltage referenced to ground (earth).
I know that sometimes ground and negative are interchanged, but in this case I am referring to earth as ground.

Great that your supply is working!
Just remember that it does have it's inherent problems (like spiking on power on, a slight overshoot when the output is turned on (could damage what you have connected)).
If you have a scope, take a look at the output to verify these symptoms... and remember it IS a cheapy chinese product, but I like it and I could afford it.

[tony3d]

I have been using mine for months now with no issues at all. I wouldn't leave it unattended pushing 4 amps, but anything under an amp or so I feel very confident with.

[pomonabill221]

I agree with not leaving it unattended pushing high current, but it's over temp seems to work pretty good (I found out), and I would feel safe at about 3 amps max., just watch the temps and feel the fan's output air temp.
Still.... it's not a bad idea to be around while it is powering something, just to be sure!

[AllanMN]

Here are some pics of the bridge (notice the slight space on the lead side), the xsistors and bridge (notice how the goop doesn't look like it was making contact on the bridge), and the contact area on the heatsink (only partial contact on the bridge).

Mine had a bolt tightened through the bridge and it was tight. Like my crushed wire, assembly has many errors without QC to catch it.

[commongrounder]

After reading about AllanMNs experience with a trapped wire, I figured I had better take a look inside mine to see how it faired the assembly process.  Mine did pretty well.  I found a couple of loose screws on the power transformer (mine is the dual supply, and the xformer is huge and heavy).  I also found the regulator board for the fixed 5-volt supply had a screw missing, which I replaced.  Here is something interesting.  I can see the area where the I/O connections are made, but mine is not a USB/RS232 version.  However, the circuit board is sitting right there, half populated.  Even the opto-isolators are there.  I wonder, if there were a way to determine what chips are used, whether my unit could be converted to have I/O capabilities.  If someone were willing to post a photo of that board that has the I/O connectors and related circuitry, I might be able to create a parts list for the conversion (assuming the parts have visible labels).   I could also post a photo of my partially loaded board for comparison.

[pomonabill221]

Wonder if the firmware is the same for all the models?  For instance, if you were to populate the I/O board, would the supply recognize commands?
If it isn't alot of trouble, might be worth trying it out!

[commongrounder]

Yes, I'm curious about the firmware compatibility, as well.  If I can ID the parts, I might try it just to see what would happen.  It would be fairly low risk and wouldn't take much time to set up.  I could also see how much the I/O board would cost as a part and just swap it out.

[pomonabill221]

Here are some pics of the I/O board.

[commongrounder]

Fantastic, pomonabill221!     Thank you very much for taking the time to do this.  I will study these and see if I can come up with a parts list for completing an I/O mod for a non-USB/RS-232 Korad.

[commongrounder]

Okay.  I studied pomonabill221's photos and pulled my board.  Surprise!    The boards are not the same...  I am assuming there is either more than one version of this board, or there is a board designed specifically for the KA300xD or P-3S triple output supplies.  I've attached a photo of my board.  If anyone has a "P" version of the triple output supply with the populated I/O board they feel like sharing a photo of, I will research the missing parts.  Thanks!

[pomonabill221]

Okay.  I studied pomonabill221's photos and pulled my board.  Surprise!    The boards are not the same...  I am assuming there is either more than one version of this board, or there is a board designed specifically for the KA300xD or P-3S triple output supplies.  I've attached a photo of my board.  If anyone has a "P" version of the triple output supply with the populated I/O board they feel like sharing a photo of, I will research the missing parts.  Thanks!

DANG IT!!!  Sorry yours is different.  I hope you find someone that has the info for you!

[pomonabill221]

Okay.  I studied pomonabill221's photos and pulled my board.  Surprise!    The boards are not the same...  I am assuming there is either more than one version of this board, or there is a board designed specifically for the KA300xD or P-3S triple output supplies.  I've attached a photo of my board.  If anyone has a "P" version of the triple output supply with the populated I/O board they feel like sharing a photo of, I will research the missing parts.  Thanks!

I betch ya that U 2_2_1_1 is the max232 uart.
Can you trace the pins from the USB/DB9 connector?  Where do they go?
Just assuming that the second flat pack isn't used?  Just a general observation on what IS there is that the vacant parts around that chip are not used.
It almost looks like the two controllers talk to each other, or share common data/address lines.  hhmmmm....
What's with that crystal being mounted diagonally???  Strange!

You really need to see someone else's board though... hopefully the second flat pack isn't used!
Good luck!

[commongrounder]

Thanks pomonabill221.  I will do some tracing out of the board when I don't need to use the supply for my work.  It may well be possible to figure out some of the parts by context.  As concerns that crooked crystal, they chose to mount it on the board up on its legs rather than directly on the surface.  It is just twisted a bit on its legs.
With any luck someone might see this and do a quick photo.

