Author Topic: Korad KA3003D redesign and upgrade  (Read 14623 times)

0 Members and 1 Guest are viewing this topic.

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #25 on: October 04, 2016, 07:43:30 pm »
The capacitor C34 at the sense inputs could be a problem. If at all it should be rather small (more like 1 nF), as a delay could be a problem.  I would more suggest diodes parallel to D11 / D12 to have the high frequency feedback from the power outputs - this helps to reduce ringing from inductance in the output lines. Though usually the user is responsible for using low inductance (e.g twisted) wiring.
I had initially thought that as the sense lines are relatively low impedance (probably <10R), a small amount of capacitance shouldn't be too much of an issue. I'm not sure I follow your parallel diode comment - I was under the impression that in normal conditions (sense lines properly connected), they would have no effect as they'd be a very low voltage across them?

D11 and D12 should be strong enough to carry the full load current - worst case the current will flow here.
They're both 3A rated schottkys.


The diode D15 could go to GND as well, just like D14 - normally the input voltage should be in the +-200 mV range, no more.
That doesn't seem right to me. I've designed the divider (R24, R28, R35) to make full use of the opamp range: -35v on VSENSE- equals -5v at the output of the divider. If D15 went to GND, that would clamp it.

I made the decision not to bother with diodes on the current amplifier - for that to have 500mV on it, you'd need 10A flowing through the shunt - at which point something has obviously gone totally wrong and it's beyond fixing with the regulation loop.

C35 only makes sense if there is a capacitor parallel to R39 as well. However one could very well end up with C35 not fitted.
Noted (and added). Always better to have not fitted components on the PCB than to have to bodge extra ones on!


I can't find the output capacitors. There should be caps directly at the power-outputs : something like 100-500 nF as low ESR /ESL caps (MLCC) and 1 or 2 low ESR electrolytics with something like 100-470 µF.
The output terminal board in the Korad supply has a 220uF, 50v electrolytic and a couple of ceramic caps, which I'm planning to keep. I might replace them with known brands though (Panasonic M series), to guarantee a nice low ESR!

In the layout, I can't find decoupling caps close to U15,U11,U12. At least U15 is rather fast and might really need one close to the chip.
The decoupling caps for U1, U2, U6, U8, U10, U11, U12, U13, U14 and U15 are mounted on the reverse of the PCB (right underneath that specific chip). I've tried to keep it mainly to a single sided load, but usually find you get the shortest current loop for decoupling caps if they're on the other side of the board.

Thanks for all the suggestions!
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #26 on: October 05, 2016, 03:38:22 pm »
If everything is cabled good the sense input will not have a problem with the extra capacitor. But it could be a problem if the cables are not perfect (e.g. long non twisted cables or even one sense line missing): here the extra cap between the sense input could cause oscillation.

Having capacitors from the sense to the power terminals could be an alternative to block RF, but without the negative effect. These caps can also provide a fast path for the feedback signal in case the cabling is not good, so they improve stability.

In the layout D11/D12 look rather small for 3 A diodes.

I agree with D15 - can not go to GND in this circuit. The circuit I used for simulation did not have the divider resistor, but just the differential amplifier with low gain (e.g. 0.2) - in this case diodes to GND are OK.

There is no real need for the voltage sensing amplifier to be a AZ type, a relatively fast precision one would be good enough and lower noise - not a big change, as the pin-out should be the same. The OP2177 is still quite slow, I would prefer a faster version (e.g. >= 1 MHz GBW and > 1 V/µs) with software trimming a low offset voltage is not that important. So more like an TLC272 / TL082.
 
The following users thanked this post: cowana

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #27 on: October 06, 2016, 02:52:44 pm »
Having capacitors from the sense to the power terminals could be an alternative to block RF, but without the negative effect. These caps can also provide a fast path for the feedback signal in case the cabling is not good, so they improve stability.
That's a really neat solution - AC coupling with the output, and taking the DC from the sense lines. I've ditched the capacitor between the sense lines, and switched to your suggestion.

In the layout D11/D12 look rather small for 3 A diodes.
Yeah, they might have been 3A rated, but a quick calculation says they'd last ~1.5 seconds at that before overheating! I've switched them for some slightly more beefy versions, plus added some via heatsinking.

