Looks like they're trying to calculate the temperature 🤣 instead of telling you the normal range in which the device must be operated.
The net dissipation in the IC is the sum of the switching and conduction losses. Since there's no information on the output impedance of the driver, we're, unfortunately, not in a position to calculate the latter. The switching losses maybe approximated as 2* Fs* Qg* Vcc.
Maybe you can tell them that you're driving IRFB4227 MOSFETs with 6.8ohm resistors at about 400kHz.
The net dissipation in the IC is the sum of the switching and conduction losses. Since there's no information on the output impedance of the driver, we're, unfortunately, not in a position to calculate the latter. The switching losses maybe approximated as 2* Fs* Qg* Vcc.
Maybe you can tell them that you're driving IRFB4227 MOSFETs with 6.8ohm resistors at about 400kHz.
Last edited:
And at a drive voltage of 15V.
Also, since the thermal resistance given in the datasheet is from junction-ambient (not junction-case), I don't see how one could've calculated the case temperature, which is what you've been measuring all along.
Maybe Infineon has more data on the IC than they've disclosed in its datasheet.
Last edited:
I have told them those facts already but will summarize again for them ... 🙂 It'll be interesting to see the final conclusion.
Here is reply:
-----------------------------------------------------------------------------------------
Thank you for the valuable information.
The power dissipation in the IRS2092 is the sum of the following terms:
PD = PMID + PLSM + PLOW + PLSH + PHIGH
Where,
PMID: Power dissipation of the input floating logic and protection circuitry
PLSM: Power dissipation of the Input Level Shifter
PLOW: Power dissipation on the low side
PLSH: Power dissipation of the High-side level Shifter
PHIGH: Power dissipation on the high side
Kindly find the link attached for the calculation of power dissipation. Refer junction temperature estimation part.
Click here
The formula shown below can be used to get the junction temperature Tj. For a given RthJA.
TJ = RthJA × Pd + TA < 150°C
RthJA: Thermal resistance from junction to ambient. for SOIC16N it is 115 degrees Celsius /W
TA: Ambient temperature
The junction temperature should be well below the maximum allowable limit for the safe operation of IRS2092(Considering a safety factor is always advisable).
-----------------------------------------------------------------------------------------
Hmmm, I wish they would just tell me if 60-65c is hot or not for the IC???
-----------------------------------------------------------------------------------------
Thank you for the valuable information.
The power dissipation in the IRS2092 is the sum of the following terms:
PD = PMID + PLSM + PLOW + PLSH + PHIGH
Where,
PMID: Power dissipation of the input floating logic and protection circuitry
PLSM: Power dissipation of the Input Level Shifter
PLOW: Power dissipation on the low side
PLSH: Power dissipation of the High-side level Shifter
PHIGH: Power dissipation on the high side
Kindly find the link attached for the calculation of power dissipation. Refer junction temperature estimation part.
Click here
The formula shown below can be used to get the junction temperature Tj. For a given RthJA.
TJ = RthJA × Pd + TA < 150°C
RthJA: Thermal resistance from junction to ambient. for SOIC16N it is 115 degrees Celsius /W
TA: Ambient temperature
The junction temperature should be well below the maximum allowable limit for the safe operation of IRS2092(Considering a safety factor is always advisable).
-----------------------------------------------------------------------------------------
Hmmm, I wish they would just tell me if 60-65c is hot or not for the IC???
Exactly as I had expected. And BTW, what was your exact question ? Did you clearly ask for the case temperature ?
Just ask for the parameter RthJC (thermal resistance from junction to case) and tell them that you'll calculate the case temperature yourself. Be blunt and brief, and maybe you'll get a straightforward answer.
Last edited:
Maybe just measure the voltage drop across 10r R237 and 2r21 R248 and fire up your calculator. That would cover the output drivers, which is probably most of the higher-current consumption.
