@jwells02
Are you testing new boards using the same test setup and equipment as you used with the old boards? (That is, has there been any change in the test equipment and procedure?)

 

(That is, has there been any change in the test equipment and procedure?)

Or where the boards are being made or a different batch of boards.

 

Haven't changed the testing. We run them under actual conditions for at least 24 hours. We've tried different batches of boards there doesn't seem to be a correlation. Different power supplies. Just keep getting about a 50% failure rate. I've attached the latest schematic where I've added flyback diodes and made sure there where no floating pins on the chip even though they have internal pull down resistors. Will be testing this morning.

 

Haven't changed the testing. We run them under actual conditions for at least 24 hours. We've tried different batches of boards there doesn't seem to be a correlation. Different power supplies. Just keep getting about a 50% failure rate. I've attached the latest schematic where I've added flyback diodes and made sure there where no floating pins on the chip even though they have internal pull down resistors. Will be testing this morning.

Trying the flyback diodes is a reasonable approach to the problem and is consistent with failures occurring while the chip is still cool. Here's another thought: Toshiba sold its chip business to China in late 2018. If the TB6560 is now being made in a different facility, then there is certainly a possibility that chip processing changed. Are you able to find date codes on new vs old chips? Or perhaps other marking that indicates a change in mfr was made? If the diodes fix the problem, that will not be conclusive proof that your design was at fault; the diodes may allow faulty chips to survive.

 

That's interesting they sold last year. Could be the quality is no longer there. I'll investigate the date codes if I can figure it out.

While I was waiting for the PCBs with flybacks I hardwired the diodes in to test and had a failure. Testing one of the actual boards now.

I've tried a variety of power supplies with failures. I had some in stock that appeared to be good quality. So far no failures on the first 6 boards. Maybe these chips are sensitive to power supplies? I suppose I could clean it up with a choke and capacitor.

Still trying a variety of things. Thanks for chiming in.

 

That's interesting they sold last year. Could be the quality is no longer there. I'll investigate the date codes if I can figure it out.

While I was waiting for the PCBs with flybacks I hardwired the diodes in to test and had a failure. Testing one of the actual boards now.

I've tried a variety of power supplies with failures. I had some in stock that appeared to be good quality. So far no failures on the first 6 boards. Maybe these chips are sensitive to power supplies? I suppose I could clean it up with a choke and capacitor.

Still trying a variety of things. Thanks for chiming in.

How about C3? Is it positioned very close to the TB6560 so that its leads are very short to minimize inductance? Is trace width adequate (both 5V and ground)? Have you tried low ESR ceramics in parallel with C3, e.g. a 0.1uF & 1uF--with short leads on each? 100uF electrolytics vary widely in how much ESR they have. Do you not have an oscilloscope to view noise across C3?

 

C3 is close to U1 but not so much U3. I'll use a scope to see how much noise is there. U3 Pin 20 is probably the best place to check I assume. I believe the trace is thick enough for the 5v side of that chip and the grounds go direct to the ground plane. I guess that will all become apparent when I check.

 

I'm seeing small spikes at pin 20 and C3 of about 0.25v as it runs. On the 12v side I'm seeing 2v+.

 

I'm seeing small spikes at pin 20 and C3 of about 0.25v as it runs. On the 12v side I'm seeing 2v+.

It would be very helpful to learn how to photograph the scope screen (or otherwise obtain a screen image) & upload the image. Since you say "spikes" of 0.25V, I assume that you mean the +5V increases for a very brief period (i.e. it does not drop). In my opinion, that magnitude spike (and also that at 12V) is not a problem. However, be sure you look carefully with a fast scope sweep speed as the spikes may be very narrow. (Actually, in many cases a very slow sweep--being brighter--will reveal narrow spikes...try both slow & fast.) A very closely placed bypass cap is nevertheless highly recommended.

 

I think it is resolved. TeeKay6 mentioned that Toshiba sold that business to a company in China the end of last year so I believe the Toshiba chips may have been more robust which could explain why I started having problems all of the sudden. Looking at the noise at C3 (5v) didn't seem to be an issue but the 12v going into the chip was wild which I expected some because that is driving the load. I added a 1uf and 100uf in parallel at the 12v supply into the chip and removed the flyback diodes and it smoothed out nicely. Haven't had anymore failures, knock on wood. The datasheet does show these capacitors in the application circuit example which I should have picked up on a few years ago which was sloppy. But the boards worked so well for so long I thought it was a good design.

Thanks everyone for helping out.