There is no real need for the voltage sensing amplifier to be a AZ type, a relatively fast precision one would be good enough and lower noise - not a big change, as the pin-out should be the same.
Staying with SOT23 packaged amps, there are a lot more options available if I go from a dual opamp to a single opamp (which is fine as I'm using separate packages for the first V and I amplifiers anyway). How about an OPA209? Noise is better (2.2nV vs 5.9nV), bandwidth is greater (18MHz vs 10MHz), slew rate is slightly worse (6.4V/us vs 8V/us).

The OP2177 is still quite slow, I would prefer a faster version (e.g. >= 1 MHz GBW and > 1 V/µs) with software trimming a low offset voltage is not that important. So more like an TLC272 / TL082.
Good point about software trimming of the offset voltage - I had discounted some opamps forgetting that. With my 16-bit ADC/DAC and the fast floating point unit in the processor, there's no issue with doing multi-point calibration (even non-linear if required). The TLC272 and TLC082 are both available with the same pinout in a TSSOP package (Farnell 2496465 and 2406736 respectively) - I've switched my footprint from the narrower MSOP to that TSSOP package, so will probably use the TLC272 instead.
« Last Edit: October 06, 2016, 02:56:27 pm by cowana »
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #28 on: October 06, 2016, 04:04:12 pm »
The OPA209 is a good low noise OP for a low impedance source, but not that suitable here. It somewhat depends what you are looking for. With the MAX5136 as DAC this will not be super high precision or very low noise anyway. As there is not much gain here the OP does not need to be fast either. So many types to chose from, just like the two regulator OPs.

With the better DAC the 1/f noise of the TLC27x could be significant. So a slightly "better" (BJT based) OP might be a good idea. No need for something fancy - so an RC4558 might do it.
 

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #29 on: October 06, 2016, 05:21:02 pm »
The OPA209 is a good low noise OP for a low impedance source, but not that suitable here. It somewhat depends what you are looking for.
What parameter is it that makes it unsuitable? The main specs I would imagine would be important for the front-end voltage amp would be a high enough slew rate for the voltage changes (which won't be that fast due to the 220uF output cap), low noise, and low drift offset (so you can calibrate it out). Power rails, current draw and output swing are fairly unimportant.

In terms of the specifications I'm designing the PSU for, it will be used as a general purpose power supply for my lab (mainly digital/low speed analogue circuits, most of my designs have their own local regulation anyway). However, in the scheme of aiming for a solid robust design, I would like to be significantly better than the original Korad's specs (if it's not, why bother redesigning it!):

Quote
Voltage:
Output Voltage Range: 0-30V Continuously adjustable
Setup Resolution: 10 mV
Setup Accuracy: <0.05% +20 mV
Ripple: < 1mVRMS
Read back Accuracy: 10 mV
Temp. Coefficient: < 0.1% + 10mV
Read back Temp. Coefficient: < 100 ppm+10 mV
 
Current:
Output Current Range: 0-5A  Continuously adjustable
Setup Resolution: 1 mA
Setup Accuracy: <0.1% +3 mA
Ripple: < 3mARMS
Read back Accuracy: 1 mA
Temp. Coefficient: < 0.1% + 5mA
Read back Temp. Coefficient: < 100 ppm+5 mA

With the MAX5136 as DAC this will not be super high precision or very low noise anyway. As there is not much gain here the OP does not need to be fast either. So many types to chose from, just like the two regulator OPs.

I presume this is just a matter of what is considered 'high accuracy'? Compared to the rough resistor-based DAC of the original Korad design (which didn't actually use any type of voltage reference), I would have thought the MAX5136 would give a decent output. Is it worth adding an active low-pass filter after the DAC output to remove any high frequency noise from the output? This could have a very low bandwidth (say 500Hz) - it's very rare you want to change voltage or current that quickly.

With the better DAC the 1/f noise of the TLC27x could be significant. So a slightly "better" (BJT based) OP might be a good idea. No need for something fancy - so an RC4558 might do it.

One of the things I liked about the TLC272 was the low drift (and the fact it specifically says it's a precision opamp) - although I do see the high noise figure... The TL082 is in the middle with noise, but much higher slew rate, higher bandwidth, and hopefully lower offset drift...
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #30 on: October 06, 2016, 06:17:23 pm »
I just realized the OP2177 is not that bad - though this was a dual version of the OP177  :palm:
It's actually substantially faster. So the OPx177 family would be Ok. The ADA4177-x is slightly faster.

