CY8C4245LQI temperature problem - bypass capacitor capacity?

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
GuNo_288966
Level 4
Level 4
25 replies posted 10 replies posted 10 questions asked

Hi, we have a custom board with CY8C4245LQI-483.

Several boards run perfectly in a temperature range from -20°C to 60°C.

Now, a new series of boards runs at room temperature but stalls at low and at high temperatures.

The only visible difference between the boards is that the bypass capacitors are mechanically bigger.

They are designed exactly as in figure 15 of the datasheet.

I can currently not measure the capacitance of the installed capacitors.

Could a too high capacitance of the bypass capacitors cause the controller to stall?

0 Likes
1 Solution
DennisS_46
Employee
Employee
100 sign-ins 50 likes received 50 solutions authored

Internal PSoC regulators (for Vccd) are absolutely stable for sufficiently large caps. I checked the samsung datasheet, available on Digikey. The X5R dielectric has a strong voltage coefficient; the graph below is for a Samsung 10uF 6.3 V cap example. Samsung does not provide a separate graph for each value and voltage rating.

DennisS_46_0-1645650259496.png

Scaling from this graph, at Vdd = 0.5* Vcap-rating (for 5V supply), the actual value of the capacitance is only 0.4 uF, instead of the specified 1.0 uF. If you are running at Vdd= 3.3V, then the cap value would be 0.6 uF.  I have found that X7R has a much lower voltage coefficient. To me, the performance difference is worth the money. I find that TDK and Kyocera capacitor datasheets are more complete for this kind of information.
My suggesting is to solder an additional 1.0 uF cap on the Vccd bypass  (just stack it on top) and see if that helps. Vccd is more sensitive to variation than Vddd or Vdda. It's clear from the PSoC datasheet that bypass of as much as 10 uF or more is safe. What is not clear is what happens when the bypass cap values are too small. The only exception to that lack of knowledge is the reference bypass cap for the ADC, where larger value caps reduce the noise in the conversion; this I have absolutely found to be true.
If a patched cap up to 2 uF works, then you should be able find a 2.2uF cap in the same package. If there is no change, check the other caps similarly. If no improvement, then it's time to return a failed chip for failure analysis.

---- Dennis Seguine
PSoC Apps Engineer 

 

 

View solution in original post

0 Likes
6 Replies
DennisS_46
Employee
Employee
100 sign-ins 50 likes received 50 solutions authored

It's unlikely that too high value would be a problem. Some capacitor dielectrics work fine at room temp, but work terribly (lower C) at high and low temperatures. Please provide manufacturer and part number for the bypass caps you are using and I'll check it out.

---- Dennis Seguine
PSoC Apps Engineer

0 Likes
GuNo_288966
Level 4
Level 4
25 replies posted 10 replies posted 10 questions asked

The capacitors that should be mounted are Samsung Electro Mechanics CL05A105MP5NNNC (Farnell part number 3013331). I can not find this part in the Samsung catalog.

I suspect that there is a different capacitor soldered to the board now. We got the reel from the same supplier as before with the same part number as before, but the capacitors are slightly larger than the ones from the previous reel. That´s why I suspect that they might have a higher capacitance.

I have already seen voltage regulators to fail when there is a too high capacitance at the output, that´s why I am concerned especially about the capacitor that is connected to VCCD, i.e. to the internal voltage regulator.

I am currently in home office and can not make any measurements of the parts.

 

0 Likes
DennisS_46
Employee
Employee
100 sign-ins 50 likes received 50 solutions authored

Internal PSoC regulators (for Vccd) are absolutely stable for sufficiently large caps. I checked the samsung datasheet, available on Digikey. The X5R dielectric has a strong voltage coefficient; the graph below is for a Samsung 10uF 6.3 V cap example. Samsung does not provide a separate graph for each value and voltage rating.

DennisS_46_0-1645650259496.png

Scaling from this graph, at Vdd = 0.5* Vcap-rating (for 5V supply), the actual value of the capacitance is only 0.4 uF, instead of the specified 1.0 uF. If you are running at Vdd= 3.3V, then the cap value would be 0.6 uF.  I have found that X7R has a much lower voltage coefficient. To me, the performance difference is worth the money. I find that TDK and Kyocera capacitor datasheets are more complete for this kind of information.
My suggesting is to solder an additional 1.0 uF cap on the Vccd bypass  (just stack it on top) and see if that helps. Vccd is more sensitive to variation than Vddd or Vdda. It's clear from the PSoC datasheet that bypass of as much as 10 uF or more is safe. What is not clear is what happens when the bypass cap values are too small. The only exception to that lack of knowledge is the reference bypass cap for the ADC, where larger value caps reduce the noise in the conversion; this I have absolutely found to be true.
If a patched cap up to 2 uF works, then you should be able find a 2.2uF cap in the same package. If there is no change, check the other caps similarly. If no improvement, then it's time to return a failed chip for failure analysis.

---- Dennis Seguine
PSoC Apps Engineer 

 

 

0 Likes
GuNo_288966
Level 4
Level 4
25 replies posted 10 replies posted 10 questions asked

Dennis, thanks for looking into this issue.

If the caps are too small, could we measure that with an oscilloscope, e.g. oscillations on VCCD?

Still in home office, no access to the affected boards.

0 Likes
DennisS_46
Employee
Employee
100 sign-ins 50 likes received 50 solutions authored

Vccd bypasses an internal regulator which supplies the bulk of the logic in the chip. The cap is required for
stability. If the cap is too small, there could be oscillation and noise on Vccd which could could upset logic
operation and also the logic part of the analog, such as in the ADC. I repeat my earlier suggestion to add a
a second 1 uF cap on Vccd to see if this helps.

---- Dennis Seguine
PSoC Apps Engineer

0 Likes
GuNo_288966
Level 4
Level 4
25 replies posted 10 replies posted 10 questions asked

I could now check the capacitors. They have 7µF.

Since the distributor obviously mixed something up, other details like material, voltage and tolerance have to be considered to be unknown.

We have replaced the questionable capacitors by an X5R 1µF 16V type, and the boards run fine now.

Dennis, thanks again for looking into this issue.

0 Likes