Cheezeball DC Load: DL24P: Pump, or Dump ???

[Dave_g8]

For interest a few more test results for the DTL150.

Calibrating the 3A current at 1.5A (see CAL_1A5.JPG). This gives accurate results down to an ISET of 0.08A, actual load current = ISET/2 = 0.04A.

Oscilloscope traces of the MOSFET Vgate voltage after recalibrating at 3A
DS1Z_CR_0A_.png      CR Mode, 0A, ON. The Vgs threshold is approximately 4.5V.
DS1Z_CC_BTN_OFF2.png   CC Mode, 2A, ON to OFF using ON/OFF button. Off time <10ms.
DS1Z_CC_BTN_ON.png   CC Mode, 2A, OFF to ON using ON/OFF button. On time 3s (0A to 2A)
DS1Z_CR_BTN_ON.png   CR Mode, 2A, OFF to ON using ON/OFF button. On time 1.2s (0A to 2A)
DS1Z_CC_PSU_OFF.png   CC Mode, 2A, ON to OFF when switching off the PSU (see note 1)

Notes
1. In this case, when the PSU is initially turned off, the Vgate voltage goes to maximum ~11V until the firmware detects that there is no sense voltage and then turns the MOSFET off in <10ms.
2. I have not fitted Zener clamping diodes between gate and source at the moment.

Regards, Dave

[jtr1962]

So apparently the DTL150 does soft start. I think that accounts for the greater stability. The DL24/P seems to jump up to its current setting very quickly. The end result of that is less stability, and sometimes some overshoot.

I put a 100uF tantalum cap over each of the sense resistors. That does seem to help a bit. On the 1 amp setting, it'll seem to stay between 0.999 and 1.000 for a greater percentage of the time. Ceramic caps would probably work just as well. There does seem to be noise on the sense resistor. This accounts for a greater value sense resistor giving more stability. The noise ends up being a smaller percentage of the sense resistor voltage at any given current setting.

I recalibrated my unit to take into account the time base error. I used 2.991A for the 3A calibration. I was curious to see how accurate I can get the device compared to a standard (i.e. the capacity testers I built which are ~0.1% accurate).

I connected to a power supply and stopped the DL24 as soon as the indicated capacity read 2000 mAh. I tried various currents from 0.5A to 10A.  Here are the actual capacities according to my tester, and the error:

0.50A - 2001.2mAh (-0.06% error)
1.00A - 2000.6mAh (-0.03% error)
2.00A - 2000.6mAh (-0.03% error)
3.00A - 1999.1mAh (+0.045% error)
4.00A - 1999.4mAh (+0.03% error)
5.00A - 1999.5mAh (+0.025% error)
7.50A - 2000.6mAh (-0.03% error)
10.00A - 2004.7mAh (-0.235% error)

These are great results. The device can match a standard to well within 0.1% over much of its range. This is what I was after. I didn't test any higher than 10A because none of my power supplies could put out that much current. Also, given how hot the reverse protection diodes were getting, plus the track size, this is really a 10A device without doing some serious modification. I'll try to go to 30A or more when I buy of the DIY devices. I can use a few A123 26650 LiFePO4 cells in parallel to supply that kind of current.

[TruslowPJ]

Well I got mine without the cooler. had a bit of a difficult time installing my H50 watercooler to it, due to the z height being significantly different than an intel CPU. I made my own backplate from 3mm fiberglass sheet since the metal backplate seemed unwise.
my observations so far are that it can definitely keep the temps under control on the fet, and have tested it to 100W. It seems mine has a reclaimed fet, but I haven't had a reason to pull it yet. I did add a 12V bidirectional TVS diode to the gate before powering it up the first time. I've also observed that without a cooler that provides airflow over the diode, it gets pretty hot at 5A, around 70 degrees C with 0.42V drop across it, which is in line with the datasheet so it might be a legit part. some napkin math says it should be around 30C/W and at 5A that's around 2W of heat, so it checks out. 10A might be possible, but I have half a mind to just short the diode, or use a Pmos, a Zener, and a resistor to make a better reverse polarity detector. maybe another of those 12V TVS diodes could be used in place of a zener, though I don't know if they are really designed to be used that way.
I wish there was a way to change how the fan output works. It would be nice if it was enabled whenever the load is enabled, as even at a watt of heat, I still want the pump running. I might just make it powered all the time.

