New DL24EW and DL150W load testers from Atorch.

[SpottedDick]

I'm aware we have the DL24(p) cheeseball thread, but this new unit from Atorch seems like such a radical redesign, I figured it deserved its own thread, especially since I can't find any proper reviews online of these modules yet.

Ignoring things like the WiFi option, there's basically a colour screen version and monochrome screen version.

Colour screen (DL24EW):
https://www.aliexpress.com/item/1005004773128494.html

Monochrome (DL150W):
https://www.aliexpress.com/item/1005004682736656.html

It's obvious from the listing photos that they've completely redesigned the circuitry. However what's more interesting and not shown in the listing (I got this from a review) is that they've now moved to 4 independent MOSFETs instead of one! This is in a very interesting arrangement that places the MOSFETs through the PCB.

See below:




Considering they've finally added a fuse, I wonder have they gone and fixed other issues like the ringing at the gate?

Yes no more CPU coolers unfortunately folks, they've gone for a custom design this time. Not a bad thing if they've actually got it right!

I can't see what the MOSFETs actually are, but considering they've stuck to 150W across 4, maybe they're actually being honest this time?

Anyone thinking of purchasing one?

[elecdonia]

I can't see what the MOSFETs actually are, but considering they've stuck to 150W across 4, maybe they're actually being honest this time?

I recently learned that some power MOSFETs are rated for “linear operation.” These have a well-defined “safe operating area” for voltage/current. In contrast most MOSFETs are intended for on/off switching only. Digikey’s search feature offers a selection parameter of “linear” which pulls up the most desirable MOSFET part numbers for applications like these linear electronic loads.

Now I’m thinking of an alternate architecture for an electronic load: Wide range DC-DC buck/boost converter feeding a very large wattage load resistor. The load current gets adjusted by varying the output voltage of the DC-DC converter. This will dissipate most of the power in the resistor rather than directly into the MOSFETs.   

[Filippo52]

Yes I plan to buy one in the near future, I am currently reviewing the DL24MP, single element.
I see both the color and monochrome models claim the same power but use different heatsinks.
Those in the color model are double the height; how is it possible that this has no effect on power?
in any case we are business as usual, there is always something badly sized and some bottle neck.
What did it take to put bigger heatsinks?
Once cultured, the DL24 burned out because a single mosfet not suitable for working in a linear way went up in smoke. Now they have put 4 mosfet that could also do something, but they have reduced the heat dissipation capacity of the heatsinks. Really strange.

I wanted to try to put better heat sinks, perhaps in copper, but I saw that they are now prohibitively expensive.

[thm_w]

Those in the color model are double the height; how is it possible that this has no effect on power?

Because the spec they give is meaningless, it will never reach that level safely. I have one of the short heatsink style, older model, I believe it was capable of 60-70W at reasonable temperatures.
The taller heatsinks might do a few watts more.

What did it take to put bigger heatsinks?
Once cultured, the DL24 burned out because a single mosfet not suitable for working in a linear way went up in smoke. Now they have put 4 mosfet that could also do something, but they have reduced the heat dissipation capacity of the heatsinks. Really strange.

Yes 4 FETs, if they are genuine ones not fakes, will handle more power than single reliably.
But that added more to the cost, so they probably cut back on the heatsink size.

I wanted to try to put better heat sinks, perhaps in copper, but I saw that they are now prohibitively expensive.

Look around for old graphics cards or computers for sale on your local marketplace. You can pull good heatsinks out of them. Should cost less than $20.
Its just that the form factor this product comes in is not very friendly for modding or installing in an enclosure.

[srsrus]

I bought this load. The thing is interesting. Unfortunately, I didn't have time to fully test it. It broke due to incorrect connection. At the moment I am corresponding with the manufacturer to restore the device.

[Filippo52]

It is possible to put a single heatsink with a 8x8cm base (even 2mm more to recover the lost space between the current heatsinks which are 4x4cm but are not placed side by side; contact is not a problem because they are all together electrically even now) as there are no components between the 4 small current heatsinks. I found an old heatsink in an electronics shop 8,4X9,5cm 4 cm high and weighing 450 gr.
I can mount it in my DL24MP (which is similar in terms of mosfet arrangement) by making 1.5cm come out from the opposite side to the connectors, obviously in this case I will have to isolate a part of the base from the expansion contacts.

then maybe I'll let you know for the same absorption how many degrees of temperature you save.