[pomonabill221]

I'm toying with the idea to replace ALL the output stage (three transistors/diodes/ballasting resistors/base resistors) with a single power fet, driven directly from the control board.

With a fet, you don't have thermal runaway and at least I know that it IS a real fet and not a counterfeit look alike!
I really don't like the use of ballasting resistors although with transistors it is necessary.  Even with several parallel fets though, you don't need ballasts as the fets will self ballast.

I'll give it a try and see what happens.

[jldesigns.eu]

Hello Folks, Since this is my first post on the EEVBLOG forum wanted to say Hi to ALL

I was a happy owner of Korad KA3005D-3S, until left half of my power supply failed. During testing of a system (12V 2A), cracking sound burning smell and smoke came out of vent holes.

Its only been used few times since I've bought it 13 months ago but distributor refused a warranty service (12 months warranty).

Since I’m electronics engineer I’ve turn it apart and fund that the microprocessor board has malfunction caused by fan short circuit (or possibly other way around). ULN2003A was blown, I’ve replaced the fan and ULN, to find out that 3 op amps are also shorted (TL082C), replaced those but that was not the end. There was something still short-circuiting 3.3V line, found out that it’s a microprocessor with sanded markings.

Can you help me in any way with fixing my PSU? I would need:
- a programmed microprocessor (probobly wont get it) or
- full microprocessor board (maybe Korad sells replacements?) or
- a firmware for the uC (second board from other half of my PSU is ok, but i doubt Korad didn't lock the firmware) or
- i could possibly write my own firmware, but reverse engineer the board would be a nightmare

Thanks in advance for any support.

PS. On the attached photo the board have desoldered 3.3 LDO and a ceramic cap on it, it was to check what is shorting out 3.3V line.

[electricar]

Hi folks,

does the newest revision of the PSU have those overshoot issues too? I bought the Korad KA3005P recently and have overshoots when enabling the output. Unfortunately, I didn’t look up the revision of the PSU yet.

[milamber]

Right. But they could also be level shifters/buffers for the R2R DAC. IMHO the 100ms transient response (no opamp is THAT slow) and the steps in the voltage after turn-on (watch Dave's video) speak a clear language.
Unfortunately Dave didn't do a transient test. Anyone who owns one of these want to do one? All you need to do is to connect a big resistor to the output and switch it with a MOSFET at ~100 Hz and trace the output voltage on a scope.

Is the Atmel clocked with a 12 MHz quarz? Did someone already try to exchange it with a 16MHz or 20MHz to tweak the transient response to 75 or 60ms?

[Andreas]

Hi folks,

does the newest revision of the PSU have those overshoot issues too? I bought the Korad KA3005P recently and have overshoots when enabling the output. Unfortunately, I didn’t look up the revision of the PSU yet.

Hello,

I have around 8% overshoot on the 5V setting.

With best regards

Andreas

[oger]

Hi everyone,

I also have this Powersupply. I have a later date, so I don't need to deal with a lot of the issues discussed here.m But I did a mod that prevents the fan from spinning permanently.
I do have however two things that bother me.
First is that the supply seems to have a capacitor at the output. I could not source a schematic so I'm not sure.
The problem is that when I connect an led for example and have the output voltage set to anything higher than 7 or 8v it will be destroyed because
current limiting did not kick in quick enough even though I set it for less than 10mA.
This might be a common problem among PSU though. I'm not sure.

Secondly, what bothers me the most is that the rotary knob for setting voltage and current behave erratically.
When I turn it, it might jump to 25v right away. Very dangerous when the output is on.
So I wonder if I could just replace the digital "pot" or is it the logic behind it at fault?
This gets progressively worse.

Thanks for info!

Stefan

[Dominic-V]

Just disassemble the rotary knob and clean it with isopropyl alcohol, then lube it and put it back in. Should work as new, at least mine did and never missed a digit after.

As for the current limit, just connect the led before turning the output ON.

[oger]

That did it!
Just had to clean the encoder. No more steps missed.

[Andreas]

in a german forum I have read that you can replace the encoder by ALPS STEC12E07 available at Conrad / Reichelt.

With best regards

Andreas

[oger]

I t does not not seem to be a special encoder.
but cleaning seems to remedy the problem quite good.
No need to buy a new one. Especially mine was hardly used.

[uv3afl]

Hello,

As this is my first post here I want to thank this forum and Dave Jones for plenty of constructive things I knew from EEVblog forum and videos!