With a 220 µF output capacitor and enough reserve for parasitic inductance (separate boards for power and output caps) one may not be able to have the regulation super fast anyway.

The TL082 has rather high offset drift. Another classic would be the RC4558 - though I could not find drift data, but as a BJT based OP, drift should not be that bad.

The way to turn off the output with a relay at the base of the transistor is a little strange for a board where you try to save space. Doing it electronic would make more sense.
 

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #31 on: October 06, 2016, 06:54:53 pm »
I just realized the OP2177 is not that bad - though this was a dual version of the OP177  :palm:
It's actually substantially faster. So the OPx177 family would be Ok. The ADA4177-x is slightly faster.
Ahh, so many opamps with very slightly differing specs! Luckily they generally have the same pinout - so if I have nice wide pads on the PCB, I can put any of the options that have been discussed so far on. The OP2177 will do nicely as a starting point.

The way to turn off the output with a relay at the base of the transistor is a little strange for a board where you try to save space. Doing it electronic would make more sense.
But you don't get a satisfying click that way ;D. The boards aren't really tight for space (there's spare digital room top left, and spare analog room top right). I do like small SMD packages though - they're neater, and easier to solder (plenty of flux, then swipe with a hot iron loaded with solder).

It looks like the design is almost there now - if no major changes or suggestions comes up, I'll try and get the board layout finished and ordered over the weekend. Kleinstein, you've been amazingly helpful with identifying a lot of the design issues before the stage where it has to involve cutting through copper tracks :)
« Last Edit: October 06, 2016, 06:57:44 pm by cowana »
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #32 on: October 06, 2016, 08:25:23 pm »
For the sense amplifier one might have to use low tolerance resistors. If wrong, the DC / low frequency output resistance can get slightly negative (e.g. like -1 mOhms) and thus could cause instabilities with an extreme short. It might help to allow for an intentional adjustment resistor to get a defined +0.1 mOhms or so.

When not using the sense inputs correctly this will also change the resistor ratio. If the sense+ line is not connected nothing bad happens - just some .1 to 1 mOhms of output resistance. If only the sense- is not connected right and thus relies on the resistor, the output resistance would turn negative.  If the voltage divider is not too low in resistance, just the plug / wire resistance can be enough to save the day.

P.S:
Forgot the conclusion from the first part. It gets quite difficult to calculate the right resistors. So it might be easier to use just a differential amplification (with gain at about 0.15)  and no extra divider first. Not much difference: no R35 and different resistor values. For extra flexibility one could  split R34 in two in series and have a cap parallel to one (for full symmetry). R24 could than be removed.
« Last Edit: October 07, 2016, 08:19:55 am by Kleinstein »
 

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #33 on: October 08, 2016, 01:04:26 pm »
I'm not overly worried maintaining ideal behaviour if the sense inputs not being used - the connection marked as 'FRONT_SENSE' is connected to the rear of the binding posts inside the unit, so that'll be the mode used 99% of the time. The rear terminals (with dedicated SENSE terminals) will only be used if I specifically need to use sense connections for a task - although having said that, it shouldn't blow the PSU/DUT up if one of the wires comes loose.

That's an interesting idea of removing the voltage divider. Am I right in thinking the attached schematic is what you have in mind?
  • C38 is small (or not fitted)
  • R38 and R39 provide the offset voltage of the output (100mV)
  • C44 matches C35 providing equal responses for each side of the amp
  • As the gain is x0.1, does the split resistor biasing need to apply 1v (with an impedance equal to R34) in order to achieve the 100mV output offset?
  • Probably depends on the above point, but what's the common mode voltage of the opamp inputs? My feeling is it is 0V, but if it's not then D14/D15 may be an issue
  • Do D14/D15 do anything in that arrangement? If not, is other input protection needed?

Apart from a slightly neater arrangement and less resistors requiring high tolerance, I'm not fully convinced of this purely differential arrangement. At the moment, I'm more comfortable with the original idea - but that could change!

 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #34 on: October 08, 2016, 02:26:53 pm »
Having just the differential amp, without the divider has a few advantages: it is much easier to calculate the right resistors (and caps) - with the divider it gets a little messy. Less low tolerance resistors needed. The input protection with D14/D15 also helps with cross over from CC to CV mode (less or no overshoot). BW requirement for the OP could be a little lower.