I with a DL24pcb dissipated with an old intel core I7 bloomfield CPU cooler at 80 W I have 44 °C and I pushed it up to 120 W with the original mosfet and a temperature of 64 °C

I am convinced that with this 4 mosfet version, if cooled well, 150W is an achievable value

[Filippo52]

Spotted Dick
I wanted to ask you something because I'm having a dispute with the seller for the following reason.

In my DL24MP there are 4 mosfets and to balance the load, as well as having independent drives on the various mosfet gates, balancing resistors have been inserted.

up to here all OK

looking closely I saw that the resistances of one side were 0.22 ohms while those of the opposite side were 0.25 ohms.
I found this rather strange and went to measure the currents. As was logical, the mosfets with 0.25 ohm resistors passed a current of about 10-15% less than those with 0.22 ohm resistors.
It's not a serious thing but not any justification and with high loads it can decrease the maximum achievable performance.
The Atorch vendor was confused and contradictory. So I opened a dispute and first he offered me, only in words, a new DL24MP at half price then he said he had spoken to the manufacturer and produced a diagram where using different resistors is a "precise choice" design. And here comes the beauty according to them it serves to balance the load for the alleged longer routes ... and he also accused me of having tried to confuse and deceive him.
Something is denied by the measurements but in pure Aliexpress style when they are in trouble they ask you for proof of a video that is practically impossible to do.
I'm afraid I won't buy anything from this seller again, having already had to say for another product where the APP is not on the Google Playstore and the PC one is positive for the antivirus.
I wanted to ask you if you can tell me in your apparatus what value the four resistors have and if they are all the same, let me take a picture of them. You can do it?
if instead they are different, have you checked if the currents are balanced between the 4 mosfets?
It is not difficult being able to put for example a global current of "n" A and then measure the voltages on the 4 resistors.
I thank you in advance

[thm_w]

Why would you open a dispute over resistors being slightly different resistance? Who cares.
Mine are 0.25R and 0.22R as well.

Unless you thought it could actually do 150W stock, which from reading this thread and your posts, you should already know it cant.

[Filippo52]

I don't think it can do 150W of dissipation especially with those tiny heatsinks.
But I don't understand why losing further capacity by unbalancing the load between the various mosfets. It seems silly to me, they claim it's a design choice to balance the load on the mosfets, but the measurements I've made say the opposite.
Keep in mind that it is a dispute on the principle, having learned that it is a choice of the manufacturer I have put a symbolic refund of 1 €.
Se nessuno glki contesta i difetti non faranno nulla per migliorare.
E' un oggetto carino ma hanno molte cose su cui lavorare.
Un altro esempio? la modalità  CV praticamente non funziona.
Il software da usare con il bluetooh è quasi inutile...

[Filippo52]

Ah just out of curiosity, by chance have you tried to see since you too have different resistances if the currents are balanced in the 4 mosfets?

[thm_w]

Its a good observation though.

With a 5A load I measure: 0.299V, 0.299V across 0.25R and 0.288V, 0.286V across 0.22R
So 1.2A and 1.3A, about 8% difference as you had predicted.

[Filippo52]

OK
Thanks for your attention.
I would like to point out that the official Atorch seller claims that the choice of putting different resistors on the two sides is a precise choice precisely to optimize the balance and that it costs more to put different resistors but this way they get the balance and they sent me a photo thermal that I am attaching to you.



I find this thing very strange and I told him that I have measured different currents and now you confirm this thing,
If the currents are different and are greater on one side, that side will have more heat dissipation and so what are they sustaining and why?
I also looked at the photo they sent. It can also be seen by eye but with Photoshop I verified that the left side (seen from below, in my case it is the side with the lowest resistances and the highest currents) has a lighter shade of yellow, a sign of a slightly higher temperature.

Who knows if they will be convinced that by putting different resistors they make a mistake?

or is there some mystery?

[thm_w]

Maybe on an earlier design, there was a bigger difference between the FETs with actual justification for it. And they just left it in, even though its no longer needed?
Or whoever tested it had screwed up the thermal paste.

I wouldn't overthink it.

[SpottedDick]

If it's like the original DL24, you can calibrate it out anyway?
Of course, first put a meter on it and see if it's already accurate

Can you get the model number of the MOSFETs?

[Filippo52]

Thanks SpottedDick,
but the problem is not the calibration but rather getting the maximum power dissipation. You can get this maximum only if the 4 mosfets work at the same temperatures otherwise the hottest ones will be the first to die followed by the others.
By changing the heatsink with something much more robust, even keeping its fan at 150 W, you can get there.
Currently I have arrived at 80 W with 40 °C with those ridiculous heatsinks that they put in and therefore even if it is not the temperature of the mosfet I think it is still a safety temperature. I have had it for a long time without any problems. By placing a 400 gr heat sink (8cmx8cmx 5cm heigth), a 50-55°C could be obtained with 130-150 W dissipated.
I test these days and then I come here to report

[SpottedDick]

Thanks SpottedDick,
but the problem is not the calibration but rather getting the maximum power dissipation. You can get this maximum only if the 4 mosfets work at the same temperatures otherwise the hottest ones will be the first to die followed by the others.
By changing the heatsink with something much more robust, even keeping its fan at 150 W, you can get there.
Currently I have arrived at 80 W with 40 °C with those ridiculous heatsinks that they put in and therefore even if it is not the temperature of the mosfet I think it is still a safety temperature. I have had it for a long time without any problems. By placing a 400 gr heat sink (8cmx8cmx 5cm heigth), a 50-55°C could be obtained with 130-150 W dissipated.
I test these days and then I come here to report

I'm not sure that a 10% difference is going to let a MOSFET run away in thermal runaway. I'm also not convinced this wasn't intentional due to you and another user here getting the same value...

Are you sure that isn't to cope with track resistance or something?

10% is that small in the grand scheme of things, that it does sound like track resistance. Remember you're talking to a seller here, and not the engineer. The engineer probably has 0 English and the seller hasn't a clue what you're talking about.

The only way you'd know for sure would be to take off a MOSFET leg and do two opposite ones and see if it makes sense.

Also, I have my original DL24 up to 350W with a water cooler and better MOSFET. Can you get me the model number they've used on the 150W version so I can check the COA curve and see where it should land, even with a 10% difference?

[SpottedDick]

Just a note on the new unit, I can't really speculate on this until you send the model number, but the delta between temperatures matters a lot, something like this could make a massive difference, but obviously mounting is a problem.

https://www.aliexpress.com/item/1005002971155868.html

You would just need a €3 temperature sensor to control the Peltier device and turn them on/off at a certain temperature to get the delta up when you need it.

[SpottedDick]

#
The only way you'd know for sure would be to take off a MOSFET leg and do two opposite ones and see if it makes sense.

Also, I have my original DL24 up to 350W with a water cooler and better MOSFET. Can you get me the model number they've used on the 150W version so I can check the COA curve and see where it should land, even with a 10% difference?

I had a few beers last night before posting this!

I meant test the actual current through each mosfet with a meter.
I also meant SOA curve, not COA

[Filippo52]

Spotted Dick,
Measuring current doesn't seem difficult to me; the 0.22 and 0.25 ohm resistors are in series with the Mosfet so just measure the voltage across them to know the currents. It is exactly in this way that I have seen that they are unbalanced and that by reporting the equal value on all 4 mosfets the currents tend to be the same.

For the mosfet model, both in the DL24 pcb and in the 4-mosfet DL24MP there are IRFP264

a greeting

[SpottedDick]

Sorry, I'm bad at explaining things.

I don't think you said which side are which, but I'll guess the 0.25 Ohms are on the left and 0.22 Ohms are on the right?

If so it's probably to compensate for the additional track resistance to better balance the circuit. 0.03 ohms would make sense in that case.

Measure the voltage from the green connector at the bottom left to the resistor on the MOSFET input side and see if that gives a better reading.