Those Gate driver waveforms (on the MOSFET fitted channel) do not inspire confidence in the unit's reliability.
Also, when contacting component manufacturers, make sure to include that it's a commercial design, and it has been in service for ~14 years. That way they don't expect you to have done all the calculations required of the original designer.😉
Cheers
Those Gate driver waveforms (on the MOSFET fitted channel) do not inspire confidence in the unit's reliability.
Also, when contacting component manufacturers, make sure to include that it's a commercial design, and it has been in service for ~14 years. That way they don't expect you to have done all the calculations required of the original designer.😉
Cheers
Yes and sent thermal images. I have also repeated this to make sure.
Will do ... will ask Monday.
Hi Rick, thanks for the post, I am here to learn ... can you explain in more detail what the problem with the waves is in reliabity terms?
I have stated most of that ... 🙂
----------------------------------------------------------------------------------------
I asked for a simpler answer and there has been one more reply:
ME: ....Hi, this is getting very complicated ... is there not a simple best practice average operating temperature range for the IRS2092 IC? .... I mean temperature of the IC itself as measured in the original post (not ambient).
INFINEON:
I understand that the theoretical process is complicated.
The junction temperature of the IC cannot be predicted with great accuracy using thermal camera data. We can get more accurate data by using thermocouples.
According to the rule of thumb, a temperature increase of 60 to 65 degrees rise is acceptable as long as it has attained its steady-state value and there is no cumulative increase in temperature due to self-heating over time.
It is always a good practice to do theoretical analysis and compare it with practical values.
ME: ... The ambient temperature here is 23c
The IC casing measures 65c steady state, that is the best measurment I can make here because there is a powerful fan blows directly over the top of the chip and any contact thermocouples that are used get cooled by the airflow so do not give an accurate measurment.
I am not designing a product here I have repaired it and it works well (all PWM waves look great) I just find the fact that the IRS2092 (65c) is at a higher temperature than the mosfets (35c) to be unusual & was seeking your opinion on if this was ok or not.
The IC case temperature has increased from 23c to 65c = 42c rise from ambient.
So are you saying that up to a 65c increase from ambient is ok? (a 65c rise would mean the IC casing measuring 88c).
---------------------------------------------------------------------------------------------
I will get the 2x new 4227's fitted over the weekend and measure R237 & R248, then ask for RthJC on Monday ....
Sure: The thing that gives Class-D its high efficiency is NOT operating between all-the-way-on, and all-the-way-off. The gradual, arching turn-on causes a lot of extra dissipation.
Assuming these aren't 'scope artifacts, these oscillograms from post 84 show the reliability concerns.
Cheers
Assuming these aren't 'scope artifacts, these oscillograms from post 84 show the reliability concerns.
Cheers
I'd call that attention deficit disorder. Where did you ask for a prediction of the junction temperature ?
This looks like a "customer care" person rather than an application engineer. You're clearly talking about measured temperature and a technically qualified person should know that what's measured is the case temperature, and not that of the junction.
Yes I was getting the feeling that they had misunderstood what was asked for or were maybe reading the question too quickly and not picking up the correct context but you would think after multiple messages they would have got the idea ... I really thought it would have been a simple one sentence answer ...🤔
Thanks ... this is all of interest to me, I'm going to write up notes from the entire thread when I'm finished. Here's a better image of the waveform (interestingly the fets don't seem to get that hot, even after 40mins with 8 ohm load, channel volume controls at max and enough bass heavy signal to flash the limiters they were still around 35c). Whereas the IRS2092 case temp measures 60c at idle to almost 65c after 40mins.
This is LO ... I don't have the diff probes to be able to measure HO w.r.t. VS on the better scope.
It's a technical forum I have been asking on, you can follow the "story" here: https://community.infineon.com/t5/C...mperature-when-driving-IRFB4227/m-p/357405#M7
New IRFB4227 fets fitted so all working now.
