An investigation about ripple current of Kunkin KP184

[doktor pyta]

I tried to check what is causing annoying current ripple in Kunkin KP184 DC electronic load.

Reference to the older thread:
https://www.eevblog.com/forum/testgear/fixing-an-manufacturer-issue-of-the-kunkin-kp184-electronic-load/

Setup:
Kunkin KP184 ver.04 with C58=10n installed by the manufacturer
Tek P6302+ AM503 current probe. Bandwidth limited to 5MHz.
The oscilloscope is AC coupled.
Linear bench power supply.
Conditions: current set on KP184 is 100mA; voltage applied 5V


Sanity check: noise of the system



Current ripple: as bought



Current ripple: without front panel



Current ripple: without front panel, without fan



Current ripple: without front panel, without fan, without 2nd MOSFET assembly



Now after staring at the PCB and some experimenting I've added a thick ground return path which looks like this



Current ripple: without front panel, without fan, without 2nd MOSFET assembly, with ground return mod



I've also added 1000uF/16V low ESR capacitor in parallel to 470uF cap on the front PCB.
Multiplexed LED display creates current spikes.
I tried to lower them by adding 10uF MLCC but without effect.
Photo of this mod below



Current ripple: with front panel C mod, without fan, without 2nd MOSFET assembly; with ground return mod



Current ripple: with front panel C mod, without fan, with 2nd MOSFET assembly; with ground return mod


Final:
Current ripple: with front panel C mod, with fan, with 2nd MOSFET assembly; with ground return mod



Conclusion:

1. due to 2 layer layout the current return path is not optimal. It can be fixed relatively easily.

2. most of the ripple comes from the fan. The rotor's magnet spins only 1cm away from current sense resistors and induces EMF in it.
It also creates HF content. I tried to add better decoupling but only electrostatic shield make things slightly better however it lowers airflow.
The fan is an integral part of the mechanical assembly.
So only solution that comes to mind is disconnecting the fan and adding another fan as far as possible in the enclosure,
however full current rating probably won't be met due to lower airflow

3. the designers didn't bothered these problems. The load is specified for Imax= 40A and even at 5A these artifacts are barely visible.

Right now I don't see any easy and cheap solution to the fan problem, so I'm staying with ground return mod and C mod applied.

[doktor pyta]

Additional photos + rare bottom side view.

[doktor pyta]

There is one interesting thing about KP184.
The crystal oscillator signal is not present.
I'm sure I'm not killing it by connecting 10:1 probe.
They are probably using internal RC oscillator.
I'd appreciate if someone confirm this.

[CDaniel]

If the fan is inducing that noise than the electronic is not well design and should be fixable somehow if you want and have time... this is not normal  Maybe missing some small caps in the current amplifier ...

[doktor pyta]

There is one possible solution to the mentioned problems: disconnect internal fan and install another fan outside enclosure, on the rear panel.

[AlexVlc]

Hi,

My unit have noise with or without load, tested with and without the capacitor and with or without the fan, noise is present always.

[guenthert]

Hi,

My unit have noise with or without load, tested with and without the capacitor and with or without the fan, noise is present always.

(Attachment Link)

    Is that truly the noise from the load?  I've seen such injected from other devices (here a Ethernet over Powerline adapter is the greatest offender by a large margin) into the power line.  I'd have a 2nd look with the usual suspects turned off.

[AlexVlc]

    Is that truly the noise from the load?  I've seen such injected from other devices (here a Ethernet over Powerline adapter is the greatest offender by a large margin) into the power line.  I'd have a 2nd look with the usual suspects turned off.

I mean that I have noise (around 50mV) only placing the probes in the input connector without connecting it to a load, the noise comes from the internal circuitry of the unit and persists without the fan and with or without the capacitor. I'd like to spend some time and find the problem but I don't have time.