I also threw out the included power adapter and used my own. It's cables were extremely suspect, which was confirmed when I tore the mains side by hand

[TruslowPJ]

Before I even turned it on, I installed a 12V bidirectional TVS diode on the gate by scraping some of the coper away with a hot soldering iron. Got it set up with an old corsair water cooler. had to fashion a backplate from 3mm fiberglass sheet and get creative with the mounting situation, but I got it working. at 5A, the diode was already getting hot, like 85 degrees hot. I was considering shorting it out, but like the protection, then I considered a mosfet based reverse protection, but PMOS would get expensive to get one with less than 0.4V drop at 10A with over 60V of breakdown voltage with not much gained, and NMOS would be a pain to drive the gate, decided instead to just deal with the heat. adding a bunch of 14AWG copper wires to spread the heat more effectively really helped (that's 2.5 square mm wire for you metric folks). dropped the diode temps by about 25 degrees C. I am seeing 62 degrees at 20V 5A, and at 5V 10A, it was about 100. hot, but not dead diode hot. I imagine it would not run as hot with airflow over it, but given that I am water cooling and there isn't any airflow there until I build an enclosure, it'll do for now to run comfortably at 5A, with 10A being a bit toasty.
burned it in for a while at 20V, 6A with the stock fet. I'm getting 44 degrees with the thermistor placed against the side or legs of the fet, and 32 degree water temperature
next will be a 3d printed enclosure that will direct air from the radiator over the board and diode, at which point I might hook it up to my 48V 10A power supply and let it rip with 200W or so.

[beanflying]

Moooooooo 



** Only works in Metricated countries 

[jdev99]

I got the plain DL24 without heatsink, but it arrived with a damaged screen. 
Asked for a refund or replacement, will see what happens,
Anyway, there are replacement screens available, I think somebody asked about it here earlier.

https://www.aliexpress.com/item/1005003275856527.html

I also found this software, not sure if somebody already tried it:
http://www.hardandsoftware.net/DL24Download.html

[vk3em]

I also found this software, not sure if somebody already tried it:
http://www.hardandsoftware.net/DL24Download.html

I tried it but I could not get it to work properly. I have a DL24MP and the factory supplied software has an option for different Protocol versions so maybe this software only supports one (the original) protocol type. This the factory software repository : https://www.mediafire.com/folder/1c04afq923397/A#yhumeijk86fj8

Luke

[Dave_g8]

After doing tests on the DTL150 electronic load, I also purchased an Atorch DL24 unit (£19 delivered to the UK) and repeated a few of the tests.

- MOSFET Vgate (see attached traces)
DS1Z_CC_0A01.png      CC Mode, 001A, ON. The Vgs ON threshold is approximately 3.7V.
DS1Z_CC_BTN_OFF1.png   CC Mode, 1A, ON to OFF using ON/OFF button. Off time <50ms.
DS1Z_CC_BTN_ON1.png   CC Mode, 1A, OFF to ON using ON/OFF button. On time <50ms.
DS1Z_CC_PSU_OFF1.png   CC Mode, 1A, ON to OFF when switching off the PSU (see note 1)
DS1Z_CC_PSU_OFF2.png   CC Mode, 1A, ON to OFF when switching off the PSU (see note 2)

Notes
1. Prior to fitting the Zener diodes between the MOSFET gate and source.
In this case, when the PSU is initially turned off, the Vgate voltage goes to maximum ~11V until the firmware detects that there is no sense voltage (500ms) and then turns the MOSFET off in <100ms.

2. After fitting back-to-back 6V8 Zener diodes between the MOSFET gate and source.
In this case, when the PSU is initially turned off, the Vgate voltage goes to the clamped voltage of ~7.5V until the firmware detects that there is no sense voltage (500ms) and then turns the MOSFET off in <100ms. As previously noted, the Zener diodes will ensure that Vgs is well within the specified limit of +/-20V, although there is no evidence that this is exceeded without the Zener diodes fitted. Clamping to 7.5V may provide some degree of current limiting.


- CC Setting
Unlike the DTL150, the DL24 worked as expected down to an ISET current of 0.01A.
As previously reported on the DL24, the current and voltage reading do have some variation, but seem perfectly acceptable.


