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

[Pukker]

Yes, exactly it is. IRFP260N seems to be fine. I use dl24p up to max 100W.

I also think 100W is an good limit to use for long life of the DL24.
Also not going above 5A..6A load, reverse protecting diode and PCB-traces will getting hot too.
In most cases hot electronics won't have an long life.

[orb]

I also think 100W is an good limit to use for long life of the DL24.
Also not going above 5A..6A load, reverse protecting diode and PCB-traces will getting hot too.
In most cases hot electronics won't have an long life.

Some people install a new heatsink with a vertical fan that does not cool the pcb. This is not good for pcb and the rest of the electronics. I prefer to have a "horizontal" fan that blows in all directions.

[Filippo52]

Orb, just your reflection. By placing vertical heat sinks, only the Mosfet is cooled, but the components of the card remain only with natural cooling.
As you have seen from the photos, I also added two small Heatsinks to the protection diode and to the shunts by placing the fins so that the fan's airflow flows well and this helps these two components which are subjected to strain when working with currents above 10A

[orb]

I found an interesting video about the weight and back of the chinese mosfets.

https://youtu.be/Jp9LGm6A3jk?t=327

[Filippo52]

Thanks for the link, dear Orb,  I'd seen it before but didn't understand much.
No serious electrical test is performed as the device used to make the measurement as well as being a cheap Chinese one does not even remotely have the power to verify the RDS-On of a power mosfet. That tool is only good to say it's a mosfet and it's not shorted or open - just nothing more.
Breaking the mosfet is of little significance to me because I don't have precise ideas on how an original mosfet should be made and what a fake is definitely made of instead.
Also the proof of the magnet. which I haven't mentioned, it's not always significant because many fakes have non-magnetic material and some originals use ferrous material; so it is not a conclusive proof.
As far as I know the weight, combined with the RDS-ON value measured with a voltage between drain and source of 10 v and a current of at least 1 A are the best combination (both must be used) to try to understand if a Mosfet it is original. Obviously even breaking them is fine if you can do it without destroying them and if you know how an original and a fake are made

[Filippo52]

this is test for RDS-on

https://tinyurl.com/hes6hdp9

for the weight it is necessary a weight scale with 0,1 gr resolution

[SpottedDick]

Is there full instructions on how to calibrate the DL24 anywhere?
I can do the voltage easily enough, but I haven't a clue on the 10A input.

[Filippo52]

There is some confusion about current calibration. If you go to the settings menu you will find on line 3 the voltage ref at 30v and immediately after on line 4 you will find Current ref and the calibration value is "3A" there is no trace of the value at 10A.
To carry out the calibration, the current must enter from the 4-wire INPUT (the two of the current will be sufficient) but you will have to let this current exit from the side socket to the right of the fan. You will have to put a load on this socket which allows you, with the input voltage on the input, to deliver a precise 3 A which you will have to measure with an ammeter of excellent precision; the setting procedure is then the same as you did for the voltage

[SpottedDick]

You absolute god, thank you so much!!!

[Filippo52]

These two new models have changed architecture and now use 4 parallel driven mosfets with separate op-amps.
The mosfet is the same as the previous models: IRFP264. It is not a mosfet for linear applications and for those who have had models with only one mosfets it has often burned out without being able to reach the declared powers of 150W and 180W.
With 4 mosfets there should be power reserve; 180w become 45W; even with some imbalance the mosfet should hold up just fine.
Too bad, however, that Athorh has mounted 4 tiny heatsinks so they become the bottleneck.

I therefore bothered to change the heatsinks.
First of all I discovered that in this period they cost a lot.
If one wanted to make a nice replacement with 4 dissipators like the original ones with a 4cmx4cm base but 4cm high instead of the 1cm of the tiny standard ones, one would spend 20€ on aluminum and almost double that on pure copper: far too much.
I looked for which one and I found an old-time heatsink 8cmx9,5cm 4cm high and very heavy ... 450gr for only 6€.
And I decided to test it with him.
I tell you right away that it is more difficult than it seems. To fix the dissipator you need precise holes which are then threaded. Furthermore, the base which ends up live (positive) because Atorch has decided to mount the mosfets without insulators for better thermal dissipation, goes over the board and it doesn't take much to damage some tiny tracks that pass underneath or even damage them with the screws you put from 'other side.
In short, attention!!
I tell you that I spent a day on it.

The results are encouraging

with 20 volts 4 A before the modification the temperature was just above 40°C, now we are just above 30°C and this sometimes makes the fan turn off, the temperature obviously rises and the fan start again.