I have recently purchased this device as I knew from several youtube videos to be not bad price/performance value. It seemed to be in working condition but out of calibration. It gave about 1,5% setting voltage error. Thanks to this theme I managed to calibrate it and it turned at last into some usable device! Anyway in couple of days playing with it I have run into some sort of a bug or what ever. I haven't seen anyone mentioned it. Thus I registered here and want to describe it and ask if anybody of you had such situation.
So. I attempted to charge a 18650 battery. The process went on normally in CC mode indicating 200mA and something about 3.45 volts on battery. At that moment I thought that there was no need to give it full charge and decided to limit charge voltage to some 'long term storage' value - 3,6 volts. I have set it up and surprisingly found out that it started indicating 3.6 immediately after setting. Multimeter attached showed that actual output voltage was still 3.47V... While indicated was already 3.6.

Some further experiments showed that when in CC there is a difference in set voltage and actual output voltage greater than 0.2 volts - everything is fine. Otherwise you will see limiting voltage instead of actual.

That's the story. Any comments?

[Kleinstein]

Still showing the set voltage is a kind of cheating - the voltage display should always show the actual voltage. The only good exception could be with the output disabled and than with a clear marking that it is the set voltage and not the measured.

So it might be worth do a few more tests on this.

[Andreas]

Hello,

that is the behaviour of the device. (also on mine).
When the difference of current / voltage is below a limit the "set" instead of "actual" value is displayed.
perhaps to supress excessive flicker near cc/cv limit.

With best regards

Andreas

[JackM]

I attempted to charge a 18650 battery. The process went on normally in CC mode indicating 200mA and something about 3.45 volts on battery. At that moment I thought that there was no need to give it full charge and decided to limit charge voltage to some 'long term storage' value - 3,6 volts. I have set it up and surprisingly found out that it started indicating 3.6 immediately after setting. Multimeter attached showed that actual output voltage was still 3.47V... While indicated was already 3.6.

Does the same thing occur when you attach a static load (ie. a resistor) and then put it in constant current mode?

[uv3afl]

Thanks for answers.

Does the same thing occur when you attach a static load (ie. a resistor) and then put it in constant current mode?

I've tested this. It behaves completely the same at the static dummy resistor load.
1. Set 2V as a limit, 0,5 Amps as current limit, attach 3,3 ohm resistor

2. Load on - everything is fine

3. Change the voltage limit to 1,9 volts
4. Good bye! 1,9 instead of actual output.


Now I really interested if it is a software bug or is it a hardware design feature. And I definitely don't think that it can be normal power source behavour...

[Ton]

It's a SW "feature" of the Korad power supplies (and the rebranded Tenma units), when ever the actual voltage output is within 0.2 V from the setpoint it will show the set-point instead of the actual value. 

already discussed here: https://www.eevblog.com/forum/testgear/korad-display-bug/msg585763/#msg585763

this "feature" is also seen on some cheap Chinese PID controllers, the speculation is that it gives the user a feeling of greater precision than the device actually gives. 

It is annoying but it is how these supplies have behaved, basically since they were released from Korad some 3-4 years ago.

Still I find them good value for the money even if this "feature" is irritating

Br Ton

ps. I have 3 of the Tenma branded version
pps A forum member have started the design of a alternative display PCB where this "feature" can be removed, but I dont know the current status of this, see the tread if interested.
https://www.eevblog.com/forum/projects/korad-ka3003d-redesign-and-upgrade/msg1011712/#msg1011712

[uv3afl]

It's a SW "feature" of the Korad power supplies (and the rebranded Tenma units), when ever the actual voltage output is within 0.2 V from the setpoint it will show the set-point instead of the actual value. 

already discussed here:
...
pps A forum member have started the design of a alternative display PCB where this "feature" can be removed, but I dont know the current status of this, see the tread if interested.
...

Oh, thanks a lot, Ton for proper directions! Now all further questions will be in 'right' places. And theme with alternative solution is of course of great interest! I will study that!

[uv3afl]

this "feature" is also seen on some cheap Chinese PID controllers, the speculation is that it gives the user a feeling of greater precision than the device actually gives. 

One more question if you please - why?     How does it work for getting the illusion of higher precision in this case?  I don't believe in flicker or noise Andreas tells about, as it will be also 'noisy' and flashing in say CV mode and also 'noisy' being far from voltage limit in CC mode... And if they care about CC/CV - 'noisy' mode switches, they will have it independently of displayed value , isn't it so?

[Ton]

Hi uv3afl

I guess that the theory, if you can call it that. Are that for a user with a pid controller set to keep for example a temperature at say 30 degC it will look more stable if it always show 30.0 degC than if it is sliding between 29.9 to 30.1 - it is all about the perceived stability.