- Serial Interface
The serial interface seems to work more or less as expected, although there are some quirks with the Atorch PC software and some anomalies with the value formats for control data compared to available information on the Internet.
The DL24 automatically sends out a 36-byte frame every second.
The DL24 responds with 7-byte frames to query frames (6-byte) sent from the PC software.
The DL24 parameters can be set with the control frames (6-byte) sent from the PC software.
The PC software allows selection of 36-byte (Protocol 1) or 7-byte (Protocol 2) processing.

Oscilloscope traces for the serial interface are attached.
DS1Z_Serial_36Byte_Auto.png : DL24 to PC, Auto sent 36-byte frame every second.

DS1Z_Serial_36Byte_SOF.png : DL24 to PC, shows the start of the 36-byte frame.
Structure = 0xFF 0x55 0x01 0x02 < Voltage 24 bits > < Current 24 bits >
Voltage = 0x000033 = Dec 51, Voltage (V) = Dec value / 10 = 5.1 V
Current = 0x0003E9 = Dec 1001, Current (A) = Dec value /1000 = 1.001 A

DS1Z_Serial_CTRL_Off.png : PC to DL24, turn load OFF.
Structure = 0xB1 0xB2 CMD D1 D2 0xB6
CMD = 0x01 = Enable/Disable
D1:D2 = 0x0000 = Disable Load

DS1Z_Serial_CTRL_On.png : PC to DL24, turn load ON.
Structure = 0xB1 0xB2 CMD D1 D2 0xB6
CMD = 0x01 = Enable/Disable
D1:D2 = 0x0100 = Enable Load

DS1Z_Serial_CTRL_SetI_1A5.png : PC to DL24, set load current.
Structure = 0xB1 0xB2 CMD D1 D2 0xB6
CMD 0x02 = Set I
D1:D2 = 0x0096 = Dec 150, Current (A) = Dec value / 100 = 1.50 A


Summary
Overall I think that the DL24 is slightly better than the DTL150, but both are excellent for the cost. The DL24 seems to provide CC control down to lower currents and the serial monitoring/control interface could be useful.

I have only tested currents up to around 5A and power of 70W. With the supplied heatsink and fan, both the MOSFET and protection diode seem to be at acceptable temperatures.

Regards, Dave

[Dave_g8]

Added missing attachment.

[Dave_g8]

For information, one observation with the “ATORCH Electronic Load Test Software  V1.0.2” is that the “Set Current” values appear to be sent incorrectly.

I believe the command structure should be as shown in the attached PX-100 reference. For the “Set Current” and “Set cut-off voltage”.
D1 = Integer part
D2 = Fractional part (Range should be 0 to 99)

This appears to be correct for the “Set cut-off voltage”. However for the “Set Current” the PC software sends the value as a 16-bit unsigned integer.

Example: Manually entering a value of 3.00 A in the PC software. The 16-bit value sent = Dec 300
D1 sent = Dec 1   This should be 3
D2 sent = Dec 44   This should be 0
The DL24 firmware interprets this as 1.44 A.

The DL24 firmware seems to interpret values correctly up to 2.55A even though the command value is sent incorrectly.
For 2.55A, the PC software will send a 16-bit value = 255, D1 = 0, D2 = 255.
I expect the calculation will be Current Setting (A) = D1 + (D2 / 100) and there will be no range check on the D2 value.

Regards, Dave

[Dave_g8]

DL24 serial interface

The MCU and Bluetooth devices are powered from 3.3V. The CH340G is powered from the Micro USB (5V). (See attached images).
Assuming that there is no Bluetooth connection established, the Bluetooth device appears to drive a logic ‘1’ on its TXD pin.
When the USB (5V) port is connected, there is a potential conflict between the CH340G TXD (output) and the Bluetooth TXD (output).

Measured voltages on the CH340G and Bluetooth pins are shown below.
CH340G Driving Logic ‘0’:
CH340G TXD      0.5 V
Bluetooth TXD      2.5 V
This means that the current through the 120 Ω = (2.5 – 0.5) / 120 = 17 mA. It is likely that the devices can sink more current than they can source and explains why the logic ‘0’ from the CH340G is dominant.

CH340G Driving Logic ‘1’:
CH340G TXD      4.2 V
Bluetooth TXD      3.3 V
This means that the current through the 120 Ω = (4.2 – 3.3) / 120 = 7.5 mA. In this case the CH340G is sourcing current via the 120 Ω connected to the Bluetooth device which is also outputting a logic ‘1’ at 3.3V.