The +input f the OP is essentially at GND - it's about 1/10 the voltage of the shunt + drop in positive output line, so well below 100 mV. So the diodes to GND are no problem. They are also well good enough to protect the OP. No extra protection is needed for the sens+ input, unless you want to protect against someone applying a large ( > 10 V) even more positive voltage to the input. If paranoid an diode to GND could be added.

Likely C25 and C44 are not need at all, if used they are quite small. The Capacitor C38 on the other side should not be that small - in my simulation it is in the 10 nF to 100 nF range. The main purpose of this cap is to improve the response with highly capacitive loads (e.g. > 5000 µF). This cap also helps that the OPs don't need to be that fast.

For a faster response it helps to have the analog to C38,R24 also in series with R34.

The offset at the output is slightly higher (e.g. 10% more) than the value set by R38,R39. To much of an offset could be a problem for the diode protection. I would stay at 50 mV which would still cover the range to -500 mV at the input. The OPx177 can also use the negative supply, so no problem to output negative values too.

For the values one should have R38||R39 = R27||R29 
and R24+R33 = R34  (+ supposed added resistor) at about 10 times R39
and R24 at about 5-10 times R33. So about R39=R33.
One could adjust  one of R27, R29 to a slightly larger value to ensure tolerances in the right direction. The 1:60 (or 1:100) divider for the offset could be around the right amount. To one could very well end up with R27=R39 and R29 left out or intentionally larger than R38.
So it is mainly 4 resistors that need low tolerance. The other 4 are less critical as they have a larger one in series or a smaller one in parallel.  A single resistor (e.g. R29) could be used to adjust for tolerances.
 

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #35 on: October 08, 2016, 03:25:01 pm »
For a faster response it helps to have the analog to C38,R24 also in series with R34.
Symmetrical design is always nice when it comes to opamps :)

The offset at the output is slightly higher (e.g. 10% more) than the value set by R38,R39. To much of an offset could be a problem for the diode protection. I would stay at 50 mV which would still cover the range to -500 mV at the input.
Sounds good. The main reason for needing any offset is just to avoid having to use the bottom codes of the ADC/DAC (where it might clip) - so it doesn't really matter what the offset voltage is.

For the values one should have R38||R39 = R27||R29 
and R24+R33 = R34  (+ supposed added resistor) at about 10 times R39
and R24 at about 5-10 times R33. So about R39=R33.
One could adjust  one of R27, R29 to a slightly larger value to ensure tolerances in the right direction. The 1:60 (or 1:100) divider for the offset could be around the right amount. To one could very well end up with R27=R39 and R29 left out or intentionally larger than R38.
So it is mainly 4 resistors that need low tolerance. The other 4 are less critical as they have a larger one in series or a smaller one in parallel.  A single resistor (e.g. R29) could be used to adjust for tolerances.
How do the values in the attached look for a starter? Once I've ordered the PCBs and am in the waiting period, I'll run some simulations and see exactly how it behaves - but it would be good to have some ballpark values to start with.

I've also moved those changes across to the layout, and it's looking very neat.
« Last Edit: October 08, 2016, 03:46:11 pm by cowana »
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #36 on: October 08, 2016, 05:40:17 pm »
The resistor values look reasonable. In the simulations I am currently looking at, I had about twice the resistance for all the values (and no offset, and no C25,C44).

For low currents the circuit might need an extra resistor from where you have the relay for output enable to ground This is to have enough working current for the transistor before that. Lower resistors on the output board could do a similar effect. If really needed one could like add those in parallel.

Otherwise the simulations look good. Only the cross over from CV to CC mode is relatively slow with the OPx177 or a similar OP. The slew rate limit can cause quite some overshoot at the current. For the voltage regulation the OP2177 seems to be fast enough - the limiting factor is the 2N3055 used for the output stage.

For the OP used for current limiting a faster OP would be an option to reduce the overshoot on CV to CC cross over.
Once in CC mode everything is fine - the output impedance is essentially set by the output capacitor and the load from voltage sensing.

The other option I tested is an PNP transistor to do anti windup. It simulates well - but it needs a negative set voltage and thus another OP or level shifting with extra resistors. Also the CC mode indication would than get slightly more complicated. One could use the voltage regulation side, so indicating that voltage regulation is not active.

Attached is a simulation file. At the output stage it still contains some extra transistors to make it a kind of limited 2 quadrant one. V5 = 0 turns this part inactive, with V5 at about 1.6 v is gets activated.