R248 (2R21) only has 80mV across it .... so not a lot of watts ......
R237 (10R) only has 440mV across it
R248 (2R21) only has 80mV across it .... so not a lot of watts ......
Well, then the main source of heat is the switching loss without which the IC would reach down to a case temperature of 45C, as already seen from the measurements in post#83.
Now, if there's no reply from the manufacturer (which is what I expect), then you're left with two choices:
1) Leave things the way they are now (functional).
2) Add an emitter follower buffer (as shown in the reference design) and relieve the ICs of these losses to obtain higher overall reliability. Once you add the buffer, you may also reduce the gate resistor to get a more rectangular waveform, as already mentioned by Rick PA Stadel.
Now, if there's no reply from the manufacturer (which is what I expect), then you're left with two choices:
1) Leave things the way they are now (functional).
2) Add an emitter follower buffer (as shown in the reference design) and relieve the ICs of these losses to obtain higher overall reliability. Once you add the buffer, you may also reduce the gate resistor to get a more rectangular waveform, as already mentioned by Rick PA Stadel.
In their usual roundabout way of doing things I have had a private message from Infineon:
"Hi Bravaudio,
I understand that your query has not been resolved yet.
Our design team has been contacted for the junction to case thermal resistance value.
As soon as I have the information, I will let you know.
Can you share the whole schematic, input, and output requirements with us in the interim?
This would help us to answer you better.
Regards,
Nishanth"
Funnily enough I cannot attach the schematic to their private messages so have had to post it to the original forum .... 🤔
"Hi Bravaudio,
I understand that your query has not been resolved yet.
Our design team has been contacted for the junction to case thermal resistance value.
As soon as I have the information, I will let you know.
Can you share the whole schematic, input, and output requirements with us in the interim?
This would help us to answer you better.
Regards,
Nishanth"
Funnily enough I cannot attach the schematic to their private messages so have had to post it to the original forum .... 🤔
R237 (10R) only has 440mV across it
OK, good -- less than I was expecting.
I the meantime I noticed something else -- 6,2V Z101 and 102 on pins 1 and 6 of the '2092, are surely providing no more than partial regulation. And the current they conduct would vary widely from part to part, yet still remain within tolerance. The 620R's from the ±18V rails are too low, setting the current at around 21mA. That's ~125mW at each pin, and it has to go somewhere. If the zeners are 5% tolerance, they could be anywhere from 5,89V to 6,51V. That's gonna dramatically change how much of that current the '2092's internal shunt regulators are asked to dissipate. (They could be 5% tolerance, too!)
But wait -- there's more! It's not just the (possibly up to) 250mW for the already hard-working '2092 to burn off -- if I'm reading page 7 of the Oct2013 PDF correctly, it shows 11mA MAX for both IAAZ and ISSZ! That could mean, if the internal reg's are at the low end, and Z101 & Z102 are toward the high end of tolerance, the IC will see almost double the 'Recommended Operating Conditions' max spec.
Do I remember correctly that one of the puzzling bits about this piece was the markedly different temps of the 2 IRS2092's? Maybe somewhat mis-matched tolerances of these parts, since it could easily cause an extra 200mW or more dissipation.
Glad the repair was successful . . Congrats!
Cheers
(sorry, I couldn't get the editor to put the 'R237 . .' line back in the box, once it jumped out 🙁)
OK, good -- less than I was expecting.
I the meantime I noticed something else -- 6,2V Z101 and 102 on pins 1 and 6 of the '2092, are surely providing no more than partial regulation. And the current they conduct would vary widely from part to part, yet still remain within tolerance. The 620R's from the ±18V rails are too low, setting the current at around 21mA. That's ~125mW at each pin, and it has to go somewhere. If the zeners are 5% tolerance, they could be anywhere from 5,89V to 6,51V. That's gonna dramatically change how much of that current the '2092's internal shunt regulators are asked to dissipate. (They could be 5% tolerance, too!)
But wait -- there's more! It's not just the (possibly up to) 250mW for the already hard-working '2092 to burn off -- if I'm reading page 7 of the Oct2013 PDF correctly, it shows 11mA MAX for both IAAZ and ISSZ! That could mean, if the internal reg's are at the low end, and Z101 & Z102 are toward the high end of tolerance, the IC will see almost double the 'Recommended Operating Conditions' max spec.
Do I remember correctly that one of the puzzling bits about this piece was the markedly different temps of the 2 IRS2092's? Maybe somewhat mis-matched tolerances of these parts, since it could easily cause an extra 200mW or more dissipation.
Glad the repair was successful . . Congrats!
Cheers
Last edited:
If I were you I would stop writing to these people. They're trained to manage the situation by saying just about anything in return, like some kind of "online receptionist" who wouldn't answer your question but redirect things to Mr. X or Mr. Y.
I have written to manufacturers (by email, not some forum) and have had mixed responses, can't say they're all the same, but most are like this. They just seem to beat around the bush, when they could just say that they do not have an answer to your question. This is because they're trained never to say 'no' to a customer, even if that's the correct answer to his/her question !!!
So, now, it's all about reliability. In the worst-case scenario, the case is at junction temperature i.e. RthJC is same as RthJA. Assuming that you dissipate 1W, the junction would be at Tc + 1W (115*C/W) = 60 + 115 = 175*C. Though this is unlikely, it is quite possible that the steady-state ambient temperature around the IC remains elevated during operation, which in the worst-case reaches the case temperature (60*C) that you've measured.
It is generally accepted that the lifetime of a component is halved for every 10*C rise in junction temperature.
Thank you, newvirus2008, +1!
Oh, and that 250mW alone represents an almost 29°C temp rise for the SO16 package.
So if you really wanted to give the Peavey engineers a 'run for their money', consider replacing Z101 and Z102, and the corresponding parts in the other channel (sorry, I'm too lazy at this point to check the schematic for their component designations) with 5,6V, or even 5,1V parts. That would very likely greatly reduce or eliminate the unnecessary dissipation of the IRS2092's internal shunt regulators.
Best Regards
Oh, and that 250mW alone represents an almost 29°C temp rise for the SO16 package.
So if you really wanted to give the Peavey engineers a 'run for their money', consider replacing Z101 and Z102, and the corresponding parts in the other channel (sorry, I'm too lazy at this point to check the schematic for their component designations) with 5,6V, or even 5,1V parts. That would very likely greatly reduce or eliminate the unnecessary dissipation of the IRS2092's internal shunt regulators.
Best Regards
Last edited:
First here is the last installment from Infineon:
"The IRS2092 can operate safely at a temperature of 65C.
I would advise reducing VCC to 12V if you want to improve it even further. The benefit is that RDS(ON) won't significantly increase while Qg of the FETs is reduced. This will further reduce the temperature.
Kindly let me know if you have any further queries."
So you think that could significantly reduce the 2092's temperature without any other problems?
"The IRS2092 can operate safely at a temperature of 65C.
I would advise reducing VCC to 12V if you want to improve it even further. The benefit is that RDS(ON) won't significantly increase while Qg of the FETs is reduced. This will further reduce the temperature.
Kindly let me know if you have any further queries."
No they have both been very similar, the marked difference was between the 2092's and the 4227's (i.e the wrong ones were getting hot).
Yeah it's been pretty pointless, I also have the feeling that he may have noticed that the schematic is a Peavey design and just decided it must be fine to run @ 65c if that's what they were selling. The answer came very quickly after I uploaded the full schematic.
Worst case has been 65c ... ok thanks for that, that's all going in the notes.
That is an easier one to try (Z101, Z102, Z201, Z202) ... I have 1W through hole zeners in 5.1 & 5.6v here.
So you think that could significantly reduce the 2092's temperature without any other problems?