Both of these situation are not too good, but the current is at least limited by the series resistors. In addition the RX input on the MCU is likely to be higher than 3.3V, but should be limited to (3.3V + 0.7V = 4.1V) by its input protection diode.

Without the USB port connected, the logic levels on the TX and RX data lines appears to be 0V and 3.3V, although there may be some leakage current flowing into the CH340G pins.

On the DL24 unit I have, the TX and RX LEDs are connected to the V3 pin of the CH340G which is its internal LDO regulator output. Without the USB connected, the TX and RX LEDs still light (dim) when there is serial data activity using the Bluetooth interface.

Also, reports on some of the other threads on the forum indicate that the USB chip gets damaged if the USB port is connected before powering the DC 12V input. This could be the case, but possibly related to the point made by Dick Geier on his control software page related to switched mode “wall wart” supplies (Reference: Dick Grier http://www.hardandsoftware.net/DL24Download.html).

Regards, Dave

[vk3em]

For information, one observation with the “ATORCH Electronic Load Test Software  V1.0.2” is that the “Set Current” values appear to be sent incorrectly.

I found the same until with this software until I changed to the specific version for my unit (https://www.mediafire.com/folder/1c04afq923397/A#yhumeijk86fj8)

Note: It would be good to see your comparisons with the 4 MOSFET DL24MP version here https://www.aliexpress.com/item/1005004644040863.html?spm=a2g0o.order_list.order_list_main.4.4e771802QW1Y74

Luke

[Dave_g8]

I finally managed to get an ESP32 module working in Bluetooth Serial Master mode, providing monitoring and control of the DL24 via WIFI using Node-RED.
I used a mixture of the 36-Byte and 7-Byte protocols to allow full control/monitoring over my local WIFI network.
For anyone interested in Node-RED, please see attached link https://nodered.org/about/
Regards, Dave

[JohnC]

Hi everyone,

I am currently beefing up my DL24 (bare board) and wanted to ask if anyone has tried using an "Ideal Diode" module as reverse polarity connection instead of the STPS41H100cg which get bloody hot at high power levels?

I found 2 particular (cheap) modules on AliExpress, both designed to go on the high-side so that would mean a straight swap. I've attached small images of the PCBs below.

SZ-D40100
Module rated <72V forward, <40V reverse (hmmm?)
Module rated 40A (1xMOS) or 60A (2xMOS)
https://www.aliexpress.com/item/1005005261264290.html
Doesn't need low-side connection (so, 2-wire)

Alternative using ON FDBL86062
Module rated 9V-80V forward, Mosfet vDS 100V
Module rated 50A
Also needs a ground connection (no problem here?)

These are intended for anti-backfeed on Solar Arrays. Neither module meets to max 100V rating of the DL24, however I am not sure that is realistic anyway, is it?

There's also a method to use a P-ch Mosfet in reverse on the high-side (schematic also attached).

The challenge of these ideas in an Electronic Load seems to be at low voltages (eg: single cell), and requires an additional Charge Pump mosfet driver to ensure that the current gets switched on properly.

My numerous other loads don't even have reverse polarity protection, or only pull ~5A so a Schottky works reasonably OK.

Any of you guys got any thoughts or suggestions?

[Pukker]

I think the "Ideal diode" without Ground connection will give problems, because the current will be interupted periodically.
It periodically turns off for a moment and uses the 0.7V drop of MOSFET's body diode as a power source.
This is needed for the operation without an Ground connection.
This module will be on for 98% of time. The DL24 will recognize a voltage dip and shuts down if
the minimum voltage is set to turn of.

