How critical is optocoupler's CTR parameter in SMPS feedback?

[chupocro]

Hi all,

I'll have to replace the a burnt optocoupler in computer monitor's SMPS but I can find only PC817 (without X) while the original was EL817X. According to the datasheet the difference is in CTR range:

Code: [Select]

817     CTR  50 - 600    <-- I have
817A    CTR  80 - 160
817B    CTR 130 - 260
817C    CTR 200 - 400
817D    CTR 300 - 600
817X    CTR 100 - 200    <-- original
817Y    CTR 150 - 300
I couldn't find the schematics of the monitor (AOC Q3279VWFD8) but I've found a schematics of another AOC monitor which is very similar. Here are the pictures of the parts that I removed because they were burnt and the schematics where I marked the path of 330V DC which went through the OZ531TG1N controller that was probably the first to be shorted.

I am waiting for NCP1271A controller to replace the original OZ531TG1N and SMK0965 transistors and I'd like to know if replacing the 817X optocoupler with 817 (without X) could cause some trouble.

[moffy]

If you measure the CTR of your non 'X' version and it falls within the 'X' version's range then your problem is solved. The CTR of the optocoupler affects the loop gain and therefore stability of the control loop, it sounds like good design practice to go for the limited CTR range of the 'X' version.

[wraper]

It should work fine. And you can measure actual CTR for a peace of mind if you want.

[chupocro]

I was thinking about measuring the CTR but wasn't sure if the operating point while measuring should be matched with the conditions of the circuit.

I saw this simple straightforward method of measuring the CTR:

https://en.everlight.com/wp-content/uploads/2021/07/General-Purpose-Photo-Coupler-CTR-Measurement-and-Calculation-Ver1.0EN.pdf

and a quite complicated one:

https://www.edn.com/measure-an-optocouplers-ctr/

but I am not sure what exactly would be the reason for using the second method over the first one.

I have many 817 optocouplers and I'll measure them all to select the one with the CTR which is the nearest to the center of the X range.

[PCB.Wiz]

I was thinking about measuring the CTR but wasn't sure if the operating point while measuring should be matched with the conditions of the circuit.

If you are comparing with the 817X, you need to match the 817X test conditions.

The opto is used in amplified mode, within a high gain difference loop, so the CTR has only a minor affect on the regulation.
If it's easy to select one, then doing so gives you more confidence to look elsewhere is there are still issues.

[moffy]

According to the PC817X datasheet: https://www.farnell.com/datasheets/73758.pdf
the CTR @ Vce=5V, If=5mA is 50% to 600%
According to the PC817 datasheet: https://www.digikey.com.au/en/htmldatasheets/production/34186/0/0/1/pc817-series
the CTR under the same conditions is: 50% to 600%
This is for the non graded versions, the PC817X has a Vceo=80V and the PC817 has a Vceo=35V, that is the difference and that is important to prevent failure due to excessive voltage.

[wraper]

According to the PC817X datasheet: https://www.farnell.com/datasheets/73758.pdf
the CTR @ Vce=5V, If=5mA is 50% to 600%
According to the PC817 datasheet: https://www.digikey.com.au/en/htmldatasheets/production/34186/0/0/1/pc817-series
the CTR under the same conditions is: 50% to 600%
This is for the non graded versions, the PC817X has a Vceo=80V and the PC817 has a Vceo=35V, that is the difference and that is important to prevent failure due to excessive voltage.

For EL817 "X" is a CTR grade  and it's a direct substitute of PC817 (that has no X CTR grade), PC817X is a different part that is rated for higher voltage and is an equivalent of EL817-G. Although it does not matter which one is used in this particular circuit.
EDIT: in old datasheets EL817 has the same CTR grading as PC817 and does not have X grade, in newer datasheets it's different.