 

Offline dorin

  • Contributor
  • Posts: 38
  • Country: ro
Re: Korad KA3003D redesign and upgrade
« Reply #37 on: February 26, 2017, 03:03:12 pm »
Any update? I am interested in hacking mine too.
 

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #38 on: April 16, 2017, 04:55:10 pm »
Unfortunately I haven't had too much time to work on this project in the last few months - but here's an update!

I ordered the boards back in October, and received them about a month later. I soldered a few up (picture attached to this post), and got to work on the software. The controller is a pretty powerful Cortex M4 running at 100MHz, which I put freeRTOS on. This was my first real experience of using an RTOS in a project, and I'm very happy with it - it's made creating a fairly complex UI very easy.

The low-level software drivers are all complete, and the UI is almost entirely finished. Compared to the original Korad firmware, some of the features include:
  • 16-bit ADCs and DACs, mean it can be calibrated to a much higher precision.
  • The memories M1-M4 act like proper memories. I use the M5 indicator to show when you're in 'adjustable' mode (entered by selecting the VOLTAGE/CURRENT button) - once in there, you can adjust parameters as you want, and optionally save them to a memory by long pressing a preset button.
  • Adjusting voltage and current show which digit you're on by flashing it between normal brightness and high brightness, which feels much more natural and less jarring than it turning off (as it did before).
  • The default adjustment unit vary depending on if the output is on or off, to prevent accidentally making big changes with the output powered on.
  • The UI is now really responsive and reactive, and is great to use.
  • The speaker now plays proper tones (and 'clicks' as you turn the rotary encoder), coming from an actual DAC.
  • I've written a decent calibration routine, which steps you through what to do (open/short measurements) and allows a good calibration to be performed easily.
  • All display readings now use moving-average filters (with adjustable size) running much faster than the display update rate, so the readings are smooth.

Up to now, I've only focused on the software, and it's now in a state suitable for initial use. Having started to look at the actual power supply response, I think I might need a bit more help from the forum!

The good bit - with no or little load (<200mA), it performs exactly as you would expect. The DAC sets the correct operating point, the relays select the correct transformer tap, and the opamps regulate in CV or CC mode as required, with a nice clean output.

However, as the current increases, things get a little bit unstable. The scope screenshots attached show the oscillation in CV mode (~1.8MHz) and CC mode (~300kHz) when an 8 ohm resistive load is connected. I had previously run transient and steady state simulations of the entire circuit, but this had seemed to show it would be stable (although that could be due to my  limited simulation expertise). I've attached the PSpice netlist (generated in TINA) - you may need to rename the file to .cir.

My immediate thought was to decrease the high frequency gain of the OP2177 opamp stages by decreasing R26/R41 - although this might just reduce the transient response of the entire power supply, so I've slightly cautious of doing so! Currently R26 and R41 are 4k7, and C36 and C41 are 200pF.

Some thoughts on where to start tackling this would be appreciated!

Andy

Edit - just realised I missed the output capacitor (a 220uF electrolytic) off my schematics and simulations - it is there in reality! Still, the fact the simulation didn't even oscillate with that missing probably shows I'm definitely doing something wrong!
« Last Edit: April 16, 2017, 05:55:29 pm by cowana »
 
The following users thanked this post: prasimix, TheDane, dorin

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #39 on: April 16, 2017, 08:51:34 pm »
There are parasitic effects that can make the simulation behave different from real life. Typical such influences are ESR/ESL of capacitors and parasitic inductance of low value resistors.

Another point is that the simulation uses perfect values for the resistors. Even small deviations for R33 / R34 , R38,R39,R27,R29 can make the difference from stable to unstable. It is a good idea if R34 is slightly larger than R33.

The first usual modification would be increasing the capacitors C36 / C41 to maybe 500 pF or even 1 n. Thus making regulation slower and thus usually more stable. Decreasing R26/R41 could be an option too - it would need a more accurate simulation or just a few tests to see. The not shown values of R36 and R47 are also important as they scale the overall gain.
 

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #40 on: April 17, 2017, 04:29:20 pm »
Thanks Kleinstein, that's very helpful. From today's findings:

CV mode
I tried decreasing the feedback frequency (via C36), but that didn't have much effect - the issue in CV mode seemed to be the gain was just too high. I played with changing values a lot, and seemed to get the best results with:
  • Increase C36 from 200pF to 680pF
  • Reduce R26 from 4k7 to 2k2
  • Increase R36 from 2k2 to 6k8

With those values, the control loop is looking fairly good. However I noticed initial transients were appearing as huge signals on the output of the differential voltage amp, so I increased R33 and R34 from 287R to 1k7. This didn't seem to effect the control loops too much, but it seems a better ratio. See attached image (1) for the VSENSE signals and opamp output during a transient.