I think the module uses the lm74610 chip.

https://www.ti.com/lit/ds/symlink/lm74610-q1.pdf

[OneGeekGuy]

I have received today the "new" version of the Atorch DL24P 150W. I did not have time to go through tall this post, so I might say some things that have been said but just in case 

Some pictures of the unit, it is the 4MOSFET version, some appreciations

• The MOSFETS are not same manufacturer looks like, and I think they are re-used components I would bet, but so far look OK
• I think another guy on this forum already posted that the shunt resistors are not equal, I have pair of 22R and pair of 25R
• The PCB was not really clean, but this is minor issue

I was doing short testing since it is a bit late here, the current measurement looks quite accurate showing:

Input Voltage 12V
DL24P Setting: 100mA
DM3058E Measurement: Min 100.77mA , Max 101.395mA

Input Voltage 12V
DL24P Setting: 250mA
DM3058E Measurement: Min 251.17 , Max 252.23mA

Input Voltage 12V
DL24P Setting: 500mA
DM3058E Measurement: Min 503.12mA , Max 503.65mA

Input Voltage 12V
DL24P Setting: 1000mA
DM3058E Measurement: Min 1005.7mA , Max 1006.5mA

Input Voltage 12V
DL24P Setting: 1500mA
DM3058E Measurement: Min 1508.1mA , Max 1508.5mA

I would say the minimum current that can be set is around 10mA, below 10mA the drift increase exponentially.

I will do some more measurements within these days, I will upload the results.

I need to check as well the PC application, till now I used the build in screen.

Someone know if there is a way to calibrate in case needed this new model?

I keep you posted!

[OneGeekGuy]

I was doing some measurements this morning, while the OFF/ON transition on my unit is OK and smooth, I observed an un expected behavior on ON/OFF transition which is creating a pulse or step while witching OFF the load. I tested only in CC Mode with different currents, and looks like it is always happening, I also measured the gate of the transistors and it is following the current shape.

In Purple channel or channel two you can see the current draw by the load and on channel 1, in Capture 1 and 2 the voltage applied to the load and in Capture 3 the Mosfet gate voltage in respect to ground.

Looks like this is created by the SW somewhere enabling when transitioning between ON to OFF.

Also curious thing is that below lets say... 30mA aprox the Load is taking some maybe seconds to enable itself. I need to check at this low currents changing the setting to low current.

Another thing, and this is expected is that the MOSFET do not heat or power dissipate similar, it is not balanced till now I did not measure the temperature of each but just touching you can feel it.

Thanks.

[JohnC]

I think the "Ideal diode" without Ground connection will give problems, because the current will be interupted periodically.
It periodically turns off for a moment and uses the 0.7V drop of MOSFET's body diode as a power source.
This is needed for the operation without an Ground connection.
This module will be on for 98% of time. The DL24 will recognize a voltage dip and shuts down if
the minimum voltage is set to turn of.

I didn't realise about that voltage detection. So that is certainly a consideration with the 2-wire module. I wonder how the 3-wire module operates? I might just buy one for test, since I can always reuse it in a battery charger circuit anyway.

I think the module uses the lm74610 chip.

Yeah I agree, and the pinout seems to match. I had already found that chip previously, and last night I downloaded heaps of module images to try and decipher the (ground off) chip markings. I failed and gave up looking, then just posted here instead

The big question I have is that, since you reckon the reverse diode is needed for other things... would paralleling up the existing diode be good enough to share some of the load, or is that just asking for trouble due to component tolerances?

[OneGeekGuy]

I was doing some measurements this morning, while the OFF/ON transition on my unit is OK and smooth, I observed an un expected behavior on ON/OFF transition which is creating a pulse or step while witching OFF the load. I tested only in CC Mode with different currents, and looks like it is always happening, I also measured the gate of the transistors and it is following the current shape.

In Purple channel or channel two you can see the current draw by the load and on channel 1, in Capture 1 and 2 the voltage applied to the load and in Capture 3 the Mosfet gate voltage in respect to ground.

Looks like this is created by the SW somewhere enabling when transitioning between ON to OFF.

Also curious thing is that below lets say... 30mA aprox the Load is taking some maybe seconds to enable itself. I need to check at this low currents changing the setting to low current.

Another thing, and this is expected is that the MOSFET do not heat or power dissipate similar, it is not balanced till now I did not measure the temperature of each but just touching you can feel it.

Thanks.

I have updated the FW to the latest Version 2.0, and the issue for the pulse when switching off the load is still there...., I have contacted them just in case....they want to fix the issue, because should be a FW/SW issue...., looks like it is not happening in all the versions they have.

Keep you posted.

Thanks!

[Pukker]

I have updated the FW to the latest Version 2.0, ..................................

Hi OneGeekGuy,
Updated FW, how / where to become that firmware?
asked Atorch for, never had an reply.