[moffy]

According to the PC817X datasheet: https://www.farnell.com/datasheets/73758.pdf
the CTR @ Vce=5V, If=5mA is 50% to 600%
According to the PC817 datasheet: https://www.digikey.com.au/en/htmldatasheets/production/34186/0/0/1/pc817-series
the CTR under the same conditions is: 50% to 600%
This is for the non graded versions, the PC817X has a Vceo=80V and the PC817 has a Vceo=35V, that is the difference and that is important to prevent failure due to excessive voltage.

For EL817 "X" is a CTR grade and it's a direct substitute of PC817 (that has no X CTR grade), PC817X is a different part that is rated for higher voltage and is an equivalent of EL817-G. Although it does not matter which one is used in this particular circuit.

Thanks I missed the change from 'PC' to 'EL'.

[wraper]

BTW NCP1271A is not a substitute for OZ531, it operates at 65kHz instead of 100kHz. NCP1271B operates at 100kHz, however I don't know if there are other significant differences as no full datasheet seem to be publicly available for OZ531.

[wraper]

Actually NCP1271 has one pin less, although it's NC on OZ531. EDIT it's NC on OZ531TGN and is absent on OZ531TG1N. However what's important is that Pin 1 has entirely different function. What baffles me is you're concerned about CTR while simply deciding to put a very different IC instead.

[wraper]

Oh, an BTW the imaginary current path that "damaged the optocoupler" is absolutely wrong, and circuit is similar only on a first uneducated glance. LD7552 and OZ531 are not alike at all. On pin 1 there is a very high resistance voltage divider for voltage sense. And There still won't be enough current to destroy optocoupler through resistors on HV pin as well (EDIT: unless there is no resistor like on reduced application diagram). What most likely happened was that MOSFET got destroyed for some reason and PWM controller got destroyed by high voltage coming from MOSFET gate, and destroyed optocoupler through the IC (if it's actually bad, which I now doubt). Also check if MOSFET source resistor to GND did not fail open, it happens very often with such failures.

[chupocro]

Quote from: wraper on Today at 05:25:02 am

Oh, an BTW the imaginary current path that "damaged the optocoupler" is absolutely wrong, and circuit is similar only on a first uneducated glance. LD7552 and OZ531 are not alike at all. On pin 1 there is a very high resistance voltage divider for voltage sense. And There still won't be enough current to destroy optocoupler through resistors on HV pin as well (EDIT: unless there is no resistor like on reduced application diagram). What most likely happened was that MOSFET got destroyed for some reason and PWM controller got destroyed by high voltage coming from MOSFET gate, and destroyed optocoupler through the IC (if it's actually bad, which I now doubt). Also check if MOSFET source resistor to GND did not fail open, it happens very often with such failures.

I meant the schematics has a similar topology but the one that I found is using a different type of controller IC. OZ531TG1N's and NCP1271A's HV pin is connected directly to the 325V (220V * sqrt(2)) at the output of the bridge rectifier without any voltage dividers.

You are most likely right when you say the MOSFET was destroyed first because there is about 380V at its drain and the signal is not DC but you never know - the controller IC might have been destroyed the first too.

The destroyed components are: controller IC, MOSFET, 0.22 Ohm current sensing resistor and the optocoupler which has a shorted transistor.

I was having problems with finding the controller IC type because it is partially destroyed - here is the picture. I posted that picture to an electronics Discord channel where one of the members forwarded it to some Facebook group for searching the unknown electronic components and they replied the IC is OZ531TG1N. When I was looking for that IC I noticed every listing had both OZ531TGN and OZ531TG1N in its heading but I couldn't find a reliable source and there wasn't any OZ531TG1N only listings so I started searching for the equivalent of OZ531TGN and I found this page:

https://www.520101.com/html/circuitry/1228042826.html

The translation said: "OZ531TGN can be completely replaced with NCP1207A. In addition it can be directly replaced with NCP1203, NCP1271 (1271P65) and FA5571N.".

Then I checked the pinout of NCP1207A NCP1271A and compared the data with the PCB and everything matched. HV pin really goes straight to the output of the bridge rectifier, pin 1 (skip/latch) really goes through the resistor to GND, DRV goes to the MOSFET's gate, FB is the feedback from the optocoupler etc. and that's why I ordered NCP1207A NCP1271A ICs.

https://www.aliexpress.com/item/4000150942652.html

[wraper]

and that's why I ordered NCP1207A ICs.

https://www.aliexpress.com/item/4000150942652.html

There are loads of OZ531TG1N on aliexpress. Why in hell would you order some substitute? NCP1271A is simply not suitable due to different operation frequency.

In addition it can be directly replaced with NCP1203, NCP1271 (1271P65) and FA5571N.".

Neither of those are equivalents of each ater and cannot be used interchangeably without circuit modifications, and NCP1207 being the least similar to the rest.
It probably can be replaced with NCP1271B, however circuit on pin 1 will need to be modified. It probably has different Vcc (that you cannot adjust without rewinding transformer), and it probably has a different threshold voltage on current sense pin, that would need drain resistor value change. But it's not possible to know if modification is needed as there is no datasheet for OZ531 to know original specs. I never trust equivalents listed on internet, they are plain wrong more than 50% of the time.

[wraper]

I couldn't find a reliable source and there wasn't any OZ531TG1N only listings

You should chose "Color" to select correct item if there are multiple in the listing. Not to say there are plenty single item listings https://www.aliexpress.com/item/4001116474991.html

so I started searching for the equivalent of OZ531TGN and I found this page:

https://www.520101.com/html/circuitry/1228042826.html

Schematic looks like someone replaced different part number with OZ531TG1N. Pin 1 name does not match with other sources, pin function table above the schematic is for something very different, and does not match with schematic and pinout above it, looks similar to NCP1207.

[chupocro]

Quote from: wraper on Today at 10:03:53 pm

There are loads of OZ531TG1N on aliexpress. Why in hell would you order some substitute? NCP1271A is simply not suitable due to different operation frequency.

Every OZ531TG1N I saw either had zero reviews or the IC had 8 pins or it had only one row of text as it isn't original. I can't find which is operating frequency of OZ531TG1N, where have you found that data?


Quote from: wraper on Today at 10:03:53 pm

Neither of those are equivalents of each ater and cannot be used interchangeably without circuit modifications, and NCP1207 being the least similar to the rest.
It probably can be replaced with NCP1271B, however circuit on pin 1 will need to be modified. It probably has different Vcc (that you cannot adjust without rewinding transformer), and it probably has a different threshold voltage on current sense pin, that would need drain resistor value change. But it's not possible to know if modification is needed as there is no datasheet for OZ531 to know original specs. I never trust equivalents listed on internet, they are plain wrong more than 50% of the time.

Where can I see that pin 1 of OZ531TG1N has a different function compared to pin 1 of NCP1207A NCP1271A that I ordered?

BTW, this parser doesn't work well, I had to remove the "bold" tags from your "NCP1271B" text because it causes line break and breaks quote tag when it's under quote.

[chupocro]

Quote from: wraper on Today at 10:46:08 pm

Schematic looks like someone replaced different part number with OZ531TG1N. Pin 1 name does not match with other sources, pin function table above the schematic is for something very different, and does not match with schematic and pinout above it, looks similar to NCP1207.

The OZ531TGN pins in the picture are drawn in mixed order but the pin numbers are exactly the same as in the NCP1271A datasheet:

https://pdf1.alldatasheet.com/datasheet-pdf/download/166534/ONSEMI/NCP1271A.html

and they match layout of the PCB.

I ordered NCP1271A ICs, not NCP1207A as I wrote before.

[wraper]

Where can I see that pin 1 of OZ531TG1N has a different function compared to pin 1 of NCP1207A that I ordered?

I can't find which is operating frequency of OZ531TG1N

In my 3rd and 4th posts. Pin 1 is VINS.
https://www.eevblog.com/forum/projects/how-critical-is-optocouplers-ctr-parameter-in-smps-feedback/msg5365073/#msg5365073

Every OZ531TG1N I saw either had zero reviews or the IC had 8 pins or it had only one row of text as it isn't original.

Part of text/logo removed does not mean anything. It's a very common practice on aliexpress and is done with both original and fake parts. Not to say there is much higher chance receiving counterfeit ON semi part than O2Micro part. And frankly I don't even trust positive reviews. Most of them are left even before checking the part, and the rest say that obvious counterfeit (obvious for me) they bought is good. The only feedback I usually trust is negative.

[wraper]

I found another datasheet preview from O2 MICRO that says 67kHz, so it's not even clear what it actually is as manufacturer itself provided discrepant information. So I only suggest ordering original part and not trying your luck as there is not enough info to properly assess compatibility with something else. Even if pinout and frequency did match, there are other electrical parameters that often don't match between similar parts.

[chupocro]

Quote from: wraper on Today at 11:12:36 pm

In my 3rd and 4th posts. Pin 1 is VINS.
>https://www.eevblog.com/forum/projects/how-critical-is-optocouplers-ctr-parameter-in-smps-feedback/msg5365073/#msg5365073

Yes but that is OZ531, seems it has a different pinout compared to OZ531TG1N which has the same pinout as NCP1271A that I ordered.

Quote from: wraper on Today at 11:12:36 pm

Part of text/logo removed does not mean anything. It's a very common practice on aliexpress and is done with both original and fake parts. Not to say there is much higher chance receiving counterfeit ON semi part than O2Micro part. And frankly I don't even trust positive reviews. Most of them are left even before checking the part, and the rest say that obvious counterfeit (obvious for me) they bought is good. The only feedback I usually trust is negative.

I will order OZ531TG1N too but before that I'll probably these days try if it works with NCP1271A which have already arrived. I'll first power the PSU up for just a few seconds to quickly measure the output voltage - what is the worst that could happen :-)

[wraper]

Different suffixes most likely mean packaging and nothing else, OZ531 with no suffix is not a full part number. Very unlikely that suffix would change functionality to the amount of pin function change. Chinese web page that says they're equivalents cannot be trusted as it lists 4 verifiably incompatible ICs as equivalents and provides wrong pin function table. Not to say it shows OZ531TGN on schematic that has 8 pins, not OZ531TG1N with 7 pins if you want to nitpick about suffixes.

[chupocro]

Quote from: wraper on Today at 12:19:42 am

Suffix most likely means means packaging and nothing else. Very unlikely that suffix would change functionality to the amount of pin function change. Chinese web page that says they're equivalents cannot be trusted as it lists 4 verifiably incompatible ICs as equivalents and provides wrong pin function table.

I know there is a risk. However, the pinout of NCP1271A from this datasheet exactly matches the connections on the PCB. The note in the datasheet says pin 1 must be connected to GND through the resistor and that is exactly as on the PCB. And everything else matches that datasheet.

[wraper]

I know there is a risk. However, the pinout of NCP1271A from this datasheet exactly matches the connections on the PCB. The note in the datasheet says pin 1 must be connected to GND through the resistor and that is exactly as on the PCB. And everything else matches that datasheet.

Does it really have parallel RC circuit from pin 1 to GND and nothing else connected? And again, beside other things similar PWM controllers have different VCC and threshold voltages.

[chupocro]

Quote from: wraper on Today at 12:31:12 am

Does it really have parallel RC circuit from pin 1 to GND and nothing else connected? And again, beside other things similar PWM controllers have different VCC and threshold voltages.

Yes, there is a 43k resistor from pin 1 to GND and C907 capacitor in parallel to the resistor. Although the capacitor is unpopulated from the factory. You can see that missing C907 capacitor in the second picture from my very first post too - the one between the MOSFET that I removed and the 450V capacitor.