I should probably have used stability in my first post instead of precision.

if you know stuff regarding process control, you are expecting some sliding around, not a static number.
Or your process is very stable with very small noise input.
The task to minimise the sliding around is sometimes a quite challenging one if your plant is not simple.

Br Ton

[tombi]

Hello,

With the overshoot problem, I wonder if putting a small cap in parallel with D1 might help here? I've seen some Agilent schematics that do this.

Tom

[Andreas]

What is D1?

if you mean the output: how much compared to the already used 330uF across the output?

Normally you would use a feedback capacitor across the voltage/current regulator.
But adjustment is critical: if too large the load regulation suffers. If too small you get overshoots.
And the largest issue gives a change from voltage <-> current mode.

with best regards

Andreas

[mos6502]

the already used 330uF across the output?

Are you saying this PSU has 330uF across the output? That's total junk ...

[tombi]

Sorry,

Way back in the thread somebody posted reverse engineered schematics and D1/D2 were between the voltage control and current control opamp and the pass transistors.

I was thinking like 1nF - nothing major.

Tom

[mos6502]

Way back in the thread somebody posted reverse engineered schematics and D1/D2 were between the voltage control and current control opamp and the pass transistors.

That would slow down the regulation even more. The reason a designer would put a ridiculous amount of capacitance like 330uF on the output of a linear power supply is because he's trying to make up for the slow regulation. Sorry guys, but everything points towards the fact that this PSU is garbage. I wouldn't consider it using for any electronics work because of the 330uF at the output alone. Maybe it's OK for charging car batteries. 

[uv3afl]

Way back in the thread somebody posted reverse engineered schematics...

I have found some better schematics for the device. And as the support for the device is far from being called 'support', I think this must be here. Attaching to the message. Maybe someone will stick it up to some 'documents' theme.

[uv3afl]

Now couple more pics about 'garbage' )) One of them:

 - garbage on output terminals when the load is soft-off and you hard-powered off the PS. The second one:

- garbage on output terminals when you momentarily press/depress power-on hardware button. How about killing something attached to PS with that? Hopefully too short, but... Anyway, do know.

[uv3afl]

Some more news!
My dispute with the seller at Aliexpress has just finished successfully. They agreed that the 'feature' of not displaying output voltage when delta between Set and Out voltages is less than 0,2volts is a constructive defect and will refund me half-price of the PS.

[mos6502]

Hey, at least something  Now you only have to program your own firmware and redesign the regulation circuitry. 

[Kleinstein]

Having something like 330 µF at the output of a lab power supply is about normal. There are supplies with even more capacitance at the output, that are still considered good. A normal lab supply is not expected to be very fast in regulation, as this makes is really difficult to have it stable with any possible load. There is also not much sense in making the supply faster than 1 m of cables. So if you have a load that needs fast current spikes, add local decoupling capacitors.  There are also PS designs that have way less capacitance at the output, but still have a poor / slow current limiting and may this way behave as if there would be an even larger capacitor.

The funny high frequency transients look a little strange. So the question really is how much of this is at the output terminals and how much is just EMI coupled to the wires to the scope.

[mos6502]

I disagree, 330µF is way too much. My TTI PL320 has 47µF and you can still kill a zener diode if you're not careful and dial down the voltage far enough. 330µF will fry a lot of stuff if there's a fault in your circuit.

I'm not quite sure what you mean by "faster than 1m of cables". But a proper bench PSU will have separate sense connections to compensate for any lead resistance. So yes, fast regulation/low output capacitance are what makes a bench PSU useful.

In the case of the Korad, you might as well buy a switching power supply like the CPS3205. It's smaller, lighter, cheaper and more efficient.

[pomonabill221]

Way back in the thread somebody posted reverse engineered schematics...

I have found some better schematics for the device. And as the support for the device is far from being called 'support', I think this must be here. Attaching to the message. Maybe someone will stick it up to some 'documents' theme.

Well my efforts were for not then?
Thanks for posting the schematics!
Finally after 4 years,, we have complete schematics!

[uv3afl]

Well my efforts were for not then?

I shouldn't say so! If you look closer to the schematics you will notice some annoying mistakes. For e.g. at main board two 'GNDF' at upper of R13. Looks obviously confusing.  And here I take your sheet and set everything into places  I'm too lazy to take CAD and draw everything like it shold be. By the way, good moment to say thanks for your rev-eng job! It was first internals info about this PS that I've seen!