[Aurgelme]

I have updated the FW to the latest Version 2.0, ..................................

Hi OneGeekGuy,
Updated FW, how / where to become that firmware?
asked Atorch for, never had an reply.

I have read through the thread and see there are several versions discussed here, not just the one the thread is marked with DL24P.

According to the information I have received, there is a firmware update for the DL24EW, which you can find here
I don't see any firmware file, maybe it's in the zip file (

http://en.atorch.cn/NewsDetail.aspx?ID=55

As for the firmware for the DL24P, it is not supplied as downloadable, only as fully programmed ship, which can be ordered from ATORCH with version firmware version 5.1 (when I checked a while ago)

I am working on a project with 4 parallel DL24P boards for 100A testing rig and have boards with firmware 5.1

[Pukker]

As for the firmware for the DL24P, it is not supplied as downloadable, only as fully programmed ship, which can be ordered from ATORCH with version firmware version 5.1 (when I checked a while ago)

I am working on a project with 4 parallel DL24P boards for 100A testing rig and have boards with firmware 5.1

No, no firmware in the Zip file.
100A with four DL24P, wow. Keep your fire extinguisher nearby. :
I have two DL24P units and when I need more than 100 Watt discharge,
I use these 2 parallel.
One has FW 5.0 and another FW 5.1.
Difference not very much.

Made some modifications: Changing Mosfets to original ones,
(mine was delivered with used fake Mosfets) (could see they where screw mounted before)
Also placed an zener to reduce spikes on the gate.
Placed an small cooling blok on the reverse protecting diode.
Using them with powerfull tower PC coolers.
I try to stay under 120W of power, staying safe......
For the money they are nice loads.
Accuracy is not bad, after calibration fine.

Software from Atorch I don't like, I prefer Testcontroller, find it here at the forum. Works fine.
https://www.eevblog.com/forum/testgear/program-that-can-log-from-many-multimeters/
DL24 Units are used and discussed there.

[Aurgelme]

No, no firmware in the Zip file.
100A with four DL24P, wow. Keep your fire extinguisher nearby. :
I have two DL24P units and when I need more than 100 Watt discharge,
I use these 2 parallel.
One has FW 5.0 and another FW 5.1.
Difference not very much.

Made some modifications: Changing Mosfets to original ones,
(mine was delivered with used fake Mosfets) (could see they where screw mounted before)
Also placed an zener to reduce spikes on the gate.
Placed an small cooling blok on the reverse protecting diode.
Using them with powerfull tower PC coolers.
I try to stay under 120W of power, staying safe......
For the money they are nice loads.
Accuracy is not bad, after calibration fine.

Software from Atorch I don't like, I prefer Testcontroller, find it here at the forum. Works fine.
https://www.eevblog.com/forum/testgear/program-that-can-log-from-many-multimeters/
DL24 Units are used and discussed there.

I am experimenting with a raspberry pi with node-red to have contact with all 4 cards and control everything, as well as have a summation of all the data.

Otherwise, I have used the DLP24 for a long time at close to max watts, and have only changed one mosfet, where it was my fault, forgetting to turn down the amperes before connecting an e-bike battery.

Now I have changed the mosfets to linear mosfets IXTK90N25L2
Removed pcb soldering mask, tinned and soldered in 0.7mm copper plates.
4x reverse diodes in parallel and water cooling, as well as cooling from the back of the pcb.
Welded me an aluminum enclosure for assembly.

[OneGeekGuy]

I contacted Atorch, directly and they answer me that latest FW for the DL24MP which is my load version is 2.0, link here:

http://en.atorch.cn/NewsDetail.aspx?ID=29

The DL24MP ia different to DL24P I would say (I posted pictures above), what I am surprissed is the pulse I see when powering off the load because I did not see on this post before... not sure if someone with the DL24MP ia having same behavior....

Thanks!!

[vk3em]

I have 2 x DL24MPs. One is in 300W configuration (master+daughter pcb) and the other is stand alone 150W configuration.

I will measure the gate FET control and see if the ON/OFF transition has the glitch you have observed.

I have been using both of them for a few months characterising some battery capacities. So far, very happy with them.

The software is crap. I am using PC SOFT V1.01 which is the only software I can get working with my DL24MP. You can configure and log OK but the software randomly crashes, mostly after the test is complete so its not so bad.