I went to 16v 8 A and the temperature is still a reasonable 41.2 °C, let's say that as soon as I find a more powerful power supply I can definitely go to 150W and maybe even 180W.

With the most suitable heat sinks separated by 4cmx4cm 4cm high (perhaps in copper) you can go even better and the aesthetic result is certainly better.
For the rest, judge for yourself from the photos.
I'll leave the old coolers next to them to give a visual idea of the difference.
Before you ask me about the part of the heatsink that protrudes from the back: I didn't want to cut it; obviously I put some insulator underneath so as not to make short circuits with the tracks intended for expansion.

[Pukker]

The small heatsink on the picture and 45 Watts of power
The fan must be working hard.
The DL24 loads are/were an good concept for the price,
shame for Atorch they don't care about reliability.

[Filippo52]

Pukker I agree with you.
DL24 series are nice and contain good ideas, but there are many things to improve and Atorch thinks above all of making new models

[BarryMac]

Got a continuous and stable 415W on my modded DL24P




Main changes where heatsinked a fair few components on the board, the Zener diode gate mod, changed the MOSFET for a FDL100N50F and for cooling I'm using a 240mm AIO.

[vk3em]

Firstly, thank you to all the contributors to this topic since its creation. I recently purchased a DL24-MP and have quite a large number of small batteries to test. Attached are some photos and interesting observations.

Control/Display Board:
- uC is a HC32L176 in a LQFP-48 pin package
- Bluetooth (Matching) - BPOK228-28A2 in a SOIC-8 (made by https://www.zh-jieli.com/#)
    - No data could be found, PCB tracks indicate this is an RF device, Bluetooth PCB track antenna is directly connected to this, maybe amplifier or filter (BP = Band Pass?)

Main Board:
- FETs are still IRFP264
- High Power Diode is a V40100C
- Analog to Digital seems to be now be performed by 2 x HLW8110 High Precision Power Metering IC (the wiki below suggests one of them is just used as an ADC ) - Datasheet attached.
- FETs are still controlled via LM321 op-amps
   - Separate Op Amps for each FET
- Each FET has a separate high power current balance resistor
- ME6203 and ME6118A are used for linear voltage regulation
- Supplied 12v DC supply has a characteristic where output turns off when no load is applied. I didn't like it so its been replaced with a NetGear 12v 2A equivalent.

I also found that someone else has written some PC software : http://www.hardandsoftware.net/DL24Download.html

And there is a small Wiki on this version that someone is maintaining here: https://sigrok.org/wiki/ATORCH_DL24MP-150W_Purple

Will now need to do some testing.

[vk3em]

I was unable to get the 3rd party software linked above to work with my DL24MP, so I have been testing the new software from ATORCH that which can be found here:
https://www.mediafire.com/folder/1c04afq923397/A#yhumeijk86fj8

Note this is totally different from the original software for earlier DL24 models, and it has an earlier version number (v1.0) to add to confusion, which caused me to start my tests on the older E-Soft-PC-Soft which has a version of 2.01 (and which is useless).

The software is called "ATORCH Electronic Load Test Software(DL24_M_Purple) V1.0 (Build 2022-03)

Firstly, it works, and it seems to work quite well. It logs data to CSV, and I have attached an example of the output. You can configure the DL24 via the software, as opposed to just reading the current status parameters. It provides a graph over time, and detects when the test stops as per your configured stop criteria.

Interestingly, the software has a "Protocol" selection option - Protocol 1 (36 bytes) or Protocol 2 (7 bytes). My DL24MP is running on Protocol 2 via Bluetooth. I don't have an early DL24 to test to see if Protocol 1 supports those models.

My only criticism is that the screen layout does not seem to work properly on my laptop with its 1366x768 display, but after playing around with it, I have it displaying ok for now.

EDIT: I have since found a later version here: https://www.mediafire.com/file/jjy6gmms76t2jsw/ATORCH-DC_Load_PC_Software_V2.0.8%2528DL24M%2528Purple%2529_DL150_DL24EW%2529.zip/file

[killingtime]

Anyone have the 'add on' board with 4 mosfets? Looks like this:

https://i.ibb.co/YQ9RRyn/Untitled.jpg

Do you know if the board has 4x op-amps on it that regulate the current though each FET?

Yuhar kindly posted the schematic of the single FET add-on board in the link below. Just wondering if the 4x FET has a similar set up.

https://www.eevblog.com/forum/testgear/cheezeball-dc-load-dl24p-pump-or-dump/?action=dlattach;attach=1589569

The boards could be used for DIY load projects if you only have to feed them 0-0.1V to set the current. A simple 10 turn POT would do it. You'd only get CC mode, but that's OK for many applications. I think the current shunt resistors are a much higher value  for some reason on the 4x FET boards (0.25R), according to msg4186963. If that's true then these resistors will be smoking at 20A (board), 5A (each). That's 6.25W each. Wonder if they're rated for that power.

Thanks.

[vk3em]

>Anyone have the 'add on' board with 4 mosfets? Looks like this:

Yes I have it - its what I have documented in the previous two posts above. Its a DL-24MP, I have two modules, the control + 1 expansion for a total of 300W

I dont have a schematic, but yes each of the 4 FETs has its own op-amp.

The control board is has 6 x LM321 and the expansion board has 4 x LM321. I can take more detailed photos if you want.

Luke

[killingtime]

Hi Luke, thanks for the reply.
I'll order a board and see if I can reverse engineer it. Should be interesting.

[thm_w]

The boards could be used for DIY load projects if you only have to feed them 0-0.1V to set the current. A simple 10 turn POT would do it. You'd only get CC mode, but that's OK for many applications. I think the current shunt resistors are a much higher value  for some reason on the 4x FET boards (0.25R), according to msg4186963. If that's true then these resistors will be smoking at 20A (board), 5A (each). That's 6.25W each. Wonder if they're rated for that power.

They are 6W rated resistors.
Two are 0.25R two are 0.22R.
You could couple them to the main heatsink with some thermal glue. Or try to get more airflow over them. If you plan to run at 20A all the time.

[Anomalia]

The device shows 0.7v instead of 2.25v. I can't fix the lower register. It reacts to the voltage change, but the lower the voltage goes, the bigger the difference becomes.
Similar issue: https://www.eevblog.com/forum/testgear/dl24p-calibration/

Any tips on what broke? I did experiments with the reverse charging of a lead-acid battery and for that I discharged the battery quite empty with moderate amps.

Edit:
Never mind. Load wires were apparently not connected during calibration. When I did the 30v calibration again with load + v-sense connected, the readings are correct again.

Edit 2:
It works but doesn't work as it should. When the battery voltage is, for example, 2.2v and I start discharging, the voltage rises by 0.05v. Of course, the multimeter says the opposite.

[deuteron]

Hello alltogether,
finally (after reading the whole thread, as well as many others about the DL24P) I'll show you my solution of the 'problem'. In my application, which is mainly testing of rather small magnet power supplies for physics purposes, it is not mandatory to push this thing to or even over its limits. Nevertheless, after using the current load for several days with very moderate power (< 100 W), I also smelled the famous magic smoke while testing a six cell Li battery at 4A yesterday. The days before I was already preparing some modification, but at that time only for prophylactic reasons. Btw I was very lucky that the failure of the Fet occured while testing a battery pack with internal protection against overcurrent. Otherwise the smoke would have been 'very' magic.

What did I do ? I made three main modifications:

1) I changed the Mosfet for the IXFK360N10T model, which SOA curve has a DC area with a usable part (>160W) for voltages <= 80V. The Fet has a larger package (TO-264) compared to the original one (TO-247AC), which btw was an IRF264 and not an IRF260. The new one is still small enough to fit at exactly the same place (including screw hole) on the board. I placed the Fet such that it directly touches the thermistor mounted on the board together with some heat compound. I'll come back to this point later.
2) I exchanged the original heat sink (the disco-led-model) for a Noctua NH-9Li, which I got used for half of the usual price. It directly fits to the (75mm x 75mm) mounting holes and touches nicely the Fet package without bending the electronic board. I used the very good but somewhat expensive Arctic MX-6 thermal compound.
3) I installed the Zener mod but with two pieces in back to back configuration. I'm not shure, but I think that during the observed Vgs ringing signal also negative voltages may occure.

Although I'm not after maximum power dissipation values, I did some testing up to 120W with currents up to 5A. The nice thing is that even at 120W the board temperature was well below 60°C (57°C to be exact). Furthermore I used the additional thermal sensor to cross check the fixed sensor on the board being in direct contact to the Fet package. I installed the sensor together with heat compound into the Fet mounting hole. This should pretty well represent the actual temperature of the Mosfet package. Another nice observation was that both temperatures were the same within +/-1°C max. That means I can rely on the board temperature sensor (comfigured as described in (1)) and have the second sensor free for other measurements e.g. the temperature of the diode. This time I measured the diode temperature with an additional sensor of a Fluke DMM.
The diode temperature (without any further cooling measure) was 62°C at 5A. From that, if one wants to keep it below 70°C, one is limited to perhaps 6A. Btw, the Noctua is a horizontally oriented cooler either, so that the diode is cooled somewhat by the vertical air stream.