CC mode
CC mode was continuously oscillating. As with the CV section, I decreased the response rate and the AC gain:
  • Increase C41 from 200pF to 680pF
  • Reduce R41 from 4k7 to 2k2
  • Increase R47 from 2k2 to 6k8

The result of that seems fairly good - now both CC and CV mode behave as you'd expect! I'm not sure exactly what performance is possible from a design like this with the right tweaking - the performance is certainly adequate for my needs as it currently is. Having said that, I'd be up for getting it better if that's possible!

At low powers, the CV/CC switchover looks very neat, as shown in the attached image (3). One area that doesn't quite look right is the response when the output set to a high voltage (24v), then a high current load is applied, causing it to enter CC mode. Image (4) shows this going from 24V, 0A, CV to 20V, 2.5A, CC - the current initially rises (as expected), CC mode activates, the current overshoots slightly, then comes under regulation - but then takes around 1.5ms to settle on it's steady state value. This seems odd, considering the CC loop is obviously working, as it gets it within 10% of the final value within 20us!
« Last Edit: April 17, 2017, 04:36:51 pm by cowana »
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6442
  • Country: de
Re: Korad KA3003D redesign and upgrade
« Reply #41 on: April 17, 2017, 08:01:32 pm »
Increasing R33 and R34 to 1.7 K is likely a good idea. This reduces the high frequency gain of the differential amplifier - with the old values, there was a significant gain in the high frequency limit and thus the GBW of that OP might have been important. It might be a good idea to have R34 slightly larger than R33 - even if it is just to make sure that due to tolerances it is not the other way around. A slightly larger value for R29 would have a similar effect, to make sure the dorp at the shunt is not overcompensated. With the larger values of R33, R34 the one might be able to reduce R36 or C36 a little again. The change in R33 and R34 is also slowing down the loop.
The voltage loops still seems to be reasonable fast.

The turn over from CC mode to CV mode looks a little slow. If it needs to be faster one could implement a kind of anti windup. However a very fast acting current limit might not be desirable in all cases. Load steps close the the limit do need sort time currents beyond the limit to get a good CV mode performance.

For the high current case, this could be due to OP internal protective diodes, that pull down the set value and than cause a rather slow recovery due to C43*R51. So a smaller value for R51 might be worth a try. I am not sure if there is such a protection diode inside the OP (the DS does not tell). It might also help to add such a diode (or NPN transistor to reduce the load to set point ?), to prevent the inverting input of the CC mode OP to go much below the set voltage. This would be a way of limiting windup.

It also looks like the voltage is dropping only very little - so that high current example might be just borderline to engage CC mode at all.

To test the CV mode regulation, the more critical cases are usually a load with a low ESR capacitor (e.g. film or ceramic) in the 10 µF range and the case with a large low ESR electrolytic cap (e.g. 1000-5000µF) together with a small DC current. Here (especially with the large capacitor) some ringing after a step is normal - but it should not be excessive of cause.
 

Offline cdxa

  • Newbie
  • Posts: 1
  • Country: pl
Re: Korad KA3003D redesign and upgrade
« Reply #42 on: January 16, 2018, 03:00:35 pm »
Hello,
@cowana, do you plan to share mcu firmware? or maybe you sell all as a Korad upgrade kit?
 

Offline cowana

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: gb
Re: Korad KA3003D redesign and upgrade
« Reply #43 on: January 19, 2018, 01:07:30 pm »
Hello,
@cowana, do you plan to share mcu firmware? or maybe you sell all as a Korad upgrade kit?

My main reason for doing this project was as a learning experience - I've learnt a lot doing it, and got a compact bench PSU with all the features I want as a result.

I don't think anyone else would benefit from the 'upgrade' - mainly due to the high the BOM cost of my design (just the opamps, ADC and DAC together are over £15), and the significant time/work required to install it (modifying the rotary encoder, adding cable looms, cutting additional holes in the rear panel for the output connectors). If you're after a high performance bench PSU, there are much simpler options if you're not interested in designing your own!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf