EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: BennoG on August 23, 2024, 08:42:45 am
-
I am working on a project that used SMD IR leds in short 500uS pulses about 1 pulse per second.
[attach=1]
The resistor is to counteract the influence of wire lengths and connector losses (so the biggest voltage drop is on the resistor).
There will be a multiple of these modules each having pulses (not at the same time) of about 10 - 15 Amp per pulse.
The total power of the resistor will be 15A * 1.2 \$\Omega\$ * 0.0005 Duty = 9mW. So in theory every SMD resistor of 1.2 \$\Omega\$ will be ok. The problem is these resistors will start to break at about 4A during the pulse.
Below normal pulse at 4A
[attach=2]
And the crumbling of the resistor some pulses later
[attach=3]
It looks like THT will be somewhat better breaking at 10A pulses.
Anybody a Idea of what resistors I can abuse for this (At the moment I have various current shunt resistors on order)
Added: resistor used for this.
https://www.digikey.nl/en/products/detail/rohm-semiconductor/ESR10EZPJ5R1/1762836 (https://www.digikey.nl/en/products/detail/rohm-semiconductor/ESR10EZPJ5R1/1762836)
Benno
-
What is the part number of the resistor you have now?
I assume you have tried the specialty high pulse power parts? (TE CRGP, Stackpole RPC, KOA SG73P, plenty of others) If not, grab those datasheets and look at the survival plots in there, and compare to your conditions.
-
Your pulses are 15*15*1.2= 270W. And the pulse duration is 0.0005S.
I opened the Vishay CRCW datasheet that defines pulse load.
https://www.vishay.com/docs/20035/dcrcwe3.pdf
(https://www.vishay.com/docs/20035/dcrcwe3.pdf)
These are standard SMD resistors. For continuous pulse load, none of the parts handle 270W for any duration. For single pulse, yes. You have probably a semi regular pulse load in normal operation, so you want to look at the continuous pulse load. This tells me that none of the standard SMD parts would handle this, including 2512, and you need to look specifically for pulse handling parts.
-
Anybody a Idea of what resistors I can abuse for this (At the moment I have various current shunt resistors on order)
Wirewound, but you already know that for sure.
The keyword here is "pulse withstanding". Have a look at various datasheets and literature from each manufacturers to figure out which can survive what. Most of them are qualified for voltage instead of current, but you may infer current from total energy or equivalent ratings.
example: https://www.vishay.com/docs/48516/_ms9702509-2401-vishaychecklistpulseload.pdf (https://www.vishay.com/docs/48516/_ms9702509-2401-vishaychecklistpulseload.pdf)
-
This is the resistor used, it says "Pulse Withstanding Thick Film"
https://www.digikey.nl/en/products/detail/rohm-semiconductor/ESR10EZPJ5R1/1762836 (https://www.digikey.nl/en/products/detail/rohm-semiconductor/ESR10EZPJ5R1/1762836)
-
I'm not sure if it's actually pulse withstanding. It only claims anti-surge which usually means surviving single IEC 60115-1 surge.
I opened the Vishay CRCW datasheet that defines pulse load.
For Vishay CRCW series it's either CRCW-IF or CRCW-HP.
Edit: CRCW2512-HP claims 150W@500us, CRCW2512-IF claims 500W@500us. It's up to you to qualify them.
FWIW CMB0207 mentioned in the vishay docs and Stackpole RPC2512 mentioned below claims 1000W@500us, huh.
-
This is the resistor used, it says "Pulse Withstanding Thick Film"
https://www.digikey.nl/en/products/detail/rohm-semiconductor/ESR10EZPJ5R1/1762836 (https://www.digikey.nl/en/products/detail/rohm-semiconductor/ESR10EZPJ5R1/1762836)
That is not a pulse-withstanding resistor. It is a surge-withstanding resistor and will not do what you need. (I am not going to attempt to justify the manufacturer nomenclature of these parts. It is insane, full stop.)
I will also note that last time I had a resistor problem like this, with a series pair of Stackpole 2512 RPC series parts occasionally, uh, exploding and spraying molten resistor guts everywhere, I was able to solve it with a single Vishay Z300-Cxx HV surge rated wirewound. Those didn't... leak... when provoked. (To be fair to the other guys, the provocation in that circuit was truly severe.)