As a summary, now with these three mods, the device seems to work reliably with an usable dissipation power of at least 150W (at save temperatures below 70°C). The SOA plot recommends to limit the voltage to about 80V for this power level (2A of current).
Below that voltage the limitation is given purely by the cooling power of the heat sink. For my applications this is more than enough, although other Mosfet types may be more suitable to work at higher voltages. Thus in my configuration the specs are: Pmax = 150W, Vmax = 100 V, Imax = 6A. For a save operation each limit has to be met separatly of coarse. I personally am pretty satisfied with the result. As a further modification I'll perhaps try to cool the diode more effectively by thermally contacting it to the heat sink, which (with the NH-9Li cooler) just needs a small copper or aluminium body with a 3.5mm height (see picture).
Greetings
Stefan

[deuteron]

Here comes part 2 of my posting. Today I made a more serious analysis of the temperature data I took. But to begin with, the first picture shows the SOA plot of the IXFK360N10T, in which the colored diagonal lines indicate the working points within the DC area at different power dissipations. My first intention was to use the load up to 100W max only. In this case, the working line doesn't come into conflict at any point of the SOA diagramm. The green line (130W) is the maximum power, for which the above statement is still true: It's only the power dissipated limiting the working point. The second picture is a diagram of the measured case temperatures vs. power. The curve is pretty straight linear and can thus nicely be extrapolated up to the original design power of 180W. In my configuration (without having tested it) the temperature here is determined to be just a bit above the 70°C 'reasonable temperature limit', 71°C to be exact. I think, one can widely trust the linear extrapolation. What's additionally interesting is to look for the 180W limiting line in the SOA diagram (red line). The line intersects the DC limit line at about 78°C (+/-1°C, I was to lazy to calculate the position of the line more exactly, because of the logarithmic scale.). Up to this voltage the chosen Mosfet is well suited for a 180W electronic load.
Today I also started the experiment to order a bare DL24P board from Aliexpress either. On arrivel (middle of July) I'll do some experiments with another (even higher power) Mosfet, the IXFN100N50P. The main difference is that it allows much higher currents between 100 and 120V (End of SOA DC). From the SOA diagram 10A should be possible at 100W. To use even only a part of that 1KW of power all the other improvements discussed here and elsewhere will become important: Cooling of the diode, strengthening the traces on the electronic board as well as a CPU cooler with mus higher cooling power. I will also report about that here.
Nice weekend !
Stefan

[deuteron]

Just a small correction. I recalculated the lines in the SOA diagram, see below.

[deuteron]

Yesterday I killed the IXFK360N10T ! I ran the DL24P with the above metioned modifications at 50V/2A and then increased the current by 1A in one step. And that was it, complete short as always. Before that the unit stably operated at 40V/3A (120W) for at least half an hour. I went to 50V, first 2A, running fine for a couple of minutes and than I went 1A up. Of coarse I intensively studied the SOA diagram before I did this. This setting is a good deal away from the DC limit, but nevertheless ...  It's absolutely unclear for me what went wrong. Vgs should be fine with the Zener mod. The published SOA diagram is valid only at 25°C. Perhaps there is a strong dependence of the DC limit on the case tempetaure and thus the SOA lines are degraded by some amount. But can it be that much ?
The Fet was somewhere between 50 and 60°C before I went up. So I don't think it was a thermal problem. Another possibilty could be some voltage spike form my power supply. With its ouptut power increasing, the power supply is doing some internal switching.
The problem is I only have one additional piece of this Mosfet and I don't want to sacrifice it for reseaching the reason of the failure.
Ok, the IXFK360N10T is not exactly a 'linear' mosfet, but it has a DC allowed region at least. Btw does anybody know, how could a fet be specified for DC operation without being capable of linear operation ? Has somebody any idea of the reason for the failure ? Would be very much appriciated !
I will now buy 2 pieces of the IXTK60N50L2, which is explicitly a linear mosfet with an additional SOA diagram for 75°C case temperature. Finally I want this thing to be limited only by the cooling power and not by an electrical problem. Let's see ...
To be continued,
Stefan

[Pukker]

I will now buy 2 pieces of the IXTK60N50L2, which is explicitly a linear mosfet with an additional SOA diagram for 75°C case temperature. Finally I want this thing to be limited only by the cooling power and not by an electrical problem. Let's see ...
To be continued,
Stefan

What about seperating the Mosfet from the board and screw it on the cooling system? A fuse in the Drain line, does not effect the accuracy I think, not damaging your board
when the Mosfet get shorted.
BTW I fused my load and connected the sense line for the fuse, so accuracy stays the same.