[basil]

I am having a strange problem with my KORAD 6005D (60 V 5A model ) !!!!
-> Power supply works perfect up to a set voltage of 19V ~ 20V .
-> If i increased the set voltage above ( ~21 till 60) & (o/p is enabled ) , i can hear a relay tick sound from the power board at a constant interval . At this time i am getting a voltage discharge ramp at the output terminal . In the front panel display values also i can see this effect ( display works good )
-> i noticed that , If i remove the voltage feedback/sense line from the connector J10 at the front panel pcb the relay ticking stops ( but the power suppy will not work as the sense lines are removed , right ? )

I can share more specific details , If someone can help me to figure out the issue .
 

[Kleinstein]

The description somewhat sounds like a problem with transformer tap switching. Maybe a worn out relay contact or a loose wire / blown fuse from the transformer. My guess is that the higher transformer tap for some reason does not deliver power.

The 60V 5A version is likely different from the smaller units in some aspects. I would expect a different transformer tap switching at least, to reduce power loss. So a little more information, a few photos of the internals might help. A visual inspection for lose wires, maybe cold solder joints or blown fuses would be the nest step anyway.

A helpful point to measure would be the voltage of the large filter cap (e.g. solder temporary wires and measure). It needs some care: the voltage could go to about 80 V and in case of a short there is plenty of power.

[basil]

 You were absolutely right and thank you very much for the inputs  !!!

 It was due to a faulty relay ( K3 in my case ) . Now the unit is working perfect . 

[uv3afl]

Hey, at least something  Now you only have to program your own firmware and redesign the regulation circuitry. 

Yes, KA3005P is really 'at least something'. And no DIY with it! It's not interesting, because it's somewhat classic )) It's a cheapie with more than reasonable functionalyty for it's cost.
And here is the PS of my real interest that I build now. Fortunately, it's completely open source and there's no need to develop something from scratch. And I am almost done with it. It works fine but there's still some mechanical work to fit it into the case. By the way it has no cap on output terminals inside by default. It is really fast in regulation and offers LED operations with current limiting. But... there's the other side - on some loads people catch some oscillations. And... Put some output cap on terminals 

[Tony Mach]

So I have a faulty KA3005D… and maybe someone finds the information helpful I have gathered so far (unfortunately I am not done yet).

Pinning down the problems with my PSU

At my workplace we bought this PSU, and it was broken out of the box. You could press OUTPUT, but it would not output anything… The voltage would stay at 00.00 V and the current at 0.000 A. Otherwise it seemed fine.

From time to time one or the other colleague took it, played around with it, couldn't accomplish anything, and then put the PSU angrily back into the shelf…

Now I had enough and declared the PSU dead – but I didn't want to give up so easily. Instead of having to put the PSU in the electronics-dumpster I was allowed to take it home. And now it is my problem…

With the help of the schematics here I did find out that the power-board (in the rear) is fine. With a voltage on pin 1 I could change the output voltage. The 7-segment displays even displayed the output voltage and current! The problem seems to be localized on the control-board (the MCU, the two discrete DACs with the comparators).

With the "calibrate" procedure I could find out that the "zero voltage" is adjustable from about -100 mV to about +200 mV in steps of about 2 mV. The three other calibration settings ("zero current", "max voltage" and "max current") do nothing (sometimes the voltage "blips" when turning the wheel). While I can not rule out a hardware problem, I suspect a software problem. (Maybe wrong values in the I2C-EEPROM?)

Going after the Firmware

Somewhere I found out that it is a Nuvoton M054 MCU. And oh wonder of wonders, at my workplace we have a Nuvoton Nu-Link-Me. I managed to connect it (see attached image), and I could connect to the MCU right away! In my PSU there is a M054LDN. Success! But no, not so fast, unfortunately the "security lock" for the flash is enabled – so I can not read out the firmware and can not debug it. At least I was able to get a PinView which I attached as a screenshot (whether it is accurate is another question).

So I could write my own firmware…

Next step: Getting into CooCox with the Nuvoton MCU.

Famous last words: How hard can it be to manage a handful of GPIOs?

So that is what I have so far. I don't promise that I will be able to follow up on that. Maybe I will, maybe not, or maybe someone else can use that information.

[Kleinstein]

It depends on the hardware and desired functions how complicated it would get wirting your own software. The main part could very well be interfacing the LCD output. Controlling the DAC / ADC should be relatively straight forward, unless they used the µC and a few external parts to implement a kind of high resolution sigma delta of other integrating ADC. Before deciding on the software, one should have a better picture on the actual hardware to control.

AFIAK there is a chain of 74HC(T)595 used as a shift register for a R2R type DAC. So one might be a able to check the serial signal - just in case the trouble is not with the MCU. One could also check just the output of the R2R chain - this should be relatively easy to identify on the board.

Edit:
Looks like there are schematics available. No more than just the DAC, even the ADCs and display seem to be external. So if really needed, the software would be relatively simple. It would really worth looking at the DACs output, to exclude a problem with the control OPs.

[Andreas]

Hello Tony,

is the output relay clicking when you activate the output?

With best regards

Andreas

[Tony Mach]

@Andreas:

This is a economic PSU, it has no relay for the output. Very economic. (One more reason I would want to write my own software, to add a relay for the output… the control-board even has an output for such a relay.)

@Kleinstein:

I have no isolation transformer, and the mains-switch is on the frontpanel – very near to the R2R DAC …

I am so tempted to write my own software (even if this PSU has an hardware fault – it would be easier to pin down the fault if I know what I output).

The main task would be to learn how to I/O on a Nuvoton with CooCox (there is some support in CooCox, but I'm an AVR32 guy, and everything in this new world feels slightly strange…)

Otherwise the tasks would be:
- Setting of the DAC (outputting with the 7495s)
- Output via LEDs (as well 7495s)
- Input of the wheel (interrupts?) and buttons (button matrix?)
- General "user interface"
- Reading the ADC for the voltage and current
- Switching the relays for the windings (voltage "preselect")
- Overcurrent handling
- Some calibration routines
- Reading/writing to the I2C EEPROM

Other niceties (which can wait) include:
- Reading the temperature and adjusting the fan
- Buzzer
- Output Relay
- Use of M5 LED to display "LOCKED"
- I/O via serial interface
…

Yeah, very doable. 



I just need to get started…

 

[Kleinstein]

If you don't won't to poke around the circuit when life, one can usually solder wires to the test-points and connect everything to a DMM / scope before powering up.

If the mains switch is so close to the logic part, one should consider insulation (e.g. heat shrink or similar) anyway.

The µC is supposed to be ARM M0 based. AFAIK it takes some time to get used to the a new type of µC. A first point would be to decide on what type of support files (e.g. HW definitions as C include files) to use. Often there are kind of higher level version with special macros for initialization - however some of these systems are controversial. So some use the higher level interface and some only the register names. The biggest part is likely getting started to get a first program.

[Tony Mach]

Every solder point I touch brings a PCB closer to death… I'm more a software person… So I'm glad the debug connection is working.

Anyway, I got a "hello world" to compile in CooCox for the M054 with setting of one GPIO (for the fan).

So after I successfully compiled, I could not resist, and the original firmware is already GOOOONNNE !!!

The first I did was to switch the fan (P4.1) between full speed and low speed, which was surprisingly easy.

I just added the switch of the columns (P1.4-P1.6 and P4.2) of the LEDs.

I am making unreasonably fast progress… I smell a trap. 

Next I have a date with the 7495 shift registers – something I have been looking forward to do long before I met the Korad PSU.

[Andreas]

This is a economic PSU, it has no relay for the output. Very economic. (One more reason I would want to write my own software, to add a relay for the output… the control-board even has an output for such a relay.)

Hello,

I have 4 of the units (2 * KA3005D and  2*KA3005P).
All 4 have a output relay.

So either you have a KD3005 or something is wrong.

With best regards

Andreas

[Tony Mach]

Well, I definitely have a KA3005D, and no output relay.

Please be advised, there are (at least) 3 version:
- The "old" version, that eevblog initially tested (only one capacitor on the power-PCB)
- The "middle" version, with updated PCBs (three capacitors on the power-PCB)
- The "new" version which has among other changes "screw type" connectors for the output (the old/middle had "plug only" connectors for the output)

I have the "middle" version. It has 2 relays for switching of the windings, but no output relay.

Meanwhile, I now have a first version of my software that can access the two DACs (for voltage and current), and I can switch the output voltage in software – while I have not figured the DACs out completely (and might still have a bug or two), I think the hardware should be OK. This must have been a configuration problem.

Achievement unlocked: 74HC595
Achievement unlocked: Control the seven segment display
Achievement unlocked: Control the DACs

As I can get voltage and current out of the PSU, if I either manually set the control voltage, or if I set the DACs in software – without switching any relays in both cases – I am quite certain that there is no output relay in my KA3005D. I even looked and visually checked all the PCBs – I did find only the 2 relays for the windings on the rear PCB, there is no output relay in the path.

I am curious: Where is this output-relay in your PSUs? On the power-PCB? Or on that little connector-PCB? On the main-PCB?

Anyway, my software needs further work… Once more onto the breach!

[Andreas]

Ok you are right,

the clicking that I hear is from the transformer switching when I set the voltage above 5V.
(What I usally do).

With best regards

Andreas

[Tony Mach]

Hello Andreas,

when you go above 5V, you hear one of the "windings relays"! Both winding relay are connected in such a way that there is always a connection from the transformer to the power electronics.