-
You need numbers, not keywords. Keywords like "pulse withstanding" may help you with search, but obviously it only means better pulse handling capability compared to some ordinary reference part, not any arbitrary or infinite pulse handling.
You also don't need any rules of thumb, all the information you need is in the datasheets, there is this plot which shows maximum average power as a function of different pulse lengths and duty cycles. Some resistor series do not show this plot at all, they are most definitely unsuitable for the job.
-
I have these in order
https://www.digikey.nl/en/products/detail/vishay-beyschlag-draloric-bc-components/SMM02070C1008FBP00/5304417 (https://www.digikey.nl/en/products/detail/vishay-beyschlag-draloric-bc-components/SMM02070C1008FBP00/5304417)
Benno
-
Instead of using standard chip or melf you should look at dedicated high power parts. One example is Bourns PWR263, thick film high pulse capability.
-
Vishay CRCW2512-HP e3 looks like it's able to withstand 100Watt peak when subjected to pulses with a 500us repetition rate. So even then you will have to put a bunch of them in some series / parallel combination.
https://www.vishay.com/docs/20043/crcwhpe3.pdf (https://www.vishay.com/docs/20043/crcwhpe3.pdf)
On a side note, have you looked into the efficiency of those LED's at 15A? Light output usually tapers off at some point. You can keep on pushing more current though a LED, but it does not increase light output anymore beyond some point.
-
Your pulses are 15*15*1.2= 270W. And the pulse duration is 0.0005S.
I opened the Vishay CRCW datasheet that defines pulse load.
https://www.vishay.com/docs/20035/dcrcwe3.pdf
(https://www.vishay.com/docs/20035/dcrcwe3.pdf)
For what its worth - I looked into this recently and the Vishay datasheets had the best information on this.
Stackpole also had a little info, but no where near as comprehensive. Their datasheet gave indications for PWM percentages, but nothing about PWM periods, simply stating "Please assure sufficient margin for use period and conditions for “Pulse limiting voltage”".
https://www.seielect.com/catalog/sei-rmcf_rmcp.pdf (https://www.seielect.com/catalog/sei-rmcf_rmcp.pdf)
-
You need numbers, not keywords. Keywords like "pulse withstanding" may help you with search, but obviously it only means better pulse handling capability compared to some ordinary reference part, not any arbitrary or infinite pulse handling.
Yes, what I mean with "keyword" is what the industry call when advertising for such capability, not that the product advertised a such are automatically qualified.
It helps to know such terms when specific capability tend to be named with non-obvious nor straightforward phrases.
-
I have these in order
https://www.digikey.nl/en/products/detail/vishay-beyschlag-draloric-bc-components/SMM02070C1008FBP00/5304417 (https://www.digikey.nl/en/products/detail/vishay-beyschlag-draloric-bc-components/SMM02070C1008FBP00/5304417)
Benno
Just looked into these.
Check out page 7 of the datasheet.
They can't even withstand 270W for 1uS let alone 500uS.
-
I setteled for this one
https://www.digikey.nl/en/products/detail/susumu/RL3720WS-1R0-F/714297 (https://www.digikey.nl/en/products/detail/susumu/RL3720WS-1R0-F/714297)
Benno
-
On a side note, have you looked into the efficiency of those LED's at 15A? Light output usually tapers off at some point. You can keep on pushing more current though a LED, but it does not increase light output anymore beyond some point.
It are IR leds those are known you can overdrive them a lot, the first one failed at 14A. (they are rated for 1.2A)
I will check the efficiency next see where they stop increasing in brightness.
Benno
-
I don't understand why 1.2 R but not 0.1 R or 0.01 R for example?
-
Per the Vishay datasheet, the single pulse graph on page give shows (as an example) the blue line at approximately 80W.
If I understand the graph correctly, it shows the power dissipation over time and how it decreases.
What exactly does this mean for the resistor?
At an old job we did derating analysis and they used some Excel spreadsheet that someone created. I didn't have faith in it being accurate, but never got a clear understanding of how long the resistor can withstand various power.
To use a ridiculous example, what if it were 250W for 10ns, how do you know whether the resistor can tolerate this?
-
Attached an application note from Vishay.
Apparently Melf is the best for peak power, because they distribute the heat around the biggest part of the surface of the part. Sounds logical, for peak power there is no time for the whole body to warm up and distribute the heat.
Maybe they are a bit inductive too because they're spiral cut. That may also help a bit to reduce peak power in some situations.
-
I don't understand why 1.2 R but not 0.1 R or 0.01 R for example?
To eliminate the resistance of wires and connectors when you put a couple in parallel.
We opt for 5 LED's in series an then a resistor of 1 ohm
Then put 2 or more of these chains in parallel.
-
To use a ridiculous example, what if it were 250W for 10ns, how do you know whether the resistor can tolerate this?
A single pulse or continuous stream of pulses?
On page 5 the Vishay datasheet (https://www.vishay.com/docs/20035/dcrcwe3.pdf (https://www.vishay.com/docs/20035/dcrcwe3.pdf)) has two pertinent graphs:
The "Continuous Pulse" graph tells the resistors tolerance on an ongoing (i.e. continuous) basis on the assumption that average power is less than the resistor's rated power.
The "Single Pulse" tells the resistors tolerance to a single almighty pulse. It's a bit ambiguous about how often you can send it that pulse, simple stating that "it can only survive this single almight pulse if average power is close to 0" and that it can only withstand less than 1000 of those pulses.
How close to 0 does average power need to be for a single pulse like that? The datasheet doesn't say. I suspect its because that's a terribly difficult thing to characterise.
-
A single pulse or continuous stream of pulses?
It is a pulse about max 1 per second. In the attached graph it is the in Red marked line (RL3720, the RL7520 is mentioned 2 times they made a error in copy paste for this graph)
https://www.susumu.co.jp/common/pdf/n_catalog_partition07_en.pdf (https://www.susumu.co.jp/common/pdf/n_catalog_partition07_en.pdf)
It looks we will probably do 8-10A on a 1 \$\Omega\$ resistor. As mentioned above to equalize the currents between different modules at the same voltage pulse. So every unit has about the same current even if the cable / connector resistance is not the same between the modules. (if it is less than about 0.1 \$\Omega\$, it makes only up for about 10% of the voltage drop)
-
A single pulse or continuous stream of pulses?
Well.... I was speaking in general. A design at another company had a startup circuit (going by memory, so I may be off slightly) that had a power resistor with a MOSFET parallel to it. Immediately after turn-on the MOSFET would turn on and bypass the power resistor. This was done to reduce inrush spikes.
When it was analyzed, I think the calculations were nothing more than extrapolating and "believing" that the resistor could handle the X us pulse.
My question was based around this circuit design and wondering how to actually read the datasheet (or calculate) whether the resistor is safe to use for a particular time constant.
-
On a side note, have you looked into the efficiency of those LED's at 15A? Light output usually tapers off at some point. You can keep on pushing more current though a LED, but it does not increase light output anymore beyond some point.
It are IR leds those are known you can overdrive them a lot, the first one failed at 14A. (they are rated for 1.2A)
I will check the efficiency next see where they stop increasing in brightness.
Ok final post.
The leds produce at 10A about 8 times the light they produce at 1.2A (rated current) so the light output is almost linear to the current put into it.
Benno
-
The leds produce at 10A about 8 times the light they produce at 1.2A (rated current) so the light output is almost linear to the current put into it.
Remember that at the same time, Vf goes up so efficiency drops more than what you would think from current alone.
And if you look at the datasheet curve, it assumes the LED is somehow kept at rated temperature (say 25degC). Which is a fair assumption with very low average duty cycle and short pulses (short pulses with large intervals). But if the LED heats up, the light output drops further than expected from the curve. I suggest you measure the light output and find a sweet spot for maximized light output but not at any cost, i.e. stop where returns are diminishing.
-
And if you look at the datasheet curve, it assumes the LED is somehow kept at rated temperature (say 25degC). Which is a fair assumption with very low average duty cycle and short pulses (short pulses with large intervals). But if the LED heats up, the light output drops further than expected from the curve. I suggest you measure the light output and find a sweet spot for maximized light output but not at any cost, i.e. stop where returns are diminishing.
Yes monitored the temperature with a thermal camera after 1 hour it i about 5C above ambient temperature.