[wolzlu]

Hi,
AlexVlc's picture shows one problem of the KP184. I have a KP184 VER:04 202007 and it also radiates a lot.
The layout of my unit is slightly different as in the photos of doktor pyta, hence the date code (202007). The buck converter of the fan and RS-232 section is too noisy. I measured about 3.8Vpp on the input side of the buck converter (internal fan not running). The measurement was not very accurate (100MHz scope and passive DIY probe), but these spikes are huge and get everywhere. Right next to the switching IC U24 (AOZ1282CI) is an unpopulated footprint for an input capacitor C34. Why?
I tied:

• Adding C34 100n input capacitor
• Adding short wire to reduce the unnecessary ground loop
• Adding snubber across the Schottky diode D13 (MDD DSK34) to reduce ringing at about 120MHz
• Adding output capacitor

Not tied:

• Add resistor in series to bootstrap capacitor C26
• Change diode D13 and inductor L1
• Use external buck with proper filtering

After all I came down to 304mVpp respectively 600mVpp with fan at full speed on the input side (only power pins of fan connected).
I propose to disable U24 (via pull down) and use a linear regulator instead  .

[Hydron]

I had some similar problems with my DCL6104A (same form factor to KP184, but different manufacturer, and a 500V version), thought I'd share incase it was of interest.

There were 2 ripple issues, the first is the fan next to the current shunt as seen here - is little enough (less than 10mA) that I decided to ignore it, and another is from poor layout of the power supply AC-DC rectification and regulation section.

The second issue is also similar to one you've identified here with the KP184 - with only a two layer board (quite different layout as this one) and some poor layout choices some of the 100Hz ripple interacts with the sensitive current sense signal (essentially I*R drops on the ground traces shift the current sense reference around). I managed to fix this issue with some PCB mods - see this thread for details: https://www.eevblog.com/forum/testgear/et5410-or-dcl6104-or-a-third-programmable-electronic-dc-load-in-the-budget-cat/

In terms of the fan driver, the DCL6104 does use linear regulation (BJT with a op-amp control loop), the noise from the fan is all magnetic (as mentioned) and auditory (hence my mod to reduce the speed, it was _way_ too loud at idle). I think the only switcher in the whole device is 5 or 3.3V for the front panel MCU/screen (can't remember exactly).

If I dig up any captures of the ripple I'll add them to a post in that thread, but PCB mods mostly solved it (though there weren't any HF issues from a switcher etc like seen here).

[vladant]

Hi, is not that black wire on the photo bypasses current shunt ?

[graybeard]

This video should start on the portion about load current noise.

I have seen similar noise in my Beich CH9720CU Electronic Load.   After doing a comprehensive test and tear down I believe it is due the digital feedback loop controlling the load.

In the Beich load I see two distinct almost periodic events; a faster one I think is do to the single A/D sampling the Voltage and current, and a slower one that I believe is feedback sending the current level update to the D/A driving the op-amp/MOSFET load cells.  The fact that it is not perfectly periodic is due to the regulation/measurement MCU having other duties like exchanging information with the control/GUIe interface MCU.

In the Beich load the circuit is laid out on two-sided PCBs (I assume to reduce cost) and lack controlled impedance transmission lines.   Thus coupling from the fast digital signals into the analog portion are happening as well as reflections from the fast edges of the digital signals.  I think the designer got around the reflections by giving the data time to settle before clocking into the next stage.

From your photos it appears the Kunkin load also uses two sided PCBs, so the issues may be similar.

[vladant]

I measured voltage on CD7/12 capacitors and just curious if it is a design "feature" to have 19V on 16V caps .

So far I have found out that KP184 design has 5 major flaws:
1) +-12V supply and its ripple and noise.
2) +5V supply, its (capacitance) impact on +-12V and in turn the front LED panel impact on +5V (which is devastating).
3) FAN supply circuit and it noise.
4) Oscillation in U16 TL074 opamp #4 in Constant Voltage (CV) mode with an input resistor > 5Ohm.
5) It exists.

Resolution for flaw #:
1) Cut off +5V from main transformer by removing diodes D15/16. Now +5V rail is not powered.
2) Install another 7-12V 5W transformer to power up +5V rail (fortunately the case has plenty of room). Connect its output via diode bridge to CD12. Insert 50-100 Ohm resistor in J12 line 1 (+5V to front panel).
3) Install unpopulated C34 100nF (if you have it on your PCB version, near R25) OR install LDO with 12-13.5V output instead of existing switching DC-DC.
4) Connect input+ to R88/R9 by 1MOhm resistor and 500-1000pF capacitor (>200V) in series.
5) Cannot be helped.

Please share your results.

[PaWill68]

I contacted the developers. C34 must be installed without fail. Rated 220nF 0805 MLCC. Can be used from position C17. When changing the fan power circuit, they forgot to install it in the first batch of products.

[aristarchus]

Can this C34 position on pcb be pinpointed on a photo ?
Tried for some time to find it and it was not easy.

[PaWill68]

Can this C34 position on pcb be pinpointed on a photo ?
Tried for some time to find it and it was not easy.

[PaWill68]

Friends, there is good news. KUNKIN has released a firmware update:
1) The voltage is read by reading one register, and the voltage value is obtained after smoothing and filtering, so there will be no sawtooth shape. The voltage acquisition rate before smoothing and filtering is 470SPS, and the voltage value after smoothing and filtering, the refresh rate is relatively low, it is 10SPS.
2) Instructions to set the battery discharge mode, half-current discharge is enabled and can be saved

[aristarchus]

Nice.
Thanks @PaWill68.
Though in my VER:04 KP184 there is no such cap C34, so I hope in VER:04 they have it solved.

Regarding the firmware update, is there any link to see it, download it and any instructions on how to upload/flash it in the KP184 ?

[Noy]

Hmmm, i have a V3 version here there is no C34..

I have an unmounted C60 and C58..  By the way whats now the bes case for C58? Unmounted or 10nF ??V on this place?

Is this new firmware also for V3 versions?

[PaWill68]

This firmware update is only for KP184 manufactured in 2019, 2020 and 2021, the model of the analog-to-digital conversion chip is AD7799. All actions are at your own peril and risk. If there are no problems, it is not recommended to update the firmware, as the changed location is not necessarily used by the user.
For questions about software bugs you can write to 64908476@qq.com

[PaWill68]

The problem of fan noise appeared when the design was changed later. Before the change, the power supply of the fan was not regulated, and the power supply voltage would rise to more than 17V due to the unstable grid voltage and burn the fan.

[Noy]

I checked ma V3 version it has an ADS1232 so this FW is probably not for my device.
Interesting observation for the FAN Supply. Maybe i should add an TVS Diode? Or i hope that our grid voltage ist stable enough here in germany...

So still the question left should i mount 10nF with xx V rating on C58? Or should i left it unpopulated... I dismounted the original cap regarding to the earlier observations / solutions..

[PaWill68]

I checked ma V3 version it has an ADS1232 so this FW is probably not for my device.
Interesting observation for the FAN Supply. Maybe i should add an TVS Diode? Or i hope that our grid voltage ist stable enough here in germany...

So still the question left should i mount 10nF with xx V rating on C58? Or should i left it unpopulated... I dismounted the original cap regarding to the earlier observations / solutions..

C58 can be omitted if C57 is present. Perhaps the scheme has changed since that time.
For ADS1232 chip, the firmware will not work and will damage the device.

[dl6lr]

This firmware update is only for KP184 manufactured in 2019, 2020 and 2021, the model of the analog-to-digital conversion chip is AD7799.

Updated my KP184 Board VER:04 20191111 with AD7799, previous firmware version shown was 1801. I had to make the following changes:

1. Baudrate 9600 only, tried higher settings with no success, maybe the comm settings are not effective in the bootloader
2. Device was protected, had to first erase the flash, otherwise the flashing did not succeed. It seems it tried to auto-erase the chip, but due to 3. the process was unsuccessful.
3. After Erasing the flash, had to reboot the device with pins 8+9 bridged to get the bootloader running again, then flashing was possible.

Regards, Bernd

[PaWill68]

Possibly related to the quality of the cable. I have RS232-USB on CH340 chip. It works at 115200 normally, but for the firmware I had to set the speed to 9600.
The manufacturer recommends cables based on the PL2303 chip. They are more stable.