The only thing that selects an output of 0 Volt is the control voltage. If this control voltage fails (e.g. faulty DAC), or if the winding relay fail (happened to one person, there is a video on youtube), you will get voltage on the output.

Unfortunately there is no output relay.

Kindest regards.
Tony

Anyway, I made further progress!

The SHCP and STCP are mislabel on the processor end…    I spend at least 2 hours trying to figure out why my 74HC595 routine did not quite work with the DACs 75HC595s… The same routine worked just fine with the LEDs 75HC595.  They mislabelled these two pins on the schematic… 

But now I can control the DACs!!! 

Achievement unlocked: Output hexadecimal numbers on the LEDs
Achievement unlocked: Control the DACs

Onwards to success! 

[Tony Mach]

Well, I reached a crucial point with my firmware for my KA3005: The basic functionality (and everything needed for this) is working!!! I can adjust voltage and current, and the respective values are displayed! 

It took a little longer than I thought to reach this point – but hey, I do this in my spare time, for free and for fun (at least my tortured sense of fun)…

There is a metric ton of caveats attached to this firmware:

• This works on my KA3005 - other hardware revisions may have critical differences (There is much strangeness in the schematic…)
• AFAIK one can not read out the original firmware - so once you overwrite the original firmware you can not go back
• There is no calibration routine, the calibration values are hardcoded
• CooCox and the debugger are strange… Relays sometimes switch when you load new firmware – this contributed to me having once shot the power transistors  The firmware was broken, so I wrote a new firmware, which broke the PSU - talk about irony… (But hey, I bought new transistors, fixed it  and learned an important lesson: Do not leave a load like a 12V 100W bulb attached when the debugger is attached - detach any bigger electrical load before you load a new firmware. A 230V 50W bulb on the other hand is fine.)
• I make mistakes (I once reached a peak value late at night of 2 bugs per 1 line of code) – this also contributed to the aforementioned incident where I shot the power transistors  But now the switching of the relays should be solid.
• Memory is not implemented (M1 to M5 not done)
• The interface to USB/serial is not implemented (my PSU model is without the interface board, and one would need to reverse-engineer the protocol, so this maybe will never happen)
• The keys are not debounced
• No LOCK, no OVP, no OCP
• The LEDs flicker under certain conditions (I know how to fix that, but haven't had the time to do it…)
• Haven't fully figured out the temperature-ADC yet
• Fan is switched on with the output, and runs aways with full speed while output is on
• Other problems I forgot
• Source code needs clean-up
• No warranties whatsoever
• No buzzer

But there are also good things:

• No buzzer
• IMHO the adjustment of voltage and of the current is nicer in my firmware (and I might improve it even further by a little bit)
• Considering the power rating of the transformer, and of the transistors, I think it could be possible in future to "overdrive" the PSU and output more voltage, or possibly even more current, under certain circumstances (if one stays within the ratings of those components)
• One could implement custom behaviour (e.g. "soft start", or specialized battery charger, or or or …)
• One could add an output relay, or maybe even an MOSFET as an output switch

If you want to use this firmware, you need CooCox (I used 2.0.7, but other versions should do fine). I had to add cmsis_core, M051_BSP_CMSIS_V300_001 and C_library to my project (the C_library might not be necessary - the boilerplate I used needed it, but I haven't checked if I still need it).

Sorry, no pre-compiled binary from me, no turn-key project – if you want to use this experimental firmware, you need to know what you are doing. Have fun tinkering! 

I might continue working on some of the missing features, but I do not promise anything. 

[xani]

I guess lessons learned here is that if you even design a power supply, make sure it is is safe even *if* your MCU outputs "please emit smoke" combination

I guess I'm lucky that only problem with my KA3005D-3S (not produced anymore, it is like 2 KA3005D glued together + 5V 3A output + option to link two to get 60V or 6A) was dodgy power switch (locked in place) and a bit loud fan

[rufusalceste]

Hi Folks,

I am currently trying to repair my Tenma 72-10495 (rebranded KORAD psu) . I found the issue but I would like to replace R3 and R4 on both outputs boards as these have heated a lot and burned some of the PCB right under them. I am afraid their value is now off...
The problem is I can't manage to figure the value of these resistors (see photo attached) as the colors are pretty unclear and also because this kind of color arrangement (low value resistor ?) is new to me.
R3 seems to be black/black/silver (or is it brown/black/silver ?)
and R4 seems to be black/black/gold (or is it brown/black/gold ?)

The picture is pretty representative of what I can see with my own eyes.

If any one of you guys has a clue that would be great.

Thanks a lot !

[Armadillo]

Hi Folks,

I am currently trying to repair my Tenma 72-10495 (rebranded KORAD psu) . I found the issue but I would like to replace R3 and R4 on both outputs boards as these have heated a lot and burned some of the PCB right under them. I am afraid their value is now off...
The problem is I can't manage to figure the value of these resistors (see photo attached) as the colors are pretty unclear and also because this kind of color arrangement (low value resistor ?) is new to me.
R3 seems to be black/black/silver (or is it brown/black/silver ?)
and R4 seems to be black/black/gold (or is it brown/black/gold ?)

The picture is pretty representative of what I can see with my own eyes.

If any one of you guys has a clue that would be great.

Thanks a lot !

R1 = 1Kohm  5W
R3 = 0.1 ohm 2W
R4 = 0.1 ohm 2W

[Andreas]

Hmm,

in the cirquit diagram attached here:
https://www.eevblog.com/forum/repair/korad-ka3005p-power-supply-calibration/msg370823/#msg370823

R3,R4 are 0.22 Ohms.

So could it also be red,red,gold?
Or have the values changed over the years.

with best regards

Andreas

[rufusalceste]

Thanks a lot guys.

Found more info in Dave's followup video on the Korad PSU.
Seems the older version (yellow PCB) had 0.1R resistors and the newer version (green pcb ) has 0.22R resistors.
See screen grab below.

Cheers !

[moeyekung]

Hi everyone.Who do it success?.
I try with coocox+jlink but no luck.
Mr.Tony Mach Can you send me .bin or .hex
to flash mcu directly sir.My mcu dead
Now I bought new mcu already.
   Regard.

Well, I reached a crucial point with my firmware for my KA3005: The basic functionality (and everything needed for this) is working!!! I can adjust voltage and current, and the respective values are displayed! 

It took a little longer than I thought to reach this point – but hey, I do this in my spare time, for free and for fun (at least my tortured sense of fun)…

There is a metric ton of caveats attached to this firmware:

• This works on my KA3005 - other hardware revisions may have critical differences (There is much strangeness in the schematic…)
• AFAIK one can not read out the original firmware - so once you overwrite the original firmware you can not go back
• There is no calibration routine, the calibration values are hardcoded
• CooCox and the debugger are strange… Relays sometimes switch when you load new firmware – this contributed to me having once shot the power transistors  The firmware was broken, so I wrote a new firmware, which broke the PSU - talk about irony… (But hey, I bought new transistors, fixed it  and learned an important lesson: Do not leave a load like a 12V 100W bulb attached when the debugger is attached - detach any bigger electrical load before you load a new firmware. A 230V 50W bulb on the other hand is fine.)
• I make mistakes (I once reached a peak value late at night of 2 bugs per 1 line of code) – this also contributed to the aforementioned incident where I shot the power transistors  But now the switching of the relays should be solid.
• Memory is not implemented (M1 to M5 not done)
• The interface to USB/serial is not implemented (my PSU model is without the interface board, and one would need to reverse-engineer the protocol, so this maybe will never happen)
• The keys are not debounced
• No LOCK, no OVP, no OCP
• The LEDs flicker under certain conditions (I know how to fix that, but haven't had the time to do it…)
• Haven't fully figured out the temperature-ADC yet
• Fan is switched on with the output, and runs aways with full speed while output is on
• Other problems I forgot
• Source code needs clean-up
• No warranties whatsoever
• No buzzer

But there are also good things:

• No buzzer
• IMHO the adjustment of voltage and of the current is nicer in my firmware (and I might improve it even further by a little bit)
• Considering the power rating of the transformer, and of the transistors, I think it could be possible in future to "overdrive" the PSU and output more voltage, or possibly even more current, under certain circumstances (if one stays within the ratings of those components)
• One could implement custom behaviour (e.g. "soft start", or specialized battery charger, or or or …)
• One could add an output relay, or maybe even an MOSFET as an output switch

If you want to use this firmware, you need CooCox (I used 2.0.7, but other versions should do fine). I had to add cmsis_core, M051_BSP_CMSIS_V300_001 and C_library to my project (the C_library might not be necessary - the boilerplate I used needed it, but I haven't checked if I still need it).

Sorry, no pre-compiled binary from me, no turn-key project – if you want to use this experimental firmware, you need to know what you are doing. Have fun tinkering! 

I might continue working on some of the missing features, but I do not promise anything. 

[VITHARANA AMILA ERANTHA]

i allso have same problam in my power supply ka3005d.

[nilsb]

Although it's an old post - I want to thank Tony for this bold attempt to rewrite the firmware.

It was really exciting to read. Sources are uploaded maybe someone might be able to continue on this one day 

Nils

[dmulligan]

I just came across this related blog post: https://m0wut.com/2019/08/25/power-supply-hacking-part-1/