On the IRFP264, that's great news. Once the heatsinks can keep up, that is certainly capable of 150W.

Having a quick look at the data sheet and assuming my math is correct, running at 100°C, each of them can still do 50W, so 200W total. The junction temperature would be 137°C, so still a bit of headroom.

Just a quick note before someone comes in, I know technically those MOSFETs are technically not rated for DC.

[Filippo52]

Spotted Dick
The resistors are in series with the mosfet, that's all you need. With this the voltage across is "exactly" the current flowing through the mosfet", you don't need to know anything else about the PCB rails.
The current is that anyway.
Yes, the 0.25 resistors are those on the left side, where there is the connector from which current and voltage enter and on the side where there is also the double schottky diode against polarity reversal.
  I agree with you that with good dissipation this object can do 150W and also little more.
But be careful when you change heatsinks. There are no insulators so the bolts and screws holding the heatsinks are live, there is positive voltage above them. The bad news is that these bolts force the PCB and the washers overlap or pass less than a millimeter from tiny tracks that carry the negative from the mosfet's Source towards the operational that drives the mosfet's gate. This is particularly serious for the mosfet which is close to the double schottky diode; if I tightened the screw too much or badly, you blow up the track and the mosfet obviously crashes, leaving you without driving the gate.

You were right to point out that those mosfets are not for DC and if you look carefully at the SOA graph, in addition to not having the curve for DC, which you obtained in the drawing like me imagining ... it is measured with the case at 25° C and this is the most serious thing, much more serious...when will the case ever stay only at 25°C? If you look at linear mosfet the SOA graph is with case at 75°C that is the real situation in a linear application like DL24) So the amps and the power they can dissipate is fortunate, but from practical use I have seen that a single IRPF264 mosfet if original holds, well dissipated, even 80 w for long times (I don't know if it only lasts a few months... we'll see) ... so in four if well dissipated the bottleneck of everything becomes the double schottky diode especially if you use high currents. It should be replaced or eliminated.
The manufacturer in the brand new DL24E-W model has replaced it with an automobile fuse and a very large diode. This way you don't heat anything and you also don't lose the 0.8V of the diode, which also allows you to discharge single-element Ni-Mh batteries. If you get the polarity wrong, hoping the diode is fast, it blows the fuse and saves everything, but you're not sure.

[SpottedDick]

Hi Filippo52,

The figures I gave where for the case temp at 100°C, I de-rated and calculated what the max they could do was

That's the reason I think the smaller heatsinks may be fine. By using more MOSFETs at a higher temperature, they can safely stay within SOA and reduce the heatsink size due to the higher delta. You never mentioned what kind of temperatures you where seeing?

For the resistor issue, is there any chance you can take a picture of the board from above, so maybe I can draw on it to explain better what I am trying to say?

Ciao!

[Filippo52]

Spotted Dick
while waiting for the photos, I ask you if you have made any calculations on what temperature those small heatsinks must reach to dissipate 150W. Perhaps the 100 °C that you hypothesized is not enough and in any case in a completely open apparatus do you know the danger represented by 4 small dissipators at 100 °C?
and do you still know how much the average life of a mosfet is reduced at those temperatures?
If you want a reliable device and not a dangerous object, you have to limit yourself to case temperatures of 70 °C, look at the temperatures used by linear mosfets in the SOA.
Then you and I invented the SOA with the DC work curve; we can't build anything else on it: it's like putting a weight on a sand castle.

A greeting

[thm_w]

https://www.eevblog.com/forum/testgear/atorch-dl24p-(or-pcb)-vs-dl24mp-function-costant-voltage/
https://www.eevblog.com/forum/testgear/atorch-electronic-load-dl24mp-e-dl24e-w/

You can post this stuff here or in the other Atorch thread, I don't see the need for new threads.

https://www.eevblog.com/forum/testgear/cheezeball-dc-load-dl24p-pump-or-dump/

[Filippo52]

ok
sorry, my mistake

I did tried to move the message but receive this information

"You cannot delete your own topics in this board. Check to make sure this topic wasn